Yolo Bypass Flood Date and Flow Volume Agricultural Impact Analysis

Full text

DRAFT REPORT
Yolo Bypass Flood Date and Flow Volume Agricultural
Impact Analysis
PREPARED FOR: Yolo County
PREPARED BY: Richard Howitt1, Duncan MacEwan1, Cloe Garnache1,
Josue Medellin Azuara1, Petrea Marchand2, and Doug Brown3
DATE: May 15, 2012
1 University of California, Davis
2 County of Yolo
3 Douglas Environmental

i
Executive Summary
The state and federal government propose to increase the frequency and duration of flooding in
the Yolo Bypass for fish habitat, both as a major component of the Bay Delta Conservation Plan
(BDCP) and also as a Reasonable and Prudent Alternative (RPA) in the federal National Marine
Fisheries Service’s Biological Opinion for winter run salmon, spring run salmon, and Central
Valley steelhead. Under both alternatives the project will have broader support and cost less if
effects on items such as flood protection, migratory waterfowl and other terrestrial species
habitat, and agriculture are minimized. This report provides a quantitative framework for
assessment of agricultural impacts from proposed flooding scenarios in the Yolo Bypass and
evaluates both the Biological Opinion RPA and BDCP Conservation Measure 2 (CM2).
Yolo Bypass Fisheries Enhancement Conservation Measure #2 (CM2), as stated in the February
2012 BDCP draft, would lower a portion of the Fremont Weir to an elevation of 17.5 feet, from
its current elevation of 32.8 feet, and construct an operable gate to allow Sacramento River water
to flow into the Yolo Bypass (BDCP 2012). CM2 also includes a number of other actions within
the Yolo Bypass including construction of fish passage improvements at the Fremont Weir. CM2
actions are designed to reduce migratory delays and loss of adult salmon, steelhead, and
sturgeon, enhance rearing habitat for Sacramento River Basin salmonids, enhance spawning and
rearing habitat for Sacramento splittail, and improve food sources for delta smelt downstream of
the Bypass. The operations pattern4 for the proposed Fremont Weir gate suggests supplemental
flooding up to 6,000 cubic feet per second (cfs) for 30 to 45 days in years when flooding occurs
naturally in the Yolo Bypass.
The Biological Opinion Reasonable and Prudent Alternatives (RPA), Actions I.6 and I.7, would
also lower a portion of the Fremont Weir and construct an operable gate to allow Sacramento
River water to flow into the Yolo Bypass. The RPA scope is to provide habitat for winter run
salmon, spring run salmon, and central valley steelhead. The RPA requires flooding for juvenile
winter run, spring run, and Central Valley steelhead from “December through April” in the
“lower Sacramento River basin.” The RPA further identifies “an initial performance measure” of
17,000 to 20,000 acres with “appropriate frequency and duration.” The RPA recommends flows
through an operable gate at the Fremont Weir for inundation of up to 20,000 acres in the Yolo
Bypass. This study provides a data-driven assessment of the effect of proposed RPA and CM2
implementation on agricultural productivity and Yolo County’s overall economy, and provides a
framework for evaluating BDCP proposals once they are further developed.
The 57,000-acre Yolo Bypass is first and foremost one of the primary means of providing flood
protection to the Sacramento region. In addition, Yolo Bypass agriculture provides significant
benefits to the local economy, migratory waterfowl, and the flood protection system. The Bypass
can carry, on average, four times the flow of the Sacramento River or approximately 420,000 cfs.
Yolo Bypass agriculture helps to maintain this flood capacity by controlling vegetation, thereby
reducing the state’s responsibility for vegetation removal. Yolo Bypass rice fields also provide
habitat and food for migratory waterfowl when flooded for straw decomposition during the
winter months.
4 See Table 3.4-3 of the February 2012 BDCP Draft Report.

ii
“Natural” flooding in the Yolo Bypass can occur at any time from the Sacramento River
overtopping the Fremont Weir and/or from tributary flows entering the Bypass from the west
during storm events. Farmers have adapted to these conditions and landowners have lowered
their lease rates to some extent to reflect the risk. Natural flooding delays planting times and
reduces crop yields in the Bypass – or even prevents planting. Late season flood events may
reduce crop yields through short-duration flooding, even if farmers prepare fields early in the
season. As such, increased frequency and duration of inundation within the Bypass for fish
habitat may translate into financial losses for farmers and the regional economy.
Current proposals recommend flows between 3,000 and 6,000 cfs through an operable gate in the
Fremont Weir. Flooding at the proposed volumes would inundate5 between 12,200 and 25,000
total6 acres, assuming no flooding from creeks on the west side of the Yolo Bypass. An increase
in flooding could result in economic losses to farmers and the local economy, dependent on
timing, frequency, volume, and duration. In addition, flooding may increase the costs of late
season rains which potentially affects land values, lending institutions, and farming in the Yolo
Bypass.
This study evaluates the expected losses of total agriculture revenue, total Yolo County revenue
(value added), tax revenue, and jobs for the twelve policy scenarios listed in Table 1. These
scenarios evaluate the Biological Opinion RPA under five proposed end dates for Fremont Weir
flows through an operable gate, and the CM2 scenario in the February 2012 BDCP draft, for two
flow rates.
Table 1. Inundation End Dates / Scenarios
3,000 cfs 6,000 cfs
Feb 15 Feb 15
Mar 24 Mar 24
Apr 10 Apr 10
Apr 30 Apr 30
May 15 May 15
CM2 Scenario CM2 Scenario
The fundamental driving factors in the analysis are total acres inundated, reduced crop yields,
and increased land fallowing. As the last day of flooding through the proposed gate in the
Fremont Weir increases, farmers must delay field preparation and planting, resulting in reduced
crop yields and increased land fallowing. Agricultural revenues fall, which translates into losses
in the Yolo County economy and employment in the region. Table 2 identifies the expected total
annual losses to the Yolo County economy (also known as value added losses) associated with
5 This study is an agricultural impact analysis and, as such, areas of inundation include the literal flooding “footprint” plus fields
that are partially inundated, discussed in Section 2.2.
6 12,200 total acres includes 4,500 acres of wetlands and Liberty Island, and 25,000 total acres includes 9,200 acres of wetlands
and Liberty Island. Thus, flooding will affect between 7,700 and 15,800 acres of land used for agricultural production. This
footprint does not include any land in Solano County.

iii
the inundation scenarios evaluated in the study. Under the RPA proposal, the effect of increased
flooding early in the season is small, less than $0.25 million with 6,000 cfs flow. Flooding
through May 15 significantly increases effects, with total losses to Yolo County economy of $3.8
million and $8.9 million under 3,000 cfs and 6,000 cfs, respectively. Under the CM2 proposal,
where flooding only occurs as an extension to natural flooding, expected annual losses range
from $0.63 to $1.5 million under 3,000 and 6,000 cfs, respectively.
Table 2. Expected Total Annual Loss to Yolo County Economy (Value Added) (Thousands of 2008
dollars)
Inundation End Date / Scenario 3,000 cfs 6,000 cfs
February 15 148 241
March 24 931 1,744
April 10 2,337 5,015
April 30 3,371 7,735
May 15 3,886 8,889
CM2 Scenario 625 1,468
This analysis does not explicitly consider changes in late season rains and management or
operation difficulties which may affect drainage and field preparation times. Consequently, the
estimates in this study are a conservative measure of the expected annual losses to the economy
from increased frequency and duration of inundation in the Yolo Bypass. We also would like to
stress the model results are sensitive to several assumptions. In particular, the areas of inundation
under different flooding scenarios may change with different hydrologic models and
consideration of tributary flows. We also use an expected crop price that is representative of an
average over the past 25 years and neither relies on recent boom price levels or earlier depressed
agricultural conditions.
In addition to expanded inundation scenarios as more information becomes available, we also
recommend the following actions:
 Create inundation scenarios that include the west side tributaries to the Bypass once existing
models are adequately reviewed.
 Create inundation scenarios that reflect potential constrained project footprints of 7,000 to
10,000 acres, since the current analysis only models unconstrained flooding and therefore
includes acres that do not directly benefit fish.
 Analyze the effect of crop insurance on farmer responses to inundation proposals.
 Analyze the response of agricultural lending institutions to inundation proposals.
 Evaluate proposed flood policies under a range of expected future crop prices.
 Compare the predicted area of inundation under the MIKE21 and HEC-RAS models.

iv
We thank Yolo County, the State and Federal Contractors Water Agency (SFCWA), and the
Conaway Preservation Group for the funding and support necessary to prepare this study.

v
This page intentionally blank.

vi
Table of Contents
Executive Summary.......................................................................................................................i
Table of Contents......................................................................................................................... vi
1Introduction....................................................................................................................... 1
1.1Scope of Analysis and Caveats............................................................................... 2
1.2Inundation Scenarios............................................................................................... 2
2Data Overview...................................................................................................................5
2.1Agricultural Sub-regions......................................................................................... 5
2.2Field Level Crop Data and Flood Footprint............................................................ 7
2.3Crop Yields........................................................................................................... 15
2.4Crop Prices............................................................................................................ 15
2.5Costs of Production............................................................................................... 17
2.6Areas of Inundation............................................................................................... 18
3Overview of the Modeling Approach............................................................................ 22
3.1Estimating Crop Yields (DAYCENT Model)....................................................... 23
3.2Bypass Production Model..................................................................................... 25
3.3Adjustments for Natural Flooding........................................................................ 26
3.4IMPLAN ............................................................................................................... 29
4Results.............................................................................................................................. 31
4.1Acreage Change Summary ................................................................................... 31
4.2Revenue Losses Summary.................................................................................... 33
4.3Employment Losses Summary .............................................................................36
4.4Tax Losses Summary............................................................................................37
5Sensitivity Analysis .........................................................................................................38
6Conclusion ....................................................................................................................... 41
7References........................................................................................................................ 42
0Technical Appendix: Overview of the Modeling Approach .......................................44
1Data Preparation............................................................................................................. 47
1.1Land Use and Production Data............................................................................. 47
1.2The DAYCENT Model......................................................................................... 47
1.3Yield Functions Regression Analysis................................................................... 47
2The Bypass Production Model (BPM) Calibration...................................................... 55
2.1Positive mathematical programming (PMP)......................................................... 55
2.2Model Calibration................................................................................................. 56
2.3Profit Maximization Program Definition.............................................................. 57

vii
3BPM Simulation.............................................................................................................. 59
4BPM Output and Expected Losses................................................................................60
4.1IMPLAN ............................................................................................................... 60

viii
Figures
Figure 1. Yolo Bypass Sub-regions..............................................................................................16
Figure 2. Agricultural Land Use, Yolo Bypass 2005....................................................................20
Figure 3. Agricultural Land Use, Yolo Bypass 2006....................................................................21
Figure 4. Agricultural Land Use, Yolo Bypass 2007....................................................................22
Figure 5. Agricultural Land Use, Yolo Bypass 2008....................................................................23
Figure 6. Agricultural Land Use, Yolo Bypass 2009....................................................................24
Figure 7. Commodity Price Trends, Monthly Prices from 1992-2012.........................................26
Figure 8. Agricultural Land Flooded under 3,000 cfs flow rates..................................................30
Figure 9. Agricultural Land Flooded under 6,000 cfs flow rates.\ ...............................................31
Figure 10. Illustration of the Fundamental Modeling Approach..................................................32
Figure 11. Example Yield Function, Rice in Region 1.................................................................35
Figure 12. Loss of Irrigated Acres in Region Affected by 3,000 cfs Flow Rate, by
Year................................................................................................................................................42
Figure 13. Loss of Irrigated Acres in Region Affected by 6,000 cfs Flow Rate, by
Year................................................................................................................................................42
Figure 14. Loss of Irrigated Acres in Region Affected by CM2 Proposal, by Year and
Flow Volume. ................................................................................................................................43
Figure 15. Price Sensitivity Analysis for Gross output Value under 3,000 cfs, All
Scenarios........................................................................................................................................49
Figure 16. Price Sensitivity Analysis for Gross output Value under 6,000 cfs, All
Scenarios........................................................................................................................................50
Figure A1. Illustration of the Fundamental Modeling Approach.................................................53
Figure A2. Model Framework Flow Chart...................................................................................54
Figure A3. Fitted Yield Function for Corn in Region 1. ..............................................................59
Figure A4. Fitted Yield Function for Pasture in Region 6............................................................60
Figure A5. Fitted Yield Function for Rice in Region 2................................................................60
Figure A6. Fitted Yield Function for Safflower in Region 1........................................................61
Figure A7. Fitted Yield Function for Sunflower in Region 1.......................................................61
Figure A8. Fitted Yield Function for Processing Tomatoes in Region 3. ....................................62
Figure A9. Fitted Yield Function for Melons (Vine Seed) in Region 4.......................................62
Figure A10. Histogram of Overtopping Date Frequencies (84-09 and (05-09). ..........................65

