ArticlePDF Available

Geostatistics-based spatial distribution of soil moisture and temperature regime classes in Mazandaran province, Northern Iran

Authors:

Abstract and Figures

Soil moisture regime (SMR) and soil temperature regime (STR) classes as soil classification criterions are required by US Soil Taxonomy because they affect genesis, use, and management of soils. The lack of sufficient soil moisture and temperature data requires the characterization of the pedoclimate on the basis of climatic data processed by simulation models. This research was conducted to consider the new approach for SMR and STR mapping. The objectives of this study were to compare the four interpolation schemes including ordinary kriging (OK), cokriging (Co-K), inverse distance weighting, and conditional simulation for interpolating the monthly mean total precipitation (MMTP) and monthly mean air temperature (MMAT) and to apply the Java Newhall simulation model for the MMTP and MMAT predictive values at each node of 1 km2 grids across the Mazandaran province, northern Iran, for delineating the SMR and STR classes. The semivariogram analyses showed moderate to strong spatial dependence of data sets. The accuracy of interpolators varied within months for both MMTP and MMAT data sets. In most cases, OK and Co-K methods had the highest accuracy with lower mean error, root mean square error, and higher concordance correlation coefficient. The predictive maps show high diversity of SMR classes including Aridic, Ustic, Udic, and Xeric. The STR classes comprise Mesic, Thermic, and Cryic regimes. Results herein indicated that geostatistical approaches can potentially provide the opportunity for mapping of SMR and STR classes in data scarce regions.
Content may be subject to copyright.
This article was downloaded by: [Hacettepe University]
On: 14 July 2015, At: 22:41
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: 5 Howick Place, London, SW1P 1WG
Click for updates
Archives of Agronomy and Soil Science
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/gags20
Geostatistics-based spatial distribution
of soil moisture and temperature
regime classes in Mazandaran province,
northern Iran
Mostafa Emadia, Ali Reza Shahriarib, Fardin Sadegh-Zadeha, Bahi
Jalili Seh-Bardana & Ali Dindarlouc
a Department of Soil Science, College of Crop Sciences, Sari
Agricultural Sciences and Natural Resources University, Sari, Iran
b Faculty of Natural Resources, University of Zabol, Zabol, Iran
c College of Agriculture and Natural Resources, Persian Gulf
University, Bushehr, Iran
Accepted author version posted online: 23 Jun 2015.Published
online: 13 Jul 2015.
To cite this article: Mostafa Emadi, Ali Reza Shahriari, Fardin Sadegh-Zadeh, Bahi Jalili Seh-Bardan
& Ali Dindarlou (2015): Geostatistics-based spatial distribution of soil moisture and temperature
regime classes in Mazandaran province, northern Iran, Archives of Agronomy and Soil Science, DOI:
10.1080/03650340.2015.1065607
To link to this article: http://dx.doi.org/10.1080/03650340.2015.1065607
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Downloaded by [Hacettepe University] at 22:41 14 July 2015
Geostatistics-based spatial distribution of soil moisture and
temperature regime classes in Mazandaran province,
northern Iran
Mostafa Emadi
a
*, Ali Reza Shahriari
b
, Fardin Sadegh-Zadeh
a
, Bahi Jalili Seh-Bardan
a
and Ali Dindarlou
c
a
Department of Soil Science, College of Crop Sciences, Sari Agricultural Sciences and Natural
Resources University, Sari, Iran;
b
Faculty of Natural Resources, University of Zabol, Zabol, Iran;
c
College of Agriculture and Natural Resources, Persian Gulf University, Bushehr, Iran
(Received 7 February 2015; accepted 9 June 2015)
Soil moisture regime (SMR) and soil temperature regime (STR) classes as soil
classification criterions are required by US Soil Taxonomy because they affect genesis,
use, and management of soils. The lack of sufficient soil moisture and temperature data
requires the characterization of the pedoclimate on the basis of climatic data processed
by simulation models. This research was conducted to consider the new approach for
SMR and STR mapping. The objectives of this study were to compare the four
interpolation schemes including ordinary kriging (OK), cokriging (Co-K), inverse
distance weighting, and conditional simulation for interpolating the monthly mean
total precipitation (MMTP) and monthly mean air temperature (MMAT) and to apply
the Java Newhall simulation model for the MMTP and MMAT predictive values at
each node of 1 km
2
grids across the Mazandaran province, northern Iran, for delineat-
ing the SMR and STR classes. The semivariogram analyses showed moderate to strong
spatial dependence of data sets. The accuracy of interpolators varied within months for
both MMTP and MMAT data sets. In most cases, OK and Co-K methods had the
highest accuracy with lower mean error, root mean square error, and higher concor-
dance correlation coefficient. The predictive maps show high diversity of SMR classes
including Aridic, Ustic, Udic, and Xeric. The STR classes comprise Mesic, Thermic,
and Cryic regimes. Results herein indicated that geostatistical approaches can poten-
tially provide the opportunity for mapping of SMR and STR classes in data scarce
regions.
