ArticlePDF Available

Delineation of Management Zones using Multiple Crop Yield Data

Article

Delineation of Management Zones using Multiple Crop Yield Data

Abstract and Figures

Most agricultural fields naturally have within-field variations that may require site-specific attention for optimum crop performance. These variations are often contiguous, extensive and not easy to manage since every spot in a paddock is unique. Delineating a paddock into a few management zones is therefore a practical management approach. In this study, yield maps of various crops and seasons were used to delineate management zones. Map standardisation and the application of geographic information systems (GIS) were necessary to combine various yield maps for management zones delineation. A user-defined fuzzy set membership function was employed to standardise yield maps. The membership function used control points derived from the percentile calculation to rescale yield data continuously between 0 and 1. Map standardization and processing was performed within Idrisi-32 GIS and image processing software program. Standardised maps were combined within raster GIS to identify areas of low, medium and high yield performances. The classified areas were identified as management zones requiring different level of management attention. It was found that the continuous rescaling of yield data, using percentile based control points, was suitable to standardise yield maps. This process derived control points from within the yield data and minimized the influences of extreme yield values. Overlaying standardised yield maps to identify areas of low, medium and high yield potential was a straightforward process within GIS. The investigation revealed that most area performed moderately and the areas requiring highest management attention were around the periphery of the investigated paddock.
Content may be subject to copyright.
International Soil Tillage Research Organisation Conference
69
Delineation of Management Zones using Multiple Crop Yield
Data
Badri Basnet12, Rob Kelly2, Troy Jensen2, Wayne Strong2, Armando Apan2, and David Butler2
1 Currently at The University of Queensland Gatton, E-Mail: b.basnet@uq.edu.au 2 Department of
Primary Industries, Toowoomba, Queensland 3 The University of Southern Queensland, Queensland
Abstract: Most agricultural fields naturally have within-field variations that may require site-specific
attention for optimum crop performance. These variations are often contiguous, extensive and not easy
to manage since every spot in a paddock is unique. Delineating a paddock into a few ‘management
zones’ is therefore a practical management approach. In this study, yield maps of various crops and
seasons were used to delineate ‘management zones’. Map standardisation and the application of
geographic information systems (GIS) were necessary to combine various yield maps for
‘management zones’ delineation. A user-defined fuzzy set membership function was employed to
standardise yield maps. The membership function used control points derived from the percentile
calculation to rescale yield data continuously between 0 and 1. Map standardization and processing
was performed within Idrisi-32 GIS and image processing software program. Standardised maps were
combined within raster GIS to identify areas of low, medium and high yield performances. The
classified areas were identified as ‘management zones’ requiring different level of management
attention. It was found that the continuous rescaling of yield data, using percentile based control
points, was suitable to standardise yield maps. This process derived control points from within the
yield data and minimized the influences of extreme yield values. Overlaying standardised yield maps
to identify areas of low, medium and high yield potential was a straightforward process within GIS.
The investigation revealed that most area performed moderately and the areas requiring highest
management attention were around the periphery of the investigated paddock.
INTRODUCTION
Variations in crop yield are common in agricultural fields. These variations could be due to ground-,
climate- and input-related factors and their interactions (Yang et al., 1998). Crop yield is the
cumulative effect of all such variations. Considering crop as the sensor of its own environment (Legg
and Stafford, 1998), the crop yield data could be used to understand the field variations and manage it
in a site-specific manner. However, some of the yield influencing factors such as input and climate
vary significantly from season to season. Consequently, crop yield in a field could be quite different
from year to year even if the crop is same (Plant, 2001). Therefore, the yield of one season is not a
reliable indicator of the field variations.
The yield and field variations are extensive and contiguous, and the discrete quantification of
variations is not easy. On the other hand, the knowledge of widespread and adjoining field variations
is not going to make it easy to manage it in a site-specific manner since every part of a paddock could
be different. Therefore, the concept of a ‘management zone’ is important. A ‘management zone’ is
defined as a portion of a field that expresses homogeneous combinations of yield-limiting factors
(Doerge, 1998). Site-specific applications of agricultural inputs can be implemented by dividing a
field into ‘management zones’ (Zhang et al., 2002).
Accumulation of yield data from a number of seasons and crops in the same field can provide an
excellent means of gaining a precise understanding of the field and yield variations (Plant, 2001).