ix
Tables
Table 1. Inundation End Dates / Scenarios....................................................................................ii
Table 2. Expected Total Annual Losses to Yolo County Economy (Value Added)
(Thousands of 2008 dollars) ..........................................................................................................iii
Table 3. Major Land Uses in Areas Affected by Flooding in the Yolo Bypass (acres) ...............18
Table 4. Crop Prices, 2009-2010 Average and 2000-2009 Average (2008 dollars per
ton).................................................................................................................................................27
Table 5. Variable Production Costs ($/ton) per acre (in 2008 dollars).........................................28
Table 6. Estimated Yield by Planting Date (last day of water releases) (tons/ac)........................34
Table 7. Fremont Weir Overtopping End Dates...........................................................................38
Table 8. Expected Annual Total Revenue Loss (2008 Dollars), RPA Scenarios.........................44
Table 9. Expected Annual Total Revenue Loss (2008 Dollars), CM2 Scenario..........................44
Table 10. Expected Annual Value Added Loss (2008 Dollars), RPA Scenarios.........................45
Table 11. Expected Annual Value Added Loss (2008 Dollars), CM2 Scenario ..........................45
Table 12. Expected Annual Agricultural Jobs Loss, RPA Scenarios...........................................46
Table 13. Expected Annual Agricultural Jobs Loss, CM2 Scenario ............................................46
Table 14. Expected Annual Total Statewide Tax Revenue Losses (2008 Dollars),
RPA Scenarios...............................................................................................................................47
Table 15. Expected Annual Total Statewide Tax Revenue Losses (2008 Dollars),
CM2 Scenario ................................................................................................................................47
Table 16. Price Sensitivity Analysis Range (2008 dollars), All Scenarios...................................48
Table A1. Pasture Yield Function Parameter Estimates (standard errors in
parentheses)....................................................................................................................................56
Table A2. Corn Yield Function Parameter Estimates (standard errors in parentheses) ...............57
Table A3. Vine Seed (Melons) Yield Function Parameter Estimates (standard errors
in parentheses) ...............................................................................................................................57
Table A4. Rice Yield Function Parameter Estimates (standard errors in parentheses)................57
Table A5. Safflower Yield Function Parameter Estimates (standard errors in
parentheses)....................................................................................................................................58
Table A6. Sunflower Yield Function Parameter Estimates (standard errors in
parentheses)....................................................................................................................................58
Table A7. Processing Tomatoes Yield Function Parameter Estimates (standard errors
in parentheses) ...............................................................................................................................58

x

1
1 Introduction
The state and federal government propose to increase the frequency and duration of flooding in
the Yolo Bypass for fish habitat, both as a major component of the Bay Delta Conservation Plan
(BDCP) and also as a Reasonable and Prudent Alternative (RPA) in the federal National Marine
Fisheries Service’s Biological Opinion for winter run salmon, spring run salmon, and Central
Valley steelhead. Under both alternatives, the project will have broader support and cost less if
impacts – such as flood protection, migratory waterfowl and other terrestrial species, and
agriculture – are minimized.
In this study, we estimate the extent of inundation, crop yield loss, and effects on the agricultural
economy from increasing the frequency and duration of flooding in the Yolo Bypass. All
estimates include the direct economic effects associated with reduced agricultural production, as
well as multiplier (direct and induced) effects associated with upstream and downstream changes
to the regional economy. We developed the framework in this study as an analytic tool which
may be used to evaluate alternative policy scenarios in the future. For example, the BDCP
proposal for the Yolo Bypass and the salmon RPA were not fully developed at the time of this
study. The analysis uses twelve specific policy scenarios to demonstrate the framework, but can
be used in the future to analyze any Bypass proposal.
Yolo Bypass Fisheries Enhancement Conservation Measure #2 (CM2), as stated in the February
2012 BDCP draft, would lower a portion of the Fremont Weir to an elevation of 17.5 feet, from
its current elevation of 32.8 feet, and construct an operable gate to allow Sacramento River water
to flow into the Yolo Bypass (BDCP 2012). CM2 also includes a number of other actions within
the Yolo Bypass including construction of fish passage improvements at the Fremont Weir. CM2
actions are designed to reduce migratory delays and loss of adult salmon, steelhead, and
sturgeon, enhance rearing habitat for Sacramento River Basin salmonids, enhance spawning and
rearing habitat for Sacramento splittail, and improve food sources for delta smelt downstream of
the Bypass. The operations pattern7 for the proposed Fremont Weir gate suggests supplemental
flooding up to 6,000 cubic feet per second (cfs) for 30 to 45 days in years when flooding occurs
naturally in the Yolo Bypass.
The Biological Opinion RPA, Actions I.6 and I.7, would also lower a portion of the Fremont
Weir and construct an operable gate to allow Sacramento River water to flow into the Yolo
Bypass. The RPA scope is to provide habitat for winter run salmon, spring run salmon, and
central valley steelhead. The RPA requires flooding for juvenile winter run, spring run, and
Central Valley steelhead from “December through April” in the “lower Sacramento River
basin.” The RPA further identifies “an initial performance measure” of 17,000 to 20,000 acres
with “appropriate frequency and duration.” The RPA recommends flows up to 8,000 cfs through
an operable gate at the Fremont Weir for inundation of up to 20,000 acres in the Yolo
Bypass. This study provides a data-driven assessment of the effect of proposed RPA under five
end dates to water flows through an operable gate at Fremont Weir under two flow volumes.
We use the HEC-RAS hydrologic model and the DAYCENT agronomic model to estimate the
extent of inundation and change in crop yield, respectively, under a series of proposed policies.
7 See Table 3.4-3 of the February 2012 BDCP Draft Report.

2
We estimate the effect on agricultural production using the Bypass Production Model (BPM),
developed specifically for the Yolo Bypass. The BPM estimates the change in crop mix,
agricultural revenues, and other factors due to crop yield loss (DAYCENT model) and the
number of acres affected (HEC-RAS and MIKE-21 models) in the Yolo Bypass. Results from
the BPM are linked to the IMPLAN regional input-output model to estimate total output, value-
added, and employment losses within the Yolo Bypass and the Yolo County economy.
1.1 Scope of Analysis and Caveats
In this report we model the effects of increased flooding on Yolo Bypass agriculture and the
county economy. Thus the geographic scope of the analysis is Yolo County and, in particular, the
Yolo Bypass. We do not consider crop production shifts out of the region. This would require, in
part, an analysis of the rice mills in West Sacramento and Woodland to determine proportion of
business from Bypass production in addition to other regional economic effects. Additionally, a
shift in rice production out of the Bypass is an agronomic question since specific soil and climate
is required.
The modeling approach we adopt is sensitive to several parameters which we have made every
attempt to clearly state and, where possible, report sensitivity analysis. We preview the important
parameters in this section and review them throughout the text. Section 5 provides sensitivity
analysis.
Subbing: Increased flooding in the Bypass may raise the groundwater table in regions out of the
Bypass. This may restrict farming and/or reduce yields in affected areas, thereby increasing
economic losses.
Late Rains: We provide expected annual loss estimates by using a time series of hydrologic
conditions in the Bypass. However, late season rains may have additional costs which we have
not captured. For example, if farmers begin field preparation late due to flooding for fish habitat
and late rains occur, this may delay planting further and increase economic losses.
Prices: Expected future crop prices are uncertain. We use a conservative estimate of 2009-2010
average prices which does not reflect recent booms or historic depressed levels. We provide
sensitivity analysis in Section 5.
Lending and Insurance: We do not evaluate the effect of increased flooding on lending and
insurance for farmers in the Bypass. This is related to late season rains and other management
difficulties Bypass farmers may face with extended policy flooding.
Drought: For the RPA scenarios we have implicitly assumed water will be available for policy
flooding in every year. Extended drought may lower the river level below the range of the
operable gate at Fremont Weir, which may decrease expected losses since flooding will not
occur in these years.
1.2 Inundation Scenarios
We consider five inundation dates and two different flow rates associated with RPA
implementation. Additionally, we consider the CM2 proposal under the same flow rates, for a

3
total of twelve policy scenarios (see Table 1). The inundation dates correspond to the last day of
Sacramento River water releases through operable gates in the Fremont Weir: February 15th,
March 24th, April 10th, April 30th, and May 15th. The two flow rates are 3,000 cfs and 6,000 cfs,
which correspond to the flows recommended for fish in Technical Study #2: Evaluation of North
Delta Migration Corridors: Yolo Bypass prepared for the BDCP Integration Team in April 2009.
We identified the five end dates to represent a range of outcomes from RPA alternatives to
flooding for fish habitat in the Yolo Bypass. The RPA only include flooding through April,
however we include a May 15th date to inform discussions related to potential flooding for
splittail. The BDCP currently proposes flooding for splittail every 7 years if flooding does not
occur naturally, although the acres of splittail flooding are not specified. Once acreage targets are
more fully refined, the model framework can be used to develop loss estimates specific to
proposed flooding scenarios.
The CM2 scenario, as described in the introduction, corresponds to supplemental flooding in
years with natural overtopping at Fremont Weir. As such, the end date in this scenario is variable
and depends on the specific water year. In Section 3.3 we describe the time series of hydrologic
conditions used to generate annual expected losses in the CM2 scenario.
Fields in the Bypass must drain before farmers can begin preparation for planting. Agricultural
fields located along the east side of the Bypass adjacent to the Tule Canal/Toe Drain tend to
drain slower than higher elevation fields to the west. Slower drainage time on the east side delays
planting date and tends to lower crop yields. On average, it takes two weeks for fields to drain on
the west side of the Bypass and four weeks on the east side of the Bypass. Field preparation takes
an additional four weeks. Thus, there is a delay of six to eight weeks between the last day water
is released through a Fremont Weir gate and planting, depending on the location of the field.
February 15th. February 15th represents an end date to Fremont Weir flooding when agriculture
is largely unaffected. Farmers have an adequate buffer for unforeseen circumstances, such as rain
or cool conditions that lengthen the time needed for field drainage. Farmers state they prefer to
start ground preparation by March 15th to allow adequate time for field work and planting. It
takes approximately 4 weeks from the date a farmer can start field work to the date of planting,
so an end date of February 15th would typically result in early April planting on the west side of
the Bypass and mid-April planting on the east side.
March 24th. March 24th represents an end date to Fremont Weir flooding when growers are
expected to experience yield losses (see Section 3). Consequently, we anticipate some land
fallowing and shift in crop mix but, in general, crop yields are high enough to cover variable
costs.
April 10th. The April 10th end date translates into planting in early June. According to farmers
interviewed, in an average year, June 10th is the last possible date to plant. With an April 10th end
date to water releases, farmers would plant in late May on the west side and by June 10th on the
east side. As such, significant yield losses and land fallowing are expected in this scenario. If any
unforeseen circumstances occur in this scenario, there is a high risk that planting will not occur.
April 30th. The April 30th end date translates into planting in late June and corresponds to the
latest flood date under the RPA. According to farmers interviewed, in an average year, June 10th

4
is the last possible date to plant. As such, significant yield losses and land fallowing are expected
in this scenario. In this scenario, planting may not occur at all on the east side of the Bypass and
there is a high risk that planting will not occur on the west side.
May 15th. The May 15th end date for water releases represents a date when farmers state they
will not plant crops. However, this date is frequently referred to in public forums as important for
splittail habitat. Yield response functions from the DAYCENT model confirm that crop yields
are not high enough to cover variable operation costs if flooding ends May 15th. Consequently,
significant land fallowing would occur. However, contracts and other fixed costs may induce
farmers to plant late in the season.
CM2. The CM2 scenario corresponds to a verbal interpretation of the proposal in the BDCP
February 2012 draft. This proposal may change in the future. In this scenario, flooding is
extended by 30 days in years with natural flooding in the Bypass to augment habitat and there is
no flooding in dry years. We use a 26 year hydrologic time series, described in Section 3.3, to
simulate this proposal. For example, with natural flooding until February 1 the CM2 proposal
extends flooding by 30 days, through March 1.