Keywords: soil moisture regime; soil temperature regime; geostatistics; mapping; Soil
Taxonomy
Introduction
Climate is one of the most important soil forming factors that affect soil processes and
properties. Soil climate, i.e., pedoclimate is directly impressed by the climate character-
istics. The pedoclimate term is the record of temporal patterns of soil moisture and
temperature. It is an important component of the structure of US Soil Taxonomy
(USDA 1999). Climate information such as monthly precipitation and air temperature
are implemented through the concept of soil moisture regime (SMR) and soil temperature
regime (STR) classes. In US Soil Taxonomy, SMR and STR classes are typically
*Corresponding author. Email: mostafaemadi@gmail.com
Archives of Agronomy and Soil Science, 2015
http://dx.doi.org/10.1080/03650340.2015.1065607
© 2015 Taylor & Francis
Downloaded by [Hacettepe University] at 22:41 14 July 2015
calculated through the use of well-known models such as Newhall (Van Wambeke 1986;
Newhall & Berdanier 1996), Billaux (Billaux 1978), International Committee on Soil
Moisture and Temperature Regimes (ICOMMOTR 1991), and Erosion Productivity
Impact Calculator (EPIC) (Costantini et al. 2002). The Newhall model estimates the
pedoclimate through the simple calculation schemes, based on the air temperature and
precipitationevapotranspiration balance.
In soil classification scheme of the World Reference Base for Soil Resources (WRB)
(IUSS Working Group WRB 2014), the pedoclimate is indirectly taken into account, and
as a result, they do not classify SMR and STR classes properly. However, the US Soil
Taxonomy is an extensive soil classification system applied in Iran as in the world
requires pedoclimate classification. Esfandiarpour et al. (2013) explained that more
accurate descriptions of the soils of arid and semi-arid regions, especially in southern
Iran, make the WRB system (IUSS Working Group WRB 2014) preferential to Soil
Taxonomy for soil survey purposes. However, almost all current conventional soil maps
in Iran were surveyed by Soil Taxonomy, emphasizing the needed attempts to update
these maps. The recent studies in Iran (Jafari et al. 2014; Pahlavan-Rad et al. 2014;
Taghizadeh-Mehrjardi et al. 2014) tried to update the conventional soil maps through
digital soil mappings. Meanwhile, the special attentions to the soil moisture and tempera-
ture regime maps that play important roles in Soil Taxonomy classification need to be
taken into account. The lack of climatic data availability, especially long-term data set, is a
major problem in many parts of Iran that made the researcher to characterize the
pedoclimate on the basis of climatic data processing by mathematical and geostatistical
approaches.
In recent years, the geostatistical methods have also received increasing interest by
soil scientists and agricultural engineers (Emery & Ortiz 2007; Emadi et al. 2010;
Cambule et al. 2014; Emadi & Baghernejad 2014). Spatial distributions of air temperature
and precipitation have been studied for a considerable time, although relatively few of
these studies have been conducted for soil taxonomic purposes (Ishida & Kawashima
1993; Lapen & Hayhoe 2003; Lloyd 2005; Benavides et al. 2007; Carrera-Hernandez &
Gaskin 2007; Bostan et al. 2012;Wu&Li2013). Benavides et al. (2007) demonstrated
that in spite of the scarce and irregularly distributed number of sampled data for air
temperature in a mountainous region of Northern Spain, the models derived from geos-
tatistical techniques with elevation as an auxiliary variable were quite accurate. Spatial
distribution of precipitation has also been applied on several geostatistical studies (Lloyd
2005; Symeonakis et al. 2009; Harris et al. 2010).
Mazandaran province borders Caspian Sea in northern Iran. Varying rates of precipita-
tion and temperature in different parts of Mazandaran province yielded a variety of
climates including the mild and humid climate of Caspian Sea shoreline and the moderate
and cold climate of mountainous regions. The soil nutrient cycling in northern part of Iran,
especially Mazandaran province, is affected by the descending precipitation gradient and
led to some differences in soil physical, chemical, and mineralogical characteristics
(Khormali & Kehl 2011; Emadi et al. 2012; Khormali et al. 2012). Although the climate
is the main factor of soil genesis in this region through a variety of direct and indirect
processes, however, only a few studies have considered it so far (Bahmaniar et al. 1999;
Khormali & Kehl 2011; Emadi et al. 2012).