Since yield monitors are becoming a standard feature of a combine harvester and the crop yields are
routinely recorded at the time of harvest, yield maps of several crops and seasons could be combined
within a geographic information systems (GIS) framework to identify consistently high/low-yielding
areas within a paddock. These areas could then be labelled as ‘management zones’ requiring different
International Soil Tillage Research Organisation Conference
70
management practices. The number and size of management zones within a field could vary since it is
a function of the natural variability within the field, the size of the field and the ability of the farmer to
differentially manage regions within a field (Zhang et al., 2002).
However, combining yield maps of various crops and seasons require making them commensurable by
bringing to a common numeric range. This process is known as map standardisation. In GIS, map
standardisation can be achieved using discrete (ie. Boolean) or continuous (ie. fuzzy set) classification
methods (Burrough, 1996). The use of strict Boolean true/false logic is often too rigid for the
continuous nature of the variation (Burrough, 1989) such as yield data. Therefore, the Boolean
classification method is less appropriate since it tends to generalise yield data prior to the analysis. The
fuzzy membership function, on the other hand, defines the possibility of membership that varies
continuously from 0 (for completely outside the set) to 1 (for the full membership) range (Burrough et
al., 1992). Due to the continuous nature of the yield data, it could be appropriately rescaled between 0
and 1 using the fuzzy membership function.
The continuous rescaling of data can be achieved using a simple linear, sigmoidal or a J-shaped
membership function (Eastman, 1999). However, the crop yield data, which is often skewed and have
extreme values, do not seem to follow any of those pre-defined functions. Therefore, there is a need to
develop an appropriate crop yield data standardisation technique. The IDRISI-32 software has the
provision to user-define a rescaling function based on the use of control points (Eastman, 1999). The
control points could be derived from the crop yield data in a number of ways. However, in
determining control points, the interest is to identify a specific value and the percentage of the data
lying below and above this value. Such specific values could be conveniently determined using
fractiles calculation that divides data into percentile (ie 100 parts) or some other convenient division
such as quartiles and deciles (Blank, 1982).
Once standardised, the yield maps of various crops and seasons can be combined to a single layer
using spatial addition and averaging techniques. The combined map layer can then be reclassified into
potentially high, medium and low performing areas to correspond to ‘management zones’ requiring
low, medium and high attention, respectively. Therefore, the main objective of this study was to
process yield data of several crops and seasons to spatially delineate ‘management zones’ for site-
specific management. While the spatial processing of yield data is preconditioned to map
standardisation and overlay analysis, the fundamental objective of the study was to develop a yield
map standardisation technique.
MATERIALS AND METHODS
Study area
This study was conducted at the Kielli-2 paddock (Figure 1) near Jimbour (approx. 151°10’E and
27°00’S) in southern Queensland. The Kielli-2 paddock is about 40.5 ha in area and it is about 1555 m
long and 260 m wide. The landscape is relatively flat located on a heavy-clay Vertisol.
Yield measurement
Site-specific yield data for barley (1998) and sorghum (1999, 2000 and 2001) were collected during
the harvest of each crop using an impact plate AgLeader yield-monitor fitted in the grain harvester.
Point yield data were collected at 1-sec intervals (ranging from 0.5-4 m apart) along the path of the
harvester.
International Soil Tillage Research Organisation Conference
71
Figure 1. Site-specific yield data for barley (1998) and sorghum (1999-2001) at Kielli-2, Jimbour
Data pre-processing
Collected yield data, stored initially in a vector point file format, was spatially interpolated to a
common 10-m grid using a local variogram model within Vesper kriging program (Minasny et al.,
1999). Spatial interpolation was necessary to aggregate excessive yield data points and to compensate
for any missing data.
International Soil Tillage Research Organisation Conference
72
Data rescaling
Kriged output data was imported into the JMP statistical program (SAS Institute Inc, 1995) to
calculate deciles for 100, 99.5, 97.5, 90.0, 75.0, 50.0, 25.0, 10.0, 2.5, 0.05 and 0.0 percents. Deciles
divided the set of data into ten equal parts. Decile values and corresponding yield data for each crop
are shown in Table 1. Yield data and corresponding decile values were entered as control points within
IDRISI-based fuzzy set decision support module to continuously rescale yield data between 0 and 1
(Figure 2). The user-defined method of continuous rescaling was chosen.