5
2 Data Overview
We collected extensive data for the Yolo Bypass to facilitate an empirical analysis of the
proposed inundation scenarios. These include the following: (i) field-level geo-referenced crop
data and agricultural region definitions, (ii) crop yields and yield change based on planting date,
(iii) crop prices, (iv) costs of production, and (v) area inundated under proposed flow volumes.
We review these data in the following section.
2.1 Agricultural Sub-regions
The Yolo Bypass slopes gradually downward from west to east and north to south. Temperatures
are generally lower in the southern end of the Bypass. Consequently, there are heterogeneous
production conditions across the region and natural differences in both yield and drainage times.
We identified 7 homogenous agricultural sub-regions in the Yolo Bypass which represent these
production conditions and, as such, form the basis of the BPM. We used soil and climate data, in
addition to interviews with Bypass farmers, to develop homogenous agricultural sub-regions.
The regions are illustrated in Figure 1.
Note that the BPM, as with the majority of agricultural production models, is a regional
economic model, defined over the 7 regions illustrated in Figure 1. Field-level yield and
production data are available for a subset of fields in the Bypass (discussed below), and these
data are used in the DAYCENT agronomic model. We discuss this point again in Section 2.3 and
again in Section 3, but want to raise the point here so the reader is not confused about the use of
field-level data versus agricultural sub-regions in the model.
As shown in Figure 1, Regions 1 and 2 are located north of Interstate 5, Regions 3 and 4 are
located between Interstate 5 and Interstate 80, and Regions 5, 6 and 7 are located south of
Interstate 80. The area south of Interstate 80 is divided into three regions due to its relatively
large width and the distinct row crop region located in its western portion, which distinguishes it
from the managed wetlands and grazing lands located to the east. Region 7 is located outside of
the flood inundation footprint and is not anticipated to be affected by the implementation of CM2
of the RPA, so it is not discussed in further detail in this report or considered in the analysis.

6
Figure 1. Yolo Bypass Sub-regions

7
2.2 Field Level Crop Data and Flood Footprint
We compiled detailed land use data for 2005-2009 from Pesticide Use Reports, the Yolo Natural
Heritage Program, the Sacramento-Yolo Mosquito and Vector Control District, the Yolo Basin
Foundation, and individual farmers. As a result of the extent of data collected, and verification
with key stakeholders, the database for this study is the most comprehensive and detailed
information on Yolo Bypass land use available.
Table 3 identifies major land uses in the area of the Bypass affected by each of the respective
flow volumes (identified by the HEC-RAS hydrologic model, discussed in Section 2.6) over the
five years of data collected for the study. Agricultural land constitutes the majority of the area
within the Bypass, followed by wetland and fallow land. The main crops in the affected area of
the Yolo Bypass are rice, irrigated pasture, processing tomato, vine seed, safflower, wild rice,
corn, and sunflowers.
We model 3,000 cfs and 6,000 cfs scenarios in this report which correspond to different total
affected acres, as estimated by the HEC-RAS model. An important consideration for the
agricultural impacts analysis is that the geography of land use in the Bypass means that a sub-set
of fields will be partially innundated. In other words, the HEC-RAS model estimates a “literal”
footprint of affected acres dependent on the flow volume, but this does not account for existing
agricultural fields. Cultivation of proportions of fields is costly and, in many cases, impossible.
Partial innundation makes it difficult or impossible to use machinery to begin field preparation
and, as such, the field is effectively entirely inundated. It is essential to account for the difference
between the literal footprint from hydrologic modeling and the effective footprint, the latter is
relevant for agricultural impact analysis.
In order to incorporate the effective flood footprint, we conducted a series of interviews with
Bypass farmers and extension specailists to determine the proportion of a field flooded at which
farmers cannot begin preparation. Farmers interviewed report the decision to prepare a partially
inundated field is different between rice and other field crops and depends on a number of factors
including relative prices, weather, and costs. We determined when 20 percent of a rice field is
flooded farmers will not begin preparation. For all other crops, 30 percent is the relevant
proportion. Fields partially inundated according to the above proportions are modeled as
effectively flooded and consequently included in the affected acres estimates.
Note that preparation of a partially innundated field includes installation of checks to control
existing flooding and other potentially costly management alternatives. We do not include these
production costs in the analysis, thus our estimates are conservative.

8
Table 3. Major Land Uses in areas affected by flooding in the Yolo Bypass (acres)
Cro
p
and Flow Volume 2005 2006 2007 2008 2009
Fallow
3,000 cfs 3,220 3,606 1,702 1,514 984
6,000 cfs 6,640 6,860 2,858 3,526 2,297
Libert
y
Islan
d
3,000 cfs 2,071 2,071 2,071 2,071 2,071
6,000 cfs 2,071 2,071 2,071 2,071 2,071
Vine
3,000 cfs 245 0 0 0 72
6,000 cfs 245 104 0 0 238
Pasture
3,000 cfs 2,026 2,026 2,026 2,026 2,284
6,000 cfs 3,890 3,890 3,987 3,890 5,166
Rice
3,000 cfs 765 173 931 968 1,531
6,000 cfs 2,358 1,254 2,920 2,409 4,263
Safflower
3,000 cfs 606 657 519 770 499
6,000 cfs 1,450 1,545 1,616 1,840 1,273
Sunflowe
r
3,000 cfs 138 0 0 0 0
6,000 cfs 138 0 0 0 0
Processin
g
Tomatoes
3,000 cfs 662 867 721 930 1,047
6,000 cfs 1,285 1,285 1,370 1,829 1,779
Wetlan
d
3,000 cfs 2,501 2,502 2,503 2,504 2,505
6,000 cfs 7,076 7,076 7,076 7,076 7,076
Wild Rice
3,000 cfs 0 195 427 494 494
6,000 cfs 0 928 2,292 2,303 2,393
Corn
3,000 cfs 0 138 584 208 0
6,000 cfs 0 138 925 208 0

9
We identified 9 major crop groups in areas affected by flooding in the Bypass, which we use for
the subsequent analysis. The 9 crops include corn, irrigated pasture, non-irrigated pasture, rice,
wild rice, safflower, sunflower, processing tomatoes, vines (melons). Fallow land is an implicit
tenth group. Approximately 100 acres of crops did not fit into these categories directly, including
dry beans and organic rice. We determined that the number of acres was not sufficient to require
an additional crop group and these acres were included in the crop group with the most similar
cost, return, and production characteristics. Specifically, organic rice acres were added to the rice
crop group and dry bean acres were added to the corn crop group.
Figures 2-6 illustrate the distribution of land use across the entire Yolo Bypass, by field, for the
years 2005 through 2009. These data show typical crop rotations across the sub-regions. In the
southern end of the Bypass, the crops are predominately pasture and in the northern sub-regions
the crops are predominately rice. The eastern sub-regions include a mix of pasture, rice, corn,
and processing tomatoes.
Crop acreage increased during the dry years of 2007 through 2009 and fallow land decreased.
Note that 2008 and 2009 were characterized by high agricultural commodity prices potentially
driving larger acreages into production, in particular for corn and wheat. Rice prices spiked in
2008, which partially explains the increase in rice acreage in the Yolo Bypass. Water year type
also affects production. The California Department of Water Resources classified 2005 as an
above normal hydrologic year type, 2006 as wet, and 2007 through 2009 as dry years. The
Fremont Weir overtopped through May 3rd in 2006, overtopped for three days in May of 2005
(resulting in a couple of weeks of inundation), and did not overtop in 2007 through 2009.

10
Figure 2. Agricultural Land Use, Yolo Bypass 2005

11
Figure 3. Agricultural Land Use, Yolo Bypass 2006

12
Figure 4. Agricultural Land Use, Yolo Bypass 2007

13
Figure 5. Agricultural Land Use, Yolo Bypass 2008

14
Figure 6. Agricultural Land Use, Yolo Bypass 2009

15
2.3 Crop Yields
Holding total area inundated constant, crop yields are the fundamental driving factor for
agricultural revenue losses due to flooding in the Yolo Bypass. We use two sources of
information on crop yields in this analysis. This procedure is outlined here, explained again in
Section 3, and all the technical details and equations are contained in Appendix A.
We observe field-level yield data and other micro-production characteristics (soil, climate, etc.)
for a subset of fields in the Bypass. These fields are used to calibrate the DAYCENT agronomic
model. The DAYCENT model estimates the yield on any given field taking into account all
production conditions, including climate and date the crop was planted. We then use the
calibrated DAYCENT model to estimate crop yields on a subset of fields in each of the 6 regions
of the BPM. We control for all other factors and allow the planting date to vary, thus the
DAYCENT model generates a series of data points, for each crop and region, which tells us the
expected yield conditional on the crop planting date.
We use the data points from the DAYCENT results to estimate a single yield function, for each
crop and region. We fit this function using non-linear regression analysis (discussed in Section 3
and Appendix A). The result is a single function, for each crop and region in the Bypass, which
relates crop yield to the planting date. These functions are included in the BPM, discussed in
Section 3.
In summary, we use field-level production observations to calibrate a field-level agronomic
model. We use the model to simulate the yield on a subset of fields for each crop and region as a
function of planting date. Finally, we fit a non-linear function to these data, for each crop and
region. Thus, we are able to determine crop yields, for each region, as a function of the planting
date.
Note that consistent data on the yields, prices and costs of growing melons for vine seed were
unavailable. Instead, we use economic information for melons grown for fruit, accordingly crop
yields and budgets are expressed in terms of melons grown for fruit. This is not a critical
assumption since melon acreage in the affected area averages less than 200 acres per year
(between 2005 and 2009).
2.4 Crop Prices
We obtained crop prices for the 9 crops considered in the analysis from the Yolo County
Agricultural Commissioner reports (Agricultural Commissioners Reports, 2012). No price data
per animal unit month (AUM) or hay production was available for pasture, thus we used the
price estimate per AUM per acre provided in the Cost and Returns study for flood irrigated
pasture grown in the Sacramento Valley (UC Cooperative Extension, 2003). Additionally,
sunflower prices are only available for 2007 and 2008 in the Agricultural Commissioner’s data.
Therefore, we used data reported by the National Agricultural Statistics Service (NASS). We
also use NASS data for wild rice because no price data are available prior to 2006.
One of the key components of this analysis is expected crop prices. Higher crop prices translate
into larger losses per acre and induce farmers to plant later in the season, thereby reducing fallow
land. The results of this study are sensitive to the choice of expected future crop prices.

16
Unfortunately, there is no general consensus for future expected crop prices. The commodity
price spike of 2007/2008 was unprecedented and followed decades of declining real commodity
prices. Prices have since declined but remain higher than pre-spike levels and appear to have
stabilized on a higher trend. Figure 7 illustrates the 20 year trend in corn and rice prices and
highlights the difficulty of selecting representative prices to use in this analysis.
Figure 7. Commodity Price Trends, Monthly Prices from 1992 - 20128
All of the impact analysis in this report uses a two year average (2009-2010) of crop prices for
each of the crop groups. There are two main reasons for this which include, (i) these years are
representative of historical average prices in Yolo County and, (ii) 2009 and 2010 crop prices
exclude the price spikes in 2008 and again in 2011. The 2009 and 2010 average prices represent
a conservative and defensible estimate for crop prices in this analysis.
Table 4 summarizes the average crop price9 (dollars per ton) for each of the crop groups included
in the analysis. Column two shows the prices used (2009-2010 average) and column three shows
the 10 year average crop price. Related to point (i), above, Table 4 shows that 2009-2010
average crop prices are representative of the recent history (2000 - 2009 average). Namely, rice
and corn prices are slightly higher than the 10 year average but other crops are generally lower.
Column four reports 2008 prices for each of the crops. With the exception of corn and safflower,
all crop prices were significantly inflated in 2008. In summary, 2009 and 2010 average prices are
8 Data compiled from http://www.indexmundi.com/
9 Rice prices do not include direct payments, counter-cyclical program payments, or marketing loan payments. Where applicable,
these are included in the data used for the analysis.