Due to the lack of studies on spatial distribution of precipitation and air temperature
for soil taxonomic purposes in Mazandaran province, this research was conducted to
consider the new approach for SMR and STR mapping. The objectives of this study were
to (i) compare the several interpolation schemes for predicting the monthly mean total
2M. Emadi et al.
Downloaded by [Hacettepe University] at 22:41 14 July 2015
precipitation (MMTP) and monthly mean air temperature (MMAT) over the province; (ii)
apply the Java Newhall Simulation Model (JNSM) as a soil climate simulation model onto
the estimated atmospheric climate data set; and (iii) assign and map the SMR and STR
classes across the Mazandaran province, northern Iran.
Materials and methods
Description of study area and data collection
This study was carried out in Mazandaran province, northern Iran. Mazandaran province
with the area of 23,756 km
2
is surrounded by the Alborz Mountain ranges in the south and
the Caspian Sea in the north. According to the Domartens classification, the western,
central, and eastern parts of Mazandaran province have a very humid, humid, and
Mediterranean climate, respectively. The annual precipitation varied from about
400 mm in Amiriye, Baladeh, and Siyah-Bisheh areas in south up to about 1350 mm in
Ramsar, Abbasabad, and Nooshahr cities in western part of the province. The annual
mean air temperature in different areas ranged from approximately 10°C up to about 18°
C. Elevation gradually increases from the coastal areas ~0 m above sea level up to about
20003500 m above sea level in highlands of Alborz mountain ranges. Mount Damavand
with 5671 m altitude is located in south of Mazandaran province.
Soil moisture classification, according to Soil Taxonomy (USDA 1999), is based on a
yearly assessment of the number of days in which the soil moisture control section is
moist, partially dry, or completely dry. Older versions of Soil Taxonomy evaluated
pedoclimate by considering most yearsor six or more out of ten years. The latest
edition of Soil Taxonomy (SSS 2014) replaces this approach with the concept of normal
year in which the annual precipitation is plus or minus one standard deviation of the long-
term (30 years or more) mean annual precipitation. The mean monthly precipitation is also
plus or minus one standard deviation of the long-term monthly precipitation for 8 of the
12 months. In the present study as assumed by Costantini et al. (2002), the term normal
yearof older versions of Soil Taxonomy was used because if we considered the criteria
of latest edition of Soil Taxonomy, only a few data set of weather stations can be defined
as normal. Therefore, the MMTP and MMAT data sets of 53 and 78 weather stations in
Mazandaran province were selected for considering the spatial distribution of temperature
and precipitation over the province, respectively. The MMTP and MMAT data sets
collected from the weather stations ranges from 10 to 43 years.
Java Newhall Simulation Model
The JNSM (Waltman et al. 2011; JNSM 2012) was developed in response to a USDA-
NRCS need to better understand soil climate in soil survey. The JNSM is an update to a
traditional soil climate simulation model called Newhall Simulation Model (NSM) by
Franklin Newhall. The mechanics of the model were not changed in the Java version, but
internal calculations and software architecture were updated and made more efficient
(Waltman et al. 2011; Winzeler et al. 2012). The NSM was used to simulate seasonal
water balance patterns and calendars for soil moisture in the calculated soil moisture
control section. More details about NSM and JNSM can be found in Van Wambeke
(1986), Newhall and Berdanier (1996), and Winzeler et al. (2012). The subclass modifiers
(Tentative Subdivisions of Moisture Regimes) as proposed in Newhall source code (Van
Wambeke 1986; JNSM 2012) was used in this study. The moisture subclasses are not used
Archives of Agronomy and Soil Science 3
Downloaded by [Hacettepe University] at 22:41 14 July 2015
in SSS (2014). The most important data inputs of the JNSM are the MMTP and MMAT
data sets. We used the geostatistical techniques to predict the MMTP and MMAT values at
the nodes of a 1 km
2
grid from the sparsely weather data sets (Figure 1). Due to the lack
of soil survey in the Mazandaran province, default values of 2.5°C (4.5°F) and 200 mm
(7.874 inches) were used for the offset between mean annual air temperature and mean
annual soil and the available water capacity, respectively. The default value of 2.5°C for
the offset was also used by Winzeler et al. (2012).
Geostatistical analysis
The statistically abnormal distribution of data sets can endanger the spatial continuity of
the semivariogram function, as well as influence the prediction accuracy. Therefore, data
transformation is necessary to normalize the data sets. Logarithmic transformation is often
applied in order to normalize positively skewed data sets (McGrath et al. 2004). However,