Table 1. Decile values and corresponding crop yield data between 1998 and 2001 at Kielli-2, Jimbour
Chosen decile
values Barley 1998
(t/ha) Sorghum 1999
(t/ha) Sorghum 2000
(t/ha) Sorghum 2001
(t/ha)
1.000 4.58 7.17 9.85 8.02
0.995 3.53 6.30 9.37 7.60
0.975 3.17 6.01 9.18 7.26
0.900 2.95 5.67 8.83 6.95
0.750 2.77 5.31 8.45 6.69
0.500 2.57 4.70 8.01 6.39
0.250 2.37 3.78 7.42 6.04
0.100 2.08 2.82 6.77 5.54
0.025 1.57 1.89 6.16 3.72
0.005 1.06 0.74 3.66 0.54
0.000 0.73 0.00 1.21 0.00
Figure 2. Rescaled yield maps for barley (1998) and sorghum (2000) crops at Kielli-2, Jimbour
GIS Analyses
Rescaled maps were combined spatially using arithmetic operators such as addition and division.
Image calculator model, available within IDRISI32 was used to perform this operation. Average yield
values were calculated on a cell-by-cell basis and a map of average cell values was created. This map
International Soil Tillage Research Organisation Conference
73
was then classified into three (low, medium and high) classes. Classification was based on equal
interval method in which each class has an equal range of values or the difference between high and
low values is the same for each class (Mitchell 1999). For example, all the cells with cell-values
ranging from 0 to 0.33 were classified as ‘low’ class. Each class was considered as a separate zone for
site-specific management.
RESULTS AND DISCUSSIONS
It was established that user-defining a membership function, using decile-based control points, was
useful to rescale yield data continuously between 0 and 1. This method of standardization is coherent
because the decile calculation produced adequate statistically comprehensible control points from
within the dataset to use in the user-defined function. Defining a new membership function was
necessary because there was not any suitable pre-existing membership function for continuous
classification of yield data. Discrete classification of yield data was not considered as an option since
this method ignored partial class membership (Fridgen et al., 2000). Better rescaling accuracy would
have been possible through the use of percentiles (ie. 100 values) rather than deciles (ie. 10 values) as
control points since the membership between any two adjacent control points is linearly interpolated
(Eastman, 1999). However, ten data points were considered adequate at this initial stage of the work.
Combining standardized yield data of several crops and seasons (ie. 1998 to 2001) produced an
average yield map (Figure 3a). The classification of average yield map into three classes (Figure 3b),
based on equal data interval method, revealed that the area of low, medium and high yield
performance were 4.2 ha, 24.7ha, and 11.6 ha respectively. It was also revealed that the edges of the
paddock were performing inadequately (Figure 3b).
(a) (b)
Figure 3. (a) Average yield performance and (b) management requirements for the Kielli-2 field
International Soil Tillage Research Organisation Conference
74
The purpose of the above classification was to discriminate paddock on the basis of cumulative yield
performances and to identify management zones. Hence, each class represented a separate
management zone. A class with low yield performance logically required higher management
attention as compared to another class with higher yield performance. Thus, low, medium and high
classes represented management zones requiring high, medium and low attention, respectively (Figure
3b). The delineation of management zones could be based on factors such as soil and field
characteristics (Fridgen et al., 2000), digital elevation model (Pilesjo et al., 2000) and yield maps
(Stafford et al., 1999). Considering crop yield as the desirable end product of the plant production
process and the level of yield as the cumulative effect of all yield-influencing factors, it seems logical
to use yield maps as the basis to delineate management zones.
Three management zones were considered appropriate for the Kielli-2 paddock that has an overall area
of 40 ha. The exact number of management zones may however vary from year to year depending on
factors such as weather and crop conditions (Fridegen et al., 2000).
Low yield performance around the edges of the Kielli-2 paddock is prominent. This could be related to
the presence of compaction zones as machinery turn between runs. Alternatively, this may be an
artefact of the yield delay implicit in the creation of yield maps. In this study, most area (ie 24.7 ha or
61%) performed moderately and required medium level of management attention. However, the exact
area requiring moderate level of management attention may vary depending on the method of
classification. The result presented in this study is based on equal interval method of classification. If
equal area method were applied, the area falling under each zone would be one-third of the total.
CONCLUSIONS
This study combined yield data of various crops and seasons within a GIS framework to delineate
‘management zones’. Combining several yield maps and classifying the output into the areas of low,
medium and high yield performances enabled the identification of management zones. Yield map
standardization is essential to combine multiple yield data within a GIS framework. In this study, a
new yield map standardization method that incorporated a number of statistically derived control
points in a user-defined fuzzy membership function was developed. Delineation of management zone
is a preliminary work in this area because it does not field validated the management zones nor it links
with actual management practices. Therefore, the future works will directed towards these objectives.