17
representative of recent prices in Yolo County and, more importantly, omit the recent price
spikes.
Table 4. Crop Prices, 2009-2010 average and 2000-2009 average (2008 dollars per ton)
Crop Group 2009-2010 Average 2000-2009 Average 2008
Corn 172.69 124.31 152.20
Irrigated Pasture 49.20 (based on $35 per
AUM) 49.20 (based on $35 per
AUM) 49.20 (based on $35
per AUM)
Non-Irrigated
Pasture 49.20 (based on $35 per
AUM) 49.20 (based on $35 per
AUM) 49.20 (based on $35
per AUM)
Rice 397.89 251.36 513.10
Wild Rice 961.85 1,275.30 1,684.20
Safflower 351.18 319.79 432.62
Sunflower 1,196.15 1,781.47 1,092.32
Processing
Tomatoes 78.81 59.15 68.81
Vine Seed (Melon
Proxy) 303.00 292.9 296.10
2.5 Costs of Production
In this report, we use Cost and Return studies developed by the UC Cooperative Extension
(UCCE) to determine crop costs of production. These studies provide production costs for
representative farmers in the Sacramento Valley and, as such, are representative of Bypass
farming. Crop budgets are prepared for various years, thus we use the NASS prices paid indices
for specific item categories to express each item cost in constant 2008 dollars.
Land prices are excluded from the input data in the model, thus the BPM represents returns to
land and management. This is common in PMP models, technical discussion is left in Appendix
A. However, note that PMP captures implicit land costs through the calibration routine, thus
these costs are not “omitted” from the model. Table 5 summarizes the variable costs of
production for each crop.

18
Table 5. Variable Production Costs ($/ton) per acre (in 2008 dollars)
Crop Group Cost
Corn $607
Melons $4,110
Pasture irrigated $269
Pasture dry $118
Rice $898
Safflower $239
Sunflower $553
Tomato, processing $1,838
Wild rice $502
2.6 Areas of Inundation
The second key driving factor in this analysis is the total number of affected acres under
proposed flow volumes from Fremont Weir water releases through an operable gate. We
consider two flow volumes, 3,000 and 6,000 cfs, in this report.
We estimate the number of affected acres using the one-dimensional HEC-RAS hydrologic
model hydraulic simulation model. We use the HEC-RAS model for two reasons including, (i)
the National Marine Fisheries Service used the HEC-RAS model to estimate acreage for the
Biological Opinion, and (ii) Yolo County is still in the process of completing an independent
review of the MIKE-21 model. An initial comparison of the MIKE-21 and HEC-RAS footprints
for 3,000 cfs and 6,000 cfs indicate the difference is relatively small.
Given the potential interest in this issue, some additional information is necessary to justify the
decision to rely on HEC-RAS. Both one-dimensional (1-D) and two-dimensional (2-D) models
are useful tools in hydraulic engineering and water resource planning studies. The accuracy of
both 1-D and 2-D models is strongly dependent upon the quality of information specified by the
user as input into the model and on the boundary conditions (flow, initial water level and channel
roughness) the user must also specify. It can therefore be difficult to compare results without
understanding how each model was developed, including how bed roughness, inflow and stage
boundary conditions were specified, and other how other assumptions and constraints were
entered as user-specified inputs to each model. Once Yolo County develops a better
understanding of the models’ assumptions, the existing framework can be used to estimate
effects for different model outputs.
Figures 8 and 9 identify the fields inundated under the 3,000 and 6,000 cfs flow rates. We
consider a field, in terms of restricting farm operations, to be effectively inundated if 30 percent
or more of the field was inundated for field crops and 20 percent or more for rice crops. As
discussed in Section 2.2, this reflects input received from Bypass farmers indicating that they
would not typically initiate field preparation efforts if a portion of their field is still partially

19
inundated. The blue areas in these figures identify the predicted flood inundation area. The red
and yellow areas identify the contiguous fields that would be affected at 20% and 30%,
respectively. Note that as the flow rate increases, the number of affected acres increases.
Consequently, planting dates are delayed on more fields and farm revenue losses are expected to
increase.

20
Figure 8. Agricultural Land Flooded under 3,000 cfs flow rates.

21
Figure 9. Agricultural Land Flooded under 6,000 cfs flow rates.

22
3 Overview of the Modeling Approach
We estimate the effect of the twelve proposed policies on Bypass agriculture based on the data
summarized in Section 2 and a series of empirical models, summarized in this section. This
section briefly reviews the modeling approach and policy scenarios evaluated. A detailed
technical overview of the modeling approach is included in Appendix A.
Figure 10 provides a simple illustration of the key steps in our analysis. Starting with input data
(including the HEC-RAS and MIKE-21 models) described in the previous section, we use a
series of linked models to estimate the effects on agriculture. The DAYCENT model is an
agronomic model used to estimate field-level yields, as a function of planting date, for subsets of
fields in each region of the Bypass. Regression analysis on the DAYCENT model output and
additional input data are used to calibrate the BPM. Output from the BPM and other input data
are used as inputs to the IMPLAN model. The fundamental results include direct, indirect, and
induced (the sum of which is total) expected effects on total agricultural output (revenues), value
added, agricultural employment, and statewide taxes.
Figure 10. Illustration of the Fundamental Modeling Approach
We briefly preview the five steps outlined in Figure 10, and provide more details in the
subsequent sections.
Data: Input data were described in Section 2. In summary, we compiled a comprehensive
economic, agronomic, and geo-referenced dataset of agricultural production in the Yolo Bypass
between 2005 and 2009.
DAYCENT Model: Field-level data were used to calibrate the agronomic DAYCENT model
(DeGryze et al 2009). We use the DAYCENT model to estimate crop yields as a function of
various agronomic conditions, including planting date. We use non-linear regression analysis to
fit a series of crop yield functions, for each crop and region in the Bypass. Technical details are
provided in Appendix A.
BPM: We use the crop yield functions estimated from the DAYCENT model, plus additional
economic data, to calibrate the BPM. The BPM is the fundamental model of this analysis. The
BPM relates changes in crop yield and total affected acres to changes in agricultural production
and, fundamentally, changes in agricultural revenues. The BPM is a Positive Mathematical
Programming (PMP after Howitt, 1995) model of agriculture in the 6 regions of the Yolo
Bypass. PMP models calibrate exactly to an observed, base year, of production conditions and
grower decisions and have been used extensively for water and agriculture policy analysis in
Data DAYCENT Model BPM IMPLAN Results

23
California and around the world. Appendix A reviews the technical details of the BPM and PMP
calibration procedure.
IMPLAN: The IMPLAN model estimates regional economic losses. Expected revenue losses
from the BPM analysis represent direct economic effects. However, upstream and downstream
industries will be affected and some agricultural workers will lose their jobs when production in
the Bypass decreases. We use the IMPLAN regional Input-Output (IO) model to estimate the
direct, indirect, and induced effects of the 12 policy scenarios. The sum of these components
represents the total effect of the policies.
The key result from this overview is that all of the analysis in this report is driven by observed
data and observed grower decisions in the Bypass. We use a sequence of linked models to
estimate the total (direct, indirect, and induced) effects of flood date and flow volume on
agriculture in the Yolo Bypass. These effects are defined and described in detail in Section 4 and
Appendix A.
3.1 Estimating Crop Yields (DAYCENT Model)
Crop yields are the fundamental driving factor for agricultural revenue losses due to flooding in
the Yolo Bypass. As farmers delay planting, crop yields decline which, in turn, leads to lower
revenues and land fallowing. We estimate crop yields, and variation based on planting date,
using the DAYCENT agronomic model and non-linear regression analysis on output data.
We can summarize the procedure as two steps, (i) estimate field-specific yields using the
DAYCENT model and, (ii) use the DAYCENT model output to perform regression analysis and
estimate crop and region-specific yield functions. These functions relate crop yield to the
planting date and are directly incorporated into the BPM. More information about this process is
available in Appendix A.
Table 6 presents the results (after both steps are completed) from the yield data analysis by sub-
region. Yields vary across regions and by planting date. Recall that after the last day of water
releases through the Fremont Weir gate, there is a 6-8 week delay before planting occurs, this is
implicitly built into the yield data summarized in Table 6.
There are crop and region specific functions underlying all of the data summarized in Table 6.
Figure 11 summarizes this function for an example crop of Rice in Region 1. Yield functions for
all the crops can be found in Appendix A. The vertical axis identifies the expected yield, the
horizontal axis identifies the date, red triangles are output data from the DAYCENT field-level
model, and the blue line represents the results of the fitted non-linear yield function.
There are several things to note from the example in Figure 11. First, one of these functions (the
blue line) exists for every crop in every region. This governs the relationship between crop yield
and planting date and, in part, drives the results of the economic (BPM) model. Second, note that
the relationship is non-linear, as expected. Over some range early in the season, farmers will
realize only a slight yield decline from a small delay in planting date. However, substantial
delays cause yields to decline rapidly.

24
Table 6. Estimated yield by planting date (last day of water releases) (tons/ac)
Last day of water releases at Fremont Weir
Yield (ton/acre) Region Feb 15th March 24th April 10th May 15th
Corn 1 5.84 4.72 0.51 0.00
Corn 2 5.90 5.84 4.05 0.01
Corn 3 5.88 4.76 0.59 0.00
Corn 4 5.73 5.48 3.09 0.02
Pasture - dry (AUM/acre) 5 0.45 0.29 0.25 0.21
Pasture - dry (AUM/acre) 6 0.55 0.33 0.28 0.22
Pasture - irrigated (AUM/acre) 5 2.23 1.44 1.26 1.05
Pasture - irrigated (AUM/acre) 6 2.77 1.64 1.38 1.10
Rice 1 4.14 3.19 1.08 0.01
Rice 2 4.15 3.98 2.88 0.09
Rice 3 4.15 3.20 1.09 0.01
Rice 4 4.12 3.92 2.76 0.09
Rice 5 3.66 2.50 1.14 0.07
Rice 6 3.74 3.42 2.41 0.21
Safflower 1 1.07 0.51 0.29 0.07
Safflower 2 1.19 1.01 0.76 0.21
Safflower 3 1.09 0.51 0.29 0.08
Safflower 4 1.09 0.74 0.48 0.14
Safflower 5 0.98 0.41 0.21 0.04
Safflower 6 1.10 0.70 0.43 0.12
Sunflower 1 0.64 0.56 0.52 0.45
Sunflower 6 0.63 0.60 0.56 0.46
Processing Tomato 1 38.57 34.60 28.79 10.35
Processing Tomato 2 38.76 37.25 33.98 17.59
Processing Tomato 3 38.99 35.06 29.18 10.29
Processing Tomato 6 38.36 36.23 32.48 17.74
Melons 2 7.52 7.52 6.55 3.55
Melons 3 6.80 6.20 4.84 2.10
Melons 4 6.65 6.65 5.77 2.97
Wild rice 1 0.92 0.71 0.24 0.00
Wild rice 2 0.92 0.88 0.64 0.02
Wild rice 3 0.92 0.71 0.24 0.00
Wild rice 4 0.92 0.87 0.61 0.02
Wild rice 5 0.81 0.56 0.25 0.02
Wild rice 6 0.83 0.76 0.54 0.05

25
Figure 11. Example Yield Function, Rice in Region 1
3.2 Bypass Production Model
The Bypass Production Model (BPM) combines the HEC-RAS data, DAYCENT yield functions,
and other economic data into a Positive Mathematical Programming (PMP) agricultural
production model of the Yolo Bypass. The model calibrates exactly to an observed base year of
input and output data which, in our analysis, is 2005 - 2009 average land use. In other words, the
model exactly replicates observed farmer behavior, in terms of input use and outputs, over this
period. Once the model calibrates, and a series of economic and numerical checks are satisfied
(see Howitt et al. 2012), we use the BPM to simulate changes in agricultural production under
the twelve proposed policy scenarios. We review the basics of the BPM in this section, the
interested reader can find technical details in Appendix A.
The BPM estimates the change in crop mix, agricultural revenues, and other factors due to crop
yield loss (DAYCENT model) and the number of acres affected (HEC-RAS model) in the Yolo
Bypass. The BPM calibrates to an average of 2005-2009 land use input data (summarized in
Section 2). All dollars are expressed in 2008 real terms. Crop prices for calibration are an
average of 2005-2007 prices in Yolo County. The 2005-2007 average prices were determined to
be representative of conditions farmers in the Yolo Bypass faced, on average, when making

26
planting decisions between 2005 and 2009. Input costs are expressed in 2008 dollars, from the
UCCE budgets. Policy simulations use 2009-2010 average crop prices, as discussed previously.
Technical details of the PMP calibration procedure and functional forms in the model are left to
Appendix A. We briefly review the estimation procedure in this section. The BPM estimation
procedure can be summarized as a series of five steps:
Step 1: Calibrate the BPM to base data (2005 - 2009, as discussed previously). Perform a series
of checks to ensure economic and numerical conditions are satisfied.
Step 2: Run the BPM for a season with known overtopping dates at Fremont Weir, and flooding
in the Yolo Bypass. This represents the base condition (e.g. natural flooding) for agriculture in
the Bypass in the absence of the proposed policy flooding scenarios (for that year). Repeat Step 2
for a series of known years. There are 26 known overtopping dates in the analysis which are
discussed in more detail in the following section.
Step 3: Over the same series of years as step two, run the BPM and impose (sequentially - one at
a time) the twelve proposed policy flooding scenarios. This represents what would have
happened to Bypass agriculture if the flooding policy was implemented in that year. Repeat Step
3 for all of the same years as Step 2.
Step 4: For each year simulated in Steps 2 and 3, calculate the difference in agricultural revenues
(and other outputs). Record the result for negative changes in revenue. Intuitively, for policy
evaluation we are interested in negative changes in revenue because a positive change in revenue
implies that the policy was “better” than nature. For example, if natural flooding occurred in the
Bypass until April 30th, imposing a policy which stops water releases from a Fremont Weir gate
on April 10th would not be possible (i.e. it would increase revenues).
Step 5: Calculate the average loss of revenue (and other changes) across all of the years
simulated in Steps 2 - 4. This represents the expected effects due to the proposed flooding
scenarios, and is the fundamental output of the BPM.
The fundamental procedure of the BPM is to generate an expected effect on agriculture by using
the calibrated model to estimate what would have happened under natural flooding, and then
asking what would have happened if a specific policy (last day of water releases) was in place.
This procedure allows us to generate an expected effect because we control for the expected
natural flood events in the Bypass. The following section illustrates this point.
3.3 Adjustments for Natural Flooding
In many years flooding occurs naturally in the Yolo Bypass and, in some years, flooding may
occur late in the season. Estimates of agricultural losses need to account for the fact that natural
conditions may result in flooding beyond the proposed policy date. We use a 26 year (1984-
2009) time-series of hydrologic conditions in the Bypass to estimate expected future revenue
losses in the Bypass. The implicit assumption is that the previous 26 years are representative of
expectations for natural flooding in the near future. The implications of this assumption and
details on the procedure used in the BPM are described in more detail in Appendix A.

27
Given the 26 year time-series, estimates represent expected annual losses due to flooding for fish
habitat in the Bypass. There are two reasons these 26 years of data were identified as reasonable,
including (i) detailed flow information over the Fremont Weir was available for these years, and
(ii) it is representative of current hydrologic conditions in the Sacramento Valley watershed.
Older hydrologic information less accurately represents current conditions because it does not
account for changes in urban development and reservoir operations that have altered flows in the
Sacramento River over time.
Table 7 summarizes the observed last day of overtopping and provides some notes about the
nature of flooding in key years. During the 26 years, there are five years (1989, 1996, 1998, 2003
and 2005) in which flooding events in the Yolo Bypass did not occur consecutively. In these
years, except for 2003, an early dry period enabled farmers to proceed with their land
preparation, but planting was delayed or significantly affected by late floods. To account for this
in the analysis, 28 days (the amount of time needed for field preparation) was credited to the
planting date in these years. This assumes that farmers had to wait for the fields to drain in these
years, but required minimal field preparation effort since this was completed earlier in the
season.

28
Table 7. Fremont Weir Overtopping End Dates
Year End Date Important Notes and Adjustments
1984 11-Jan
1985 -
1986 25-Mar
1987 -
1988 -
1989 14-Mar Early dry year, followed by late flooding, farmers able to prepare fields early
reducing the effect of late flooding
1990 -
1991 -
1992 -
1993 6-Apr
1994 -
1995 13-May
1996 24-May Early dry year, followed by late flooding, farmers able to prepare fields early
reducing the effect of late flooding
1997 13-Feb
1998 8-Jun Early dry year, followed by late flooding, farmers able to prepare fields early
reducing the effect of late flooding
1999 14-Mar
2000 17-Mar
2001 -
2002 10-Jan
2003 7-May Flooding confined to the Toe Drain; minimal effect on agriculture
2004 10-Mar
2005 24-May Early dry year, followed by late flooding, farmers able to prepare fields early
reducing the effect of late flooding
2006 5-May
2007 -
2008 -
2009 -

29
3.4 IMPLAN
We use the Impact Analysis for Planning model (IMPLAN) Professional Version 3 and a 2009
database for Yolo County. We link the IMPLAN model to results from the BPM, in order to
estimate changes in total output value, value added, employment, and tax revenues as a result of
the proposed flood policies. IMPLAN is an input-output model which accounts for relationships
between sectors of the economy in order to estimate the effects of a change (e.g. reduced
agricultural output) in another sector of the economy. IMPLAN is widely used by State and
Federal agencies including the California Department of Water Resources, the California
Regional Water Quality Control Boards, the U.S. Army Corps of Engineers, the U.S. Bureau of
Reclamation, the U.S. Bureau of Economic Analysis, and the U.S. Bureau of Land Management.
We summarize four key outputs for this analysis: changes in total output value, changes in
“value added”, changes in employment, and changes in statewide tax receipts. For each output
we report direct, indirect, and induced effects, the sum of which is the total effect. We define
these components below, further technical details can be found in Appendix A.
Total Output Value (e.g. Gross Revenues): The gross value of agricultural production in the
Yolo Bypass to the “global” economy. For example, this is price multiplied by yield/acre
multiplied by the total number of acres.
Total Value Added: The net value of agricultural production in the Yolo Bypass to the Yolo
County economy. This measure recognizes that many inputs/outputs are produced or consumed
outside of Yolo County and, as such, are not relevant effects for the flood policy analysis. For
example, food production is exported out of the county, state, or country for many crops.
Similarly, tractors are produced outside of the county, fertilizers are produced in another state,
etc. The measure of value added controls for these effects. Total value added includes
compensation for employees, income to business and landowners, and other business, specific to
Yolo County.
Total Employment: The change in agricultural employment in Yolo County due to changes in
agricultural production in the Yolo Bypass. Specifically, this includes NAICS classification
system sector 111 - agricultural employment.
Total Statewide Tax Revenue: The change in tax receipts due to reduced output in the Yolo
Bypass.
Each of these components has a direct, indirect, and induced effect on the Yolo County
economy. The sum of the three is the total effect and sometimes the indirect and induced effects
are jointly referred to as “multiplier” effects. We define these terms below.
Direct: Immediate effects on the relevant agricultural economy. For example, gross farm
revenue losses due to reduced yields in the Bypass.
Indirect: Changes in related sectors as a result of direct changes to production in the Bypass. For
example, reduced production in the Bypass will cause farmers to purchase fewer inputs, this is an
indirect effect.

30
Induced: Changes in all other sectors of the economy as a result of the direct changes to
production in the Bypass. For example, reduced production in the Bypass will lead to reduced
hours for farm workers who will, in turn, purchase fewer goods and services in the region.
Total: Direct + Indirect + Induced

31
4 Results
We summarize the results of the analysis in this section. Results correspond to each of the 12
policy scenarios (water release end date and flow volume) for the four measures detailed in
Section 3.4. First, we summarize changes in acreage across the Bypass.
Results are annual expected losses, reported in constant 2008 dollars.
4.1 Acreage Change Summary
Figures 12 and 13 summarize the change in total irrigated acres, by year, under the eight policy
scenarios (four end dates for water releases through a Fremont Weir gate and two flow rates)
corresponding to the RPA scenarios. Farmers may fallow land or shift small amounts of land to
alternative crops in response to delayed planting due to flooding. These figures highlight the
decision to fallow land. Note the effect of the 26 year simulation on the analysis. Years with late
natural flooding reduce (or eliminate) losses to Yolo Bypass agriculture due to the policy
scenarios.
There is a base level of average fallow acres in any given year within each of the affected 3,000
and 6,000 cfs flood areas. Specifically, in the 3,000 cfs flood region, the 2005 through 2009 base
(calibration) data shows that an average of 2,200 acres are fallow in any given year. Similarly, in
the 6,000 cfs flood region, 4,400 acres are fallow in any given year. These additional fallow
acres are not included in Figures 12 and 13.
Figure 12 shows the time-series of change in total irrigated acres in the region affected by the
3,000 cfs flow rate. As the last day of water releases from a Fremont Weir gate is delayed, the
total number of irrigated acres removed from production increases. An average of 2,580 acres are
removed from production per year if the last day of water releases is May 15th.
Figure 13 illustrates the time-series of change in total irrigated acres in the region affected under
the 6,000 cfs flow rate. As the last day of water releases is delayed, the total number of irrigated
acres removed from production increases. The total number of irrigated acres removed from
production is larger than the 3,000 cfs inundation region. On average, 7,400 acres are removed
from production under a last day of water releases on May 15th. The maximum number of acres
out of production is estimated at over 15,200 with a last day of water releases on May 15th.
Figures 12 and 13 also illustrate the importance of natural flooding in the Bypass. In years where
the is natural flooding, the effects of the policy are minimal. For example, February and March
flooding have a limited effect in many years. Averaging over these 26 years allows us to
generate an expected annual effect of the proposed policies.

32
Figure 12. Loss of Irrigated Acres in Region Affected by 3,000 cfs Flow Rate, by Year.
Figure 13. Loss of Irrigated Acres in Region Affected by 6,000 cfs Flow Rate, by Year.
We also evaluated the CM2 scenario where water flows through an operable gate at Fremont
Weir are only imposed for an additional 30 days in years when there is natural flooding. As
expected, the losses under this proposal are minimal. Figure 14 illustrates the change in total
irrigated acres over the 26 years used in our analysis for both 3,000 and 6,000 cfs CM2
scenarios. Note that losses only occur in years when there is natural flooding.
The largest losses occur in years when natural overtopping occured late into the season. For
example, in 1999 and 2000 heavy rains caused Fremont Weir overtopping through March 14 and
March 17, respectively. The CM2 proposal calls for an additional 30 days of flooding which
means flooding through the middle of April. This results in crop yield losses and an increase in
fallow acres.

33
Figure 14. Loss of Irrigated Acres in Regions Affected by CM2 Proposal, by Year and Flow Volume.
4.2 Revenue Losses Summary
We summarize the expected agricultural revenue losses for each flow rate and last day of water
releases from the Fremont Weir gate in Table 8. As shown, total output value (gross farm
revenue) expected losses range from $0.28 to $17.3 million per year in the RPA scenarios,
depending on the last day of water releases from the Fremont Weir gate and the flow rate. As
expected, a later water release date delays planting and, consequently, reduces crop yields and
increases farm revenue losses. Similarly, higher flow rates affect more fields and increase farm
revenue losses.
Losses for the RPA scenarios should be interpreted as annual expected losses from continuous
flooding up to the identified end date.