Acknowledgements
The authors are grateful to the Queensland Department of Primary Industries and Grain Research and
Development Corporation for providing financial support to this study. Yield data was kindly provided
by Jamie Grant (Jimbour) through Bill Town (Wesfarmers Landmark).
References
Blank, L., Statistical Procedure for Engineering, Management and Science, International Student
Edition, McGraw Hill International Book Company, 33pp., Japan, 1982.
Burrough, P.A., Fuzzy Mathematical Methods for Soil Survey and Land Evaluation, Journal of Soil
Science, 40, 477 - 492, 1989.
Burrough, P.A., Natural Objects with Intermediate Boundaries: Geographic Objects with Intermediate
Boundaries, EDs: P.A. Burrough and A.U. Frank, Taylor and Francis Publishers, London, 1996.
Burrough, P.A., R.A. MacMillan, and W. van Deursen., Fuzzy Classification Methods for
Determining Land Suitability from Soil Profile Observations and Topography, Journal of Soil
Science, 43, 193 210, 1992.
Doerge, T., Defining Management Zones For Precision Farming, Crop Insights, 8, 21, 1998.
Eastman, J.R., IDRISI 32 Guide to GIS and Image Processing, Volume 2, Clark Labs, Clark
University, 1 50 pp., 1999.
International Soil Tillage Research Organisation Conference
75
Fridgen, J.J., C.W. Fraisse, N.R. Kitchen, and K.A. Sudduth, Delineation and Analysis of Site-specific
Management Zones, Second International Conference on Geo-spatial Information in Agriculture
and Forestry, Florida, January 10-12, 2000.
Legg, B.J., and J.V. Stafford, Precision Agriculture New Technologies, Proceedings of the Brighton
Crop Protection Conference, 1998.
Minasny, B., A.B. McBratney, and B.M. Whelan, VESPER by Australian Centre for Precision
Agriculture, The University of Sydney, (http://www.usyd.edu.au/su/agric/acpa), Version 1, 1999.
Mitchell, A., The ESRI Guide to GIS Analysis, Volume 1: Geographic Patterns and Relationships,
Environmental Systems Research Institute Inc, ESRI Press, 37 67 pp., NY, 1999.
Plant, R.E., Site-Specific Management: The Application Of Information Technology to Crop
Production, Computers and Electronics in Agriculture, 30, 9 29, 2001.
Pilesjo, P., L. Thylen, A. Persson, Digital Elevation Models for Delineation of Agricultural
Management Zones, Second International Conference on Geo-spatial Information in Agriculture
and Forestry, Florida, January 10-12, 2000.
SAS Institute Inc., Statistical Discovery Software JMP Statistics and Graphics, Copyright 1995 by
SAS Institute Inc., NC, 1995.
Stafford, J.V., R.M. Lark, H.C. Bolam, Using Yield Maps to Recognise Fields into Potential
Management Units, Proceedings of the Fourth International Conference on Precision Agriculture,
American Society of Agronomy, Madison. 1999.
Yang, C., C.L. Peterson, G.J. Shropshire, and T. Otawa, Spatial Variability of Field Topography and
Wheat Yield in the Palouse Region of the Pacific Northwest, Transactions of the ASAE, 41, 17
27, 1998.
Zhang, N., M. Wang, and N. Wang, Precision Agriculture a Worldwide Overview, Computers and
Electronics in Agriculture, 36, 113 132, 2002.
... Within-field crop yield variability has been reported even in well-managed fields (Amado et al. 2007). This observation is attributed to zones with different soil types, soil quality differences within the same soil type, past soil erosion, microclimate, input-related factors and the complex interactions between all previous factors (Basnet et al. 2003). Fiorin et al. (1997) reported that the A horizon depth of an Ultisol in central Rio Grande do Sul State was the main driver for non-irrigated corn (Zea mays L.) yield variability. ...
... In addition, slope length has also been reported as likely to be partly responsible for crop yield variability in the rolling fields that are typical of Southern Brazil (Alba et al. 2011). Homogeneous areas with different crop yield performances have been used to delineate management zones as a consequence of the frequency and intensity of plant growth limiting factors (Basnet et al. 2003;Amado et al. 2009;Neto et al. 2011). In this sense, the use of yield maps of various crops and seasons becomes a key tool for understanding withinfield variability (Zhang et al. 2002;Molin 2002;Milani et al. 2006;Amado et al. 2007) and establishing site-specific interventions. ...