34
Table 8. Expected Annual Total Revenue Loss (2008 dollars), RPA Scenarios
Expected Total Revenue Loss (Output Value) ($2008)
3,000 cfs 6,000 cfs
February 15
Direct 172,278 280,530
Indirect+Induced 116,463 189,826
Total 288,741 470,356
March 24
Direct 1,081,960 2,026,110
Indirect+Induced 731,777 1,370,310
Total 1,813,737 3,396,420
April 10
Direct 2,713,780 5,823,400
Indirect+Induced 1,835,472 3,938,499
Total 4,549,252 9,761,899
April 30
Direct 3,915,080 8,981,760
Indirect+Induced 2,647,896 6,074,741
Total 6,562,976 15,056,501
May 15
Direct 4,512,650 10,333,200
Indirect+Induced 3,052,140 6,988,682
Total 7,564,790 17,321,882
Expected losses for the CM2 scenario range between $1.2 to $2.8 million per year. The CM2
scenario corresponds to supplemental releases only in years where natural flooding occurs. As
such, loss estimates are much lower, between $1.2 and $2.8 million per year. Note that in some
years losses are zero (when there is no natural flooding) and in other years losses are substantial
(when there is late natural flooding). These loss estimates correspond to expected annual losses,
summarized in Table 9.
Table 9. Expected Annual Total Revenue Loss (2008 dollars), CM2 Scenario
Expected Total Revenue Loss (Output Value) ($2008)
3,000 cfs 6,000 cfs
CM2 Scenario
Direct 725,930 1,704,640
Indirect+Induced 490,987 1,152,982
Total 1,216,917 2,857,622

35
A proportion of Yolo Bypass production and crop consumption occurs within Yolo County. As
such, losses to Yolo County are expected to be less than total revenue losses. The proper measure
of the effect on the Yolo County economy is change in “value added” (defined in section 3.4).
Table 10 summarizes the change in value added under the proposed flooding policies. In the
RPA scenarios expected losses in value added range from $0.14 to $8.9 million per year.
Table 10. Expected Annual Value Added Loss (2008 dollars), RPA scenarios
Expected Total Yolo County Revenue Loss (Value Added) ($2008)
3,000 cfs 6,000 cfs
February 15
Direct 74,648 121,954
Indirect+Induced 73,568 119,914
Total 148,216 241,868
March 24
Direct 469,589 879,285
Indirect+Induced 462,261 865,620
Total 931,850 1,744,905
April 10
Direct 1,177,877 2,527,185
Indirect+Induced 1,159,463 2,487,936
Total 2,337,340 5,015,121
April 30
Direct 1,699,112 3,898,193
Indirect+Induced 1,672,667 3,837,395
Total 3,371,779 7,735,587
May 15
Direct 1,958,644 4,484,527
Indirect+Induced 1,928,028 4,414,727
Total 3,886,672 8,899,254
Comparable to the output value losses, value added losses in the CM2 scenario are lower than
many of the RPA scenarios. Table 11 summarizes the CM2 results. Expected annual losses to
value added range from $0.63 to $1.5 million per year.
Table 11. Expected Annual Value Added Loss (2008 dollars), CM2 scenario
Expected Total Yolo County Revenue Loss (Value Added) ($2008)
3,000 cfs 6,000 cfs
CM2 Scenario
Direct 315,084 739,971
Indirect+Induced 310,155 728,336
Total 625,239 1,468,307

36
4.3 Employment Losses Summary
Table 12 summarizes the corresponding expected annual agricultural job losses under the
proposed flooding policies. Employment effects are generally small, ranging from no effect to
130 jobs lost.
Table 12. Expected Annual Agricultural Jobs Loss, RPA scenarios
Expected Total Employment Loss
 3,000 cfs 6,000 cfs
February 15
Direct 1 2
Indirect+Induced 1 2
Total 2 4
March 24
Direct 7 13
Indirect+Induced 7 12
Total 13 25
April 10
Direct 17 37
Indirect+Induced 16 35
Total 34 73
April 30
Direct 25 58
Indirect+Induced 24 55
Total 49 112
May 15
Direct 29 66
Indirect+Induced 27 63
Total 56 129
Table 13 summarizes the CM2 scenario employment losses. Direct expected gross revenue
losses are less than $1.5 million per year and the corresponding job losses are small.
Table 13. Expected Annual Agricultural Jobs Loss, CM2 Scenario
Expected Total Employment Loss
3,000 cfs 6,000 cfs
CM2 Scenario
Direct 5 11
Indirect+Induced 4 10

37
Total 9 21
4.4 Tax Losses Summary
Table 14 summarizes the total expected annual losses in tax revenues to the state under the
proposed flooding scenarios in the RPA. Annual tax revenue losses can be as high as $0.82
million under the 6,000 cfs flow scenario that extends flooding as late as May 15. For the 3,000
cfs flow regime scenario, annual tax revenue losses are less than $0.36 million.
Table 14. Expected Annual Total Statewide Tax Revenue Losses (2008 dollars), RPA Scenarios
Expected State and Local Tax Revenue Loss ($2008)
3,000 cfs 6,000 cfs
February 15 13,604 22,193
March 24 85,515 160,130
April 10 214,496 460,241
April 30 309,428 709,892
May 15 356,677 816,686
Table 15 summarizes the expected annual tax revenue losses to the state for the CM2 scenario.
Table 15. Expected Annual Total Statewide Tax Revenue Losses (2008 dollars), CM2 Scenario
Expected State and Local Tax Revenue Loss ($2008)
3,000 cfs 6,000 cfs
CM2 Scenario 57,377 134,744

38
5 Sensitivity Analysis
Results of the analysis are sensitive to parameters and assumptions listed in Section 1.1. Some
overstate and others understate expected losses however, overall, we feel that our estimates are
conservative. Nonetheless, some sensitivity analysis is warranted.
Expected loss estimates are most sensitive to changes in area inundated, yield loss, and crop
prices. Area inundated is driven by HEC-RAS model results which are based on RPA and CM2
scenarios. As such, we don’t have a basis to vary the number of affected acres. Similarly, yield
loss is a function of planting date which is driven by agronomic data and non-linear regression
analysis. Thus we do not have a justifiable basis to vary this relationship. Prices, as discussed in
Section 2.2 are highly uncertain and we perform sensitivity analysis on these parameters.
We select 2005-2006 average prices to represent a “low” price scenario and 2008 prices to
represent a “high” price scenario. Note that some crop prices are actually higher (lower) than the
base scenario for the lower (higher) sensitivity analysis scenarios. This is expected since some
crop prices are correlated and we typically don’t expect to observe all prices trending in the same
direction. In other words, a sensitivity analysis where all crop prices are 10 percent higher is not
relevant sensitivity analysis. Table 16 summarizes the low and high prices used for sensitivity
analysis, in addition to the base (2009-2010) prices used in the analysis. Note that the largest
uncertainty occurs with the price of rice, which experienced a large spike in 2008 following
years of lower prices.
Table 16. Price Sensitivity Analysis Range (2008 dollars), All Scenarios
Crop Group 2005-2006 Average
(LOW) 2009-2010 Average
(BASE) 2008
(HIGH)
Corn 141.00 172.69 152.20
Irrigated Pasture 49.20 (based on $35 per
AUM) 49.20 (based on $35 per
AUM) 49.20 (based on $35
per AUM)
Non-Irrigated
Pasture 49.20 (based on $35 per
AUM) 49.20 (based on $35 per
AUM) 49.20 (based on $35
per AUM)
Rice 274.80 397.89 513.10
Wild Rice 1,469.30 961.85 1,684.20
Safflower 314.80 351.18 432.62
Sunflower 1,056.10 1,196.15 1,092.32
Processing
Tomatoes 67.75 78.81 68.81

39
Vine Seed (Melon
Proxy) 349.80 303.00 296.10
Figure 15 summarizes the results of the price sensitivity analysis for the 3,000 cfs scenarios.
Sensitivity analysis corresponds to the output of the BPM model, gross agricultural revenues
(gross output value), or the direct effects listed in Table 8. The base estimate has been
normalized to 1, thus the bars show the percentage deviation due to prices. For example, in the
April 10 RPA scenario low prices reduce losses by 24 percent (0.76) and high prices increase
losses by 23 percent (1.23).
Figure 15. Price Sensitivity Analysis for Gross output Value under 3,000 cfs, All Scenarios.
Figure 15 summarizes the results of the price sensitivity analysis for the 6,000 cfs scenarios.
Again, sensitivity analysis corresponds to the output of the BPM model, gross agricultural
revenues (gross output value), or the direct effects listed in Table 8. The base estimate has been
normalized to 1, thus the bars show the percentage deviation due to prices. For example, in the
April 10 RPA scenario low prices reduce losses by 13 percent (0.87) and high prices increase
losses by 25 percent (1.25). Figures 15 and 16 indicate that results are slightly sensitive to crop
prices, as expected. Our estimates based on 2009-2010 average prices are generally conservative
since the deviation from the base is generally above 1.

40
Figure 16. Price Sensitivity Analysis for Gross output Value under 6,000 cfs, All Scenarios.
Other areas where we are unable to perform sensitivity analysis include weather shocks and
changes in the cost of production. The latter raises an important point, namely we have implicitly
assumed that the costs of production in the Bypass remain constant even with late flooding.
However, if production costs go up, for example due to overtime labor or increased preparation
costs, loss estimates will increase.

41
6 Conclusion
This study has assembled extensive data on cropping, water use, and the economics of the
agricultural industry in the Yolo Bypass. We then use this data to calibrate and link four models.
Namely, an engineering model of field flood inundation (HEC-RAS ), an agronomic model of
yield loss due to shorter growing seasons (DAYCENT), an economic production model of farm
crop decisions in the Yolo bypass (BPM), and finally a regional economic model of the Yolo
County economy (IMPLAN). The net economic results from these four models are measured as
a set of output values for twelve alternative flood scenarios that cover two different volumes of
flooding and five different ending dates for the RPA, plus an evaluation of the CM2 proposal.
The five overtopping dates analyzed were selected to span the full range from no effect on
cropping, to the cost of flooding that prevents any cropping, and intermediate values.
For each of the twelve scenarios the net dollar effect on the Yolo County economy is measured
in terms of value-added. The loss in employment is measured in terms of full-time equivalent
jobs, and the effect on the State tax receipts. The expected economic value added losses range
widely from $0.15 to $8.9 million per year. The effect on job losses and tax receipts also varies
widely, depending on the scenario.
Despite our efforts to assemble the very best data set, we would like to stress that the model
results are sensitive to several assumptions. In particular, we would like to note that the areas of
inundation under different flooding scenarios may well change with different engineering models
and better data. In addition, we have attempted to use a weighted price for future crops that is
representative of an average over the past 25 years and neither relies on recent boom price levels
or earlier depressed agricultural conditions.
We would also like to emphasize that this study is only able to measure the expected cost to the
Yolo County economy, and is not able to account for changes in risk, management difficulties,
and other factors facing the county and the agricultural industry in the Bypass. As such, the
results of this study should be regarded as a conservative measure of the expected annual
economic costs to the county economy of changes in flood policy in the Bypass.

42
7 References
Bay-Delta Conservation Plan (BDCP). (2012) February 2012 Draft Report. Available at :
http://www.baydeltaconservationplan.com. Last Accessed May 5, 2012.
DeGryze, S, R. Catala, R. E. Howitt, J. Six. (2009) Assessment of Greenhouse Gas Mitigation in
California Agricultural Soils California Energy Commission (2009) Issue: January,
Publisher: UC Davis for Public Interest Energy Research (PIER) Program, California
Energy Commission, Pages: 160
Buysse, J., Huylenbroeck, G., and Lauwers, L. 2007. Normative, Positive and Econometric
Mathematical Programming as Tools for Incorporation of Multifunctionality in
Agricultural Policy Modelling. Agriculture, Ecosystems, and Environment. 120:70-81.
Heckelei, T. & W. Britz. 2005. Models Based on Positive Mathematical Programming: State of
the Art and Further Extensions. 21. Bonn, Germany: University of Bonn, Institute for
Agricultural Policy, Market Research and Economic Sociology.
Heckelei, T. & H. Wolff (2003) Estimation of Constrained Optimisation Models for Agricultural
Supply Analysis based on Generalised Maximum Entropy. European Review of
Agricultural Economics, 30, 27-50.
Howitt, R.E. 1995. Positive Mathematical-Programming, American Journal of Agricultural
Economics, 77(2), 329-342.
Howitt, R.E., J. Medellin-Azuara, D. MacEwan, and J.R. Lund. 20112. “Calibrating
Disaggregate Models of Irrigated Production and Water Use: The California Statewide
Agricultural Production Model.” Working Paper. University of California, Davis.
IMPLAN website: http:www.implan.com
Leontief, W., 1941. The structure of American economy, 1919-1929; an empirical application of
equilibrium analysis. Harvard University Press, Cambridge, Mass.
Merel, P. and Bucaram, S. (2010) Exact Calibration of Programming Models of Agricultural
Supply Against Exogenous Supply Elasticities. European Review of Agricultural
Economics, 37, 395-418.
Merel, P., Simon, L., Yi, F. (2011) A Fully Calibrated Generalized CES Programming Model of
Agricultural Supply. American Journal of Agricultural Economics, (forthcoming).

43
University of California Cooperative Extension (UCCE). Cost of Production Studies. Various
Crops and Dates. Department of Agricultural and Resource Economics. Davis,
California. URL = http://coststudies.ucdavis.edu.
County Agricultural Commissioners (AgCom). Annual Crop Reports. Various years, various
counties. Available at http://www.nass.usda.gov/Statistics_by_State/California/
Publications/AgComm/Detail/index.asp.

44
0 Technical Appendix: Overview of the Modeling Approach
Evaluation of agricultural policies requires a modeling framework which can be used to simulate
losses and estimate costs. In this report, we adopt a modeling framework driven entirely by a
rich, empirical dataset, highlighted by Figure A1. We estimate the effect of 12 proposed policies
of flood level and date for fish habitat on Bypass agriculture. The scenarios include flow rates of
3,000 and 6,000 cfs from the Sacramento River passing through an operable gate in the Fremont
Weir. The last day of overtopping at Fremont Weir occurs on February 15, March 24, April 10,
April 30 or May 15. Additionally, we evaluate the CM2 proposal which does not correspond to a
specific end date.
Figure A1 provides a simple illustration of the key steps in the analysis. Starting with input data
(including the HEC-RAS model), we use a series of linked models to estimate the impacts to
agriculture. The DAYCENT model is an agronomic model used to estimate field-level yields, as
a function of planting date, for subsets of fields in each region of the Bypass. Regression analysis
on the DAYCENT model output and additional input data are used to calibrate the BPM. Output
from the BPM and other input data are used as inputs to the IMPLAN model.
Figure A1. Illustration of the Fundamental Modeling Approach
Production and geo-referenced land use data, HEC-RAS output, DAYCENT simulations, and
regression analysis are used as inputs to the Bypass Production Model (BPM). The BPM is the
fundamental economic model in the analysis. The technical details of the analysis can be
summarized in four phases including, (i) data preparation, (ii) calibration, (iii) estimation, and
(iv) output. The flow chart in Figure A2 illustrates this process, which we review in detail in this
technical appendix.
Data preparation involves the compilation and synthesis of model data, including geo-referenced
land use data, production data, and HEC-RAS model output. This stage additionally includes
field-level simulations with the DAYCENT model and regression analysis. Model calibration
includes development of the Bypass Production Model (BPM) and exact calibration, through
Positive Mathematical Programming, in inputs and outputs to a known base year. Estimation
involves simulation of the calibrated BPM over a series of known water years (nature) and
sequentially imposing the 12 proposed policies on the model. The difference between the base
and policy simulations is recorded for all years with revenue losses. The output phase estimates
losses from the BPM and generates expected annual gross revenue losses. Output from the BPM
are input to the IMPLAN model to estimate Yolo County direct, indirect, and induced economic
effects.
Data DAYCENT Model BPM IMPLAN Results

45
Data
Preparation
Simulation
(Over26WY)

Base:Flooding
duetonature

Policy:Proposed
floodingforfish
habitat
Calibration

Output

(Calculate
difference
betweenbase
andpolicy)
Soil,price,
cost,
HEC_RAS
DAYCENT
MODELYields Economicfunctions
Bypass
Production
Model(BPM)
S
tandardcalibrationchecks
AverageBPM
forcalibration
3KCFS 6KCFS
BasePolicy
BPM
WY84

WY09
Policy
Scenario
BPM
WY84

WY09
Policy
Scenario
Base Policy
Recordnegativerevenue Recordnegativerevenue
PolicyOutput
SummaryMatrix
(6×2)
A
verage
A
verage
3KCFS6KCFS
Policy
.
.

47
1 Data Preparation
We collected extensive data for the Yolo Bypass in order to conduct an empirical analysis of the
proposed inundation scenarios. These include the following: (i) field-level geo-referenced crop
data and region definitions, (ii) crop yields and yield change based on planting date, (iii) crop
prices, (iv) costs of production, and (v) area inundated under 3,000 and 6,000 cfs flow volumes.
We review these data in the following section.
1.1 Land Use and Production Data
Production and land use data are summarized in the main text of this report, we provide a brief
summary in this section. Land use data are from a series of years, 2005-2009, of land use for
major crops, fallow land, and wetland in the Yolo Bypass. We identified 6 agricultural sub-
regions in the Yolo Bypass which represent homogeneous production conditions and form the
basis of the BPM. We used soil and climate data, in addition to interviews with Bypass farmers,
to develop homogenous agricultural sub-regions.
1.2 The DAYCENT Model
The DAYCENT model (DeGryze et al. 2009) is an agronomic model of field-level yields for
specific agricultural production regions. Johan Six and Juhwan Lee in the Plant Sciences
Department at UC Davis were responsible for model analysis and simulations.
The DAYCENT model calibrates to observed production conditions on a sub-set of fields in the
Yolo Bypass. The sub-set of fields is selected to represent heterogeneous production conditions
in the Bypass. The model is calibrated against data for corn, rice, safflower, sunflower,
processing tomato, alfalfa and mixed melons. The model does not explicitly simulate pasture so
we use alfalfa grown on a yearly rotation to proxy for irrigated pasture. Based on interviews with
farmers we determined that the yearly yield of dry pasture in AUM/acre is a fifth that of irrigated
pasture. The model does not simulate vine seed so we use the yield for mixed melons (honeydew
and watermelon) as a proxy for vine seed.
The DAYCENT model estimates the yield on any given field taking into account all production
conditions, including climate and date the crop was planted. We use the calibrated DAYCENT
model to estimate crop yields on a subset of fields in each of the 6 regions of the BPM. We
control for all other factors and allow the planting date to vary, thus the DAYCENT model
generates a series of data points, for each crop and region, of the expected yield given the crop
planting date.
1.3 Yield Functions Regression Analysis
We use the data points from the DAYCENT model results to estimate a single yield function, for
each crop and region. We fit this function using non-linear regression analysis which results in a
single function, for each crop and region in the Bypass, which relates crop yield to the planting
date. The yield response functions are included in the BPM.

48
We control for all other factors and specify yield as a function of the planting date. We estimate
the yield function by pooling all field observations, from the DAYCENT model, in each region
for the years 2005-2009. This is because we want to estimate the average yield response to the
planting date over a range of years rather than capturing yearly weather effects. The objective of
this study is to estimate the expected effects on agriculture due to increased flooding for fish
habitat and, as such, we do not want to capture weather or other effects in the yield response
functions.
For each crop i and region
g
, define ,ig
yas crop yield and ,ig
das the planting date. Note that the
planting date is the last day of over-topping plus region-specific drainage and preparation times.
Model parameters include 01
,, ,
,,and
α
ββ
ig ig ig. The estimated model for all crops except pasture is
defined as
0, 1, ,
,
,.
1ig ig ig
ig
ig d
ye+
=+
ββ
α
(0.0)
Pasture exhibits a different response than the other crops due to its resistance to delayed planting
date. We define the yield response function for pasture as
1, ,
,
,.
1ig ig
ig
ig d
ye
=+
β
α
(0.0)
We experimented with a series of functional forms for the yield response functions and
determined that the exponential provided the best fit of the data. Specifically, the AIC (and, AIC-
corrected for small sample sizes) indicated that the models in Equations 1.1 and 1.2 were the best
fit for the data.
We perform nonlinear regression analysis in Stata to generate parameter estimates. Not all crops
are gown in all regions, thus yield functions only apply to regions where crops are grown. Dry
and irrigated pasture have the same yield functions. Rice and wild rice have the same yield
functions. These simplifications are made because there is limited data availability for these
crops. The following tables summarize the parameter estimates and standard errors.
Table A1. Pasture Yield Function Parameter Estimates (standard errors in parentheses)
Pasture in Region Alpha Beta-0 Beta-1 Observations
5 0.900 2.784 -0.024 35
(0.350) (0.597) (0.009)
6 0.886 2.803 -0.025 35
(0.350) (0.602) (0.009)

49
Table A2. Corn Yield Function Parameter Estimates (standard errors in parentheses)
Corn in Region Alpha Beta-0 Beta-1 Observations
1 5.837 -32.354 0.222 43
(0.037) (12.347) (0.092)
2 5.905 -31.547 0.217 45
(0.031) (9.015) (0.067)
3 5.885 -31.247 0.214 45
(0.038) (10.278) (0.076)
4 5.731 -24.544 0.172 46
(0.081) (9.789) (0.073)
Table A3. Vine Seed (Melons) Yield Function Parameter Estimates (standard errors in parentheses)
Vine Seed in Region Alpha Beta-0 Beta-1 Observations
2 10.907 -5.012 0.032 37
(1.786) (1.197) (0.006)
3 8.871 -6.218 0.039 37
(1.811) (2.107) (0.010)
4 9.327 -5.544 0.036 37
(1.801) (1.576) (0.008)
Table A4. Rice Yield Function Parameter Estimates (standard errors in parentheses)
Rice in Region Alpha Beta-0 Beta-1 Observations
1 4.157 -19.492 0.132 54
(0.014) (1.065) (0.007)
2 4.160 -19.616 0.132 53
(0.015) (1.125) (0.008)
3 4.162 -19.571 0.132 53
(0.015) (1.111) (0.008)
4 4.140 -18.971 0.129 54
(0.016) (1.139) (0.008)
5 3.768 -22.392 0.154 47
(0.009) (1.614) (0.012)
6 3.821 -21.303 0.145 49
(0.008) (1.053) (0.007)

50
Table A5. Safflower Yield Function Parameter Estimates (standard errors in parentheses)
Safflower in Region Alpha Beta-0 Beta-1 Observations
1 1.472 -5.498 0.044 51
(0.244) (1.364) (0.008)
2 1.256 -8.812 0.059 51
(0.073) (1.501) (0.009)
3 1.531 -5.350 0.044 51
(0.272) (1.369) (0.008)
4 1.391 -5.830 0.046 51
(0.200) (1.360) (0.008)
5 1.278 -6.526 0.052 51
(0.311) (2.606) (0.016)
6 1.521 -5.429 0.045 51
(0.294) (1.487) (0.008)
Table A6. Sunflower Yield Function Parameter Estimates (standard errors in parentheses)
Sunflower in Region Alpha Beta-0 Beta-1 Observations
1 1.816 0.000 0.006 55
(0.077) (0) (0.000)
6 0.676 -5.104 0.025 55
(0.054) (1.968) (0.010)
Table A7. Processing Tomatoes Yield Function Parameter Estimates (standard errors in
parentheses)
Processing Tomatoes in Region Alpha Beta-0 Beta-1 Observations
1 39.29 -10.09 0.06 55
(0.536) (0.720) (0.004)
2 39.49 -10.09 0.06 55
(0.568) (0.756) (0.004)
3 39.68 -10.25 0.06 55
(0.557) (0.762) (0.004)
6 39.76 -8.44 0.05 55
(0.638) (0.592) (0.003)

51
Equations (1.1) and (1.2), and the parameter estimates in Tables A1-A7, show that the best fit of
the DAYCENT yield data is with a logistic-type functional form. Over a small range of planting
delay there is a small effect on yields. Yields decline at an increasing rate over some
intermediate range and, at some point, asymptote towards zero. Figures A3-A9 illustrate the
yield functions for each crop in an example region. Data points are in red, fitted functions in
blue.
Figure A3. Fitted Yield Function for Corn in Region 1

52
Figure A4. Fitted Yield Function for Pasture in Region 6
Figure A5. Fitted Yield Function for Rice in Region 2

53
Figure A6. Fitted Yield Function for Safflower in Region 1
Figure A7. Fitted Yield Function for Sunflower in Region 1

54
Figure A8. Fitted Yield Function for Processing Tomatoes in Region 3
Figure A9. Fitted Yield Function for Melons (Vine Seed) in Region 4

55
2 The Bypass Production Model (BPM) Calibration
We use the crop yield functions estimated from the DAYCENT model, plus additional economic
data, to calibrate the BPM. The BPM is the fundamental model of the analysis. The BPM relates
changes in crop yield and total affected acres to changes in agricultural production and,
fundamentally, changes in agricultural revenues. The BPM is a Positive Mathematical
Programming (PMP after Howitt, 1995) model of agriculture in the 6 regions of the Yolo
Bypass.
Note that a model is, by definition, a simplified representation of a real system. In the process of
abstracting and simplifying a real system a model loses some information; thus even with
theoretically consistent structure it is highly unlikely that a model will calibrate closely to
observed (base year) data. The problem is well documented in the agricultural production
modelling literature (Hazell and Norton 1986, Kasnakoglu 1990). One solution is to use
observed farmer behavior, in the form of observed land use patterns, and additional exogenous
information in order to calibrate the parameters of the structural model that exactly reproduce
observed base-year conditions. The method of Positive Mathematical Programming is a common
calibration method for structural agricultural production models (Howitt 1995), which we use in
the BPM.
2.1 Positive mathematical programming (PMP)
The BPM self-calibrates using a three-step procedure based on Positive Mathematical
Programming (PMP) (Howitt 1995) and the assumption that farmers behave as profit-
maximizing agents. A traditional optimization model would have a tendency for
overspecialization in production activities relative to what is observed empirically. PMP
incorporates information on the marginal production conditions that farmers face, allowing the
model to exactly replicate a base year of observed input use and output. Marginal conditions may
include inter-temporal effects of crop rotation, proximity to processing facilities, management
skills, farm-level effects such as risk and input smoothing, and heterogeneity in soil and other
physical capital. In the BPM, PMP is used to translate these unobservable marginal conditions, in
addition to observed average conditions, into region and crop-specific exponential cost functions.
Calibrating production models using PMP has been reviewed extensively in the recent literature.
Buyssee et al. (2007) and Heckelei and Wolff (2003) argue that shadow values from calibration
and/or resource constraints are an arbitrary source of information for model calibration.
Subsequent research suggests using exogenous information such as land rents instead of shadow
values (Heckelei and Britz 2005, Kanellopoulos et al. 2010). When multiple years of
observations are available Heckelei and Britz (2005) propose a generalized maximum entropy
formulation to estimate resource and calibration constraint shadow values. Merel and Bucaram
(2010) and Merel et al. (2011) propose calibration against exogenous supply elasticity estimates.
The BPM model is calibrated using traditional PMP with exogenous supply (acreage response)
elasticity information.

56
2.2 Model Calibration
PMP is fundamentally a three-step procedure for model calibration that assumes farmers
optimize input use for maximization of profits. In the first step a linear profit-maximization
program is solved. In addition to basic resource availability and non-negativity constraints, a set
of calibration constraints is added to restrict land use to observed values. In the second step, the
dual (shadow) values from the calibration and resource constraints are used to derive the
parameters for an exponential "PMP" cost function. In the third step, the calibrated model is
combined into a full profit maximization program. The exponential PMP cost function captures
the marginal decisions of farmers through the increasing cost of bringing additional land into
production (e.g. through decreasing quality).
The BPM framework requires that additional land brought into production faces an increasing
marginal cost of production. The most fertile land is cultivated first, additional land brought into
production is of lower “quality” because of poorer soil quality, drainage or other water quality
issues, or other factors that cause it to be more costly to farm. This is captured through an
exponential land cost function (PMP cost function) for each crop and region. The exponential
function is advantageous because it is always positive and strictly increasing, consistent with the
hypothesis of increasing land costs. The PMP cost function is both region and crop specific,
reflecting differences in production across crops and heterogeneity across regions. Functions are
calibrated using information from acreage response elasticities and shadow values of calibration
and resource constraints. The information is incorporated in such a way that the average cost data
(known data) are unaffected.
Formally, the exponential PMP cost functions are, for each crop i and region g, defined as
() ,
g
igi
x
gi gi gi
Cx e=
γ
φ
(0.0)
where gi
φ
and gi
γ
are parameters estimated by the PMP calibration routine described above and
gi
x
are total acres observed in production during the calibration base years.
The BPM calibrates to average observed land use between 2005 and 2009. We determined that
2005-2009 are representative of the full dataset (1984-2009) in terms of flood occurrence in the
Yolo Bypass and, as such, are representative of land use in 3,000 and 6,000 cfs affected areas of
the Bypass. Furthermore, detailed geo-referenced land use data were only available for 2005-
2009 in the Yolo Bypass. The histogram in Figure A10 shows that the sub-set of years which we
use for calibration (2005-2009) is representative of all years in the data (1984-2009) and, as
such, represents a reasonable set of years to use for model calibration. While the data do omit
some years of intermediate flood dates, Figure A10 shows that we capture the lower and upper
bounds of inundation reasonably well. As such, we feel that calibration to average 2005-2009
land use accurately reflects base conditions in the Bypass.

57
Figure A10. Histogram of Overtopping Date Frequencies (84-09 and 05-
09)
Standard calibration checks follow model calibration (see Howitt et al. 2012). These checks
verify that the base year of observed data is reproduced by the calibrated model and that
economic optimization requirements are satisfied.
We use a three year average of prices in the BPM, 2005-2007. These prices were determined to
be representative of the average production conditions between 2005 and 2009 and, as such, are
representative of the calibration data used in the model.
2.3 Profit Maximization Program Definition
The BPM solves for the cropping pattern that maximizes the agricultural profit across all regions
subject to regional land constraints and yield functions estimated from the DAYCENT data. Data
are as described previously. We assume the flood agency announces the policy it chooses for that
year (or series of years) before farmers make their planting decisions. Therefore, farmers know
the last day of overtopping for that year (with the exception of years where nature results in
overtopping past the policy date) and the yields associated with that planting date. The objective
function for the profit maximization program in the BPM is
⋅⋅− −
∑∑ ∑∑ ∑∑
γ
φ
max ,
ig ig
ig
x
iigig ig igig
xgi gi gi
py x e vcx (0.0)
where subscripts and variables are as previously defined, i
p are individual crop prices, and
ig
vc are region and crop-specific variable costs of production per acre. Yields ig
(y ) vary by
planting date, as defined above, according to the yield functions estimated with DAYCENT
model output as,

58
0, 1, ,
,
,,
1ig ig ig
ig
ig d
ye+
=+
ββ
α
ipasture,
∀
≠ (0.0)
and
01,
ig ig ig
d
ig ig
ye
+
=+
ββ
α
for ipasture
=
, (0.0)
where subscripts, variables, and parameters are as previously defined. Finally, land constraints in
each region are defined as
,
ig g
i
x
b≤
∑
,g
∀
(0.0)
where
g
bis the total number of acres (crop acres plus fallow) observed in each region.
In summary the procedure in the calibrated BPM model is to maximize Equation (1.4) subject to
Equations (1.5) - (1.7) by selecting the optimal crop mix, .
ig
x
Simulating the model over the base
calibration data reproduces the observed base allocation.

59
3 BPM Simulation
BPM model simulations proceed for two flow volumes separately: 3k CFS and 6k CFS, given
the calibrated model defined in Equations (1.4) - (1.7). we defined the simulation procedure in
the main text of the report, and repeat here for completeness.
Step 1: Run the BPM for a season with known overtopping dates at Fremont Weir, and flooding
in the Yolo Bypass. This represents the base condition (e.g. natural flooding) for agriculture in
the Bypass in the absence of the proposed policy flooding scenarios (for that year). Repeat Step 1
for a series of known years, there are 26 total.
Step 2: Over the same series of years as step two, run the BPM and impose (sequentially - one at
a time) the 12 proposed policy flooding scenarios. This represents what would have happened to
Bypass agriculture if the flooding policy was implemented in that year. Repeat Step 2 for the all
of the same years as Step 1.
Step 3: For each year simulated in Steps 1 and 2, calculate the difference in agricultural revenues
(and other outputs). Record the result for negative changes in revenue. Intuitively, we only want
negative changes in revenue because a positive change in revenue implies that the policy was
“better” than nature. For example, if natural flooding occurred in the Bypass until April 30th
then imposing a policy which stops overtopping at Fremont Weir on April 10th would not be
possible (i.e. it would increase revenues).
Step 4: Calculate the average loss of revenue (and other changes) across all of the years
simulated in Steps 1 - 3. This represents the expected impacts to agriculture due to the proposed
flooding scenarios, and is the fundamental output of the BPM.
The fundamental procedure of the BPM is to generate expected losses to agriculture by using the
calibrated model to estimate what would have happened under natural flooding, and then asking
what would have happened if a specific policy (last day of overtopping) was in place. This
procedure allows us to generate expected losses because we control for the expected natural
flood events in the Bypass. The following section illustrates this point.

60
4 BPM Output and Expected Losses
The final phase in the analysis is to use the BPM simulations to estimate the change in
agricultural gross revenues and acreage as a result of each of the policies (last overtopping date
for RPA, or CM2 scenario) under both flow volumes (3k and 6k CFS). We estimate regional
economic effects (jobs and income) using the IMPLAN model.
Economic losses are interpreted as expected annual losses in our analysis. The key assumption is
that the previous 26 year hydrology in the Yolo Bypass is representative of expected future
conditions. Specifically, natural overtopping at Fremont Weir will occur with the same expected
frequency, duration, and volume. There are two reasons these 26 years of data were identified as
reasonable, including (i) detailed flow information over the Fremont Weir was available for these
years, and (ii) it is representative of current hydrologic conditions in the Sacramento Valley
watershed. Older hydrologic information less accurately represents current conditions because it
does not account for changes in urban development and reservoir operations that have altered
flows in the Sacramento River over time. If better data become available we can revisit this
assumption.
The policy analysis output in the report is the average, over 26 years, of annual losses as
estimated by the individual policy scenarios in the BPM.
4.1 IMPLAN
The IMPLAN model estimates regional economic changes in production, value added,
employment, and tax receipts. Expected revenue losses from the BPM analysis represent direct
economic effects. However, upstream and downstream industries will be affected and some
agricultural workers will lose their jobs when production in the Bypass decreases. We use the
IMPLAN regional Input-Output model to estimate the direct, indirect, and induced effects of the
12 policy scenarios. The sum of these components represents the total effect of the policies.
IMPLAN is a multiplier model, which accounts for interrelationships among sectors and
institutions in the regional economy. The input-output representation of the economy was first
proposed by Leontief (1941). Production in this setting is assumed to occur by using fixed
proportions of factors, such that the same amount of a production input.
Coverage if the IMPLAN area for this study is exclusive to Yolo County. We used the NAICS
classification system and groped agricultural production into a single sector, NAICS 111. We
employed IMPLAN Professional Version 3 and a 2009 database for Yolo County.

61
This page intentionally blank.