... In this sense, the use of yield maps of various crops and seasons becomes a key tool for understanding withinfield variability (Zhang et al. 2002;Molin 2002;Milani et al. 2006;Amado et al. 2007) and establishing site-specific interventions. Basnet et al. (2003) suggested that low crop performance zones should have priority in soil management. ...
Article
Precision agriculture relies on site-specific interventions determined by the spatial variability of factors driving plant growth. The main objective of this study was to assess the efficiency of variable-rate seeding of corn (Zea mays L.) with delineated management zones. This study involved two experiments carried out in Na˜o-Me-Toque, Rio Grande do Sul, Brazil. For the first experiment, carried out in 2009/2010, management zones were delineated by the farmer’s knowledge of the crop field. The field was split into low (LZ), medium (MZ) and high (HZ) crop performance zones. In the second experiment, carried out in 2010/2011, management zones were delineated by overlaying standardized yield data from nine crop seasons (seven of soybean and two of corn). The experiment was carried out with a randomized block design with three management zones and five corn seeding rates ranging from 50 000 to 90 000 seeds per ha-1. The soil was a Rhodic Hapludox with a subtropical climate. Optimization of the corn plant population within the field increased grain yield compared to the reference plant population (70 000 plants ha-1). Yield increases in the LZ, due to corn plant population reduction in relation to the target population, were 1.20 and 1.90 Mg ha-1 for first and second experiments, respectively. This resulted in economic gains of 19.8 and 28.7 %, respectively. Yield increases in the HZ were 0.89 and 0.94 Mg ha-1, respectively, and were due to an increase in plant population in relation to the target population. This resulted in economic gains of 5.6 and 6.6 % for the first and second experiments, respectively. In the MZ, the adjustment of the target plant population was not necessary. Optimizing corn population according to management zones is a promising tool for precision agriculture in Southern Brazil.
Article
Full-text available
The rate of soil degradation is increasing in the Himalayan ecosystem and inducing soil nutrient loss due to numerous environmental effects. The site-specific soil management zones (MZs) are necessary for such terrain with variability in spatial soil nutrient distribution. The present study investigates the spatial variability of soil nutrients and delineates the MZs in a mid-Himalayan watershed. Overall, 100 m grid 65 surface soil samples (0–30 cm) were collected from an area of 102 ha mini watershed located in the Tehri Garhwal district of Uttarakhand state in India. The samples were processed and analyzed for soil pH, electrical conductivity (EC), aggregate stability (AS), clay content (Cl), total carbon (TC), total nitrogen (TN), available phosphorous (AP) and available potassium (AK). The soil pH, EC, AS, Cl, TC, AP, AK and TN had mean values of 5.15, 104.57 dS cm⁻¹, 0.80, 17.78%, 2.55%, 32.14 kg ha⁻¹, 163.59 kg ha⁻¹ and 0.24%, respectively. Geostatistical analysis showed a spatial distribution pattern of soil variables with moderate to weak spatial dependency. The principal component analysis (PCA) divulged five principal components (PCs) with eigenvalue > 1 with 69.48% total variance. The fuzzy c means algorithm for the scores of the selected PCs was carried out to establish the optimum number of clusters, i.e., management zones (MZs). The geostatistical analysis, PCA and fuzzy clustering resulted in four optimum soil management zones. The analysis of variance indicated that there is a significant difference between the soil and terrain parameters among the MZs. The soil pH and TN among MZ1 and MZ4, whereas the EC, AK, TC and elevation were significantly different among all the delineated MZs. In addition, the delineated MZs using cluster analysis resulted in within-zone variability and could be used as a guide for farmers to adopt site-specific soil management.
Chapter
Nematode distribution varies significantly throughout a field and is highly correlated to soil texture and other edaphic factors. Field-wide application results in nematicides being applied to areas without nematodes and the application of sub-effective levels in areas with high nematode densities. Efforts to use grid maps as a guide to site-specific application have proven to be too expensive to be cost effective. Recently, the availability of GPS –GIS has allowed the use of soil electrical conductivity systems to rapidly and inexpensively develop cost effective soil texture maps. These maps are used to project where nematodes are likely to occur within a field. Variable-rate application systems for granular and fumigant nematicides have been developed and tied via software to soil texture maps providing a mechanism for the effective delivery of nematicides in a site-specific , variable-rate manner in individual fields. Efforts in South Carolina, Georgia, and Arkansas are further developing this system and refining our knowledge of how soil texture and other edaphic factors affect the distribution of cotton nematodes .
Article
The main objective of this thesis is to examine and discuss if an improved fanning strategy, in an arid area, affects the maize yield significantly. The basis for the discussion is the "case study" and the spatial (GIS)-analysis of the farming practices and the maize yields. The study was performed in Makanya in the north of Tanzania. The result from the GIS-analysis indicates that there is a high correlation between water supply and biomass production. The GIS-maps also show that the farmers adjacent the river outlet (high water supply) invests more resources in farming methods compared to fanners far from the outlet. The increase of yield due to fanner input, where the water is insufficient, is however minor. The results from the "case study" illustrate nevertheless the importance of proper farming strategies if the water supply is sufficient. The findings do not verify the general statements about a significant increase in biomass production with the introduction of judicious fanning methods. This is due to the water constraints in Makanya.
Article
Site-specific management requires the development of agronomic strategies for sub-field management zones that are subject to a unique combination of potential yield-limiting factors. Creation of unique management areas has focused mainly on soil type surveys, yield mapping, and soil fertility management based on grid soil sampling. This paper discusses the use of the fuzzy k-means unsupervised continuous clustering algorithm to delineate within-field management zones using soil and field characteristics (i.e., landscape features). Soil electrical conductivity and topographic attributes such as elevation and slope measured in two claypan soil fields were used for the clustering process. Measures of cluster performance indicated no advantage of dividing these fields into more than four or five management zones. Classification using this procedure explained 10 to 35% of the variation in grain yield. Year to year differences in the appropriate number of management zones was attributed to weather and crop type.
Article
Measurement and analysis of the spatial variability of fields are two important aspects of site-specific crop management. In this study, data for wheat yield, ground slope and aspect were intensively collected from five fields in the Palouse region of Washington using a set of sensors and a Global Positioning System (GPS) receiver. A Geographic Information System (GIS) database was developed to manage the field-collected data and information from topographic maps and to group the data from each field into different regions for analysis. Relative elevation differences across the five fields ranged from 29 to 70 m. Slope steepness ranged from 0 to 30°with field means of 9 to 12°. Geostatistical analyses showed that the spatial patterns of variability in wheat yields differed not only from one field to another but also from one region to another within a field. The ranges of influence in wheat yield for the five whole fields were from 55 to 86 m, while those for the within-field regions varied from 20 to over 100 m. Regression analyses indicated that topographic attributes including elevation, slope and aspect had significant effects on wheat yield. These topographic attributes could explain 13 to 35% of the variability in wheat yield for the whole fields, though 49 to 84% of the yield variability could be explained by topography in some regions within the five fields. These results provided useful quantitative information about the influence of topography on crop yield in the Palouse region.
Article
The rigid-data model consisting of discrete, sharply bounded internally uniform entities that is used in hierarchical and relational databases of soil profiles, choropleth soil maps and land evaluation classifications ignores important aspects of reality caused by internal inhomogeneity, short-range spatial variation, measurement error, complexity and imprecision. Considerable loss of information can occur when data that have been classified according to this model are retrieved or combined using the methods of simple Boolean algebra available in most soil and geographical information systems. Fuzzy set theory, which is a generalization of Boolean algebra to situations where data are modelled by entities whose attributes have zones of gradual transition, rather than sharp boundaries, offers a useful alternative to existing methodology. The basic principles of fuzzy sets, operations on fuzzy sets and the derivation of membership functions according to the Semantic Import Model are explained and illustrated with data from case studies in Venezuela and Kenya.
The ESRI Guide to GIS Analysis Geographic Patterns and Relationships, Environmental Systems Research Institute Inc
  • A Mitchell
Mitchell, A., The ESRI Guide to GIS Analysis, Volume 1: Geographic Patterns and Relationships, Environmental Systems Research Institute Inc, ESRI Press, 37 – 67 pp., NY, 1999.
VESPER by Australian Centre for Precision Agriculture
  • B Minasny
  • A B Mcbratney
  • B M Whelan
Minasny, B., A.B. McBratney, and B.M. Whelan, VESPER by Australian Centre for Precision Agriculture, The University of Sydney, (http://www.usyd.edu.au/su/agric/acpa), Version 1, 1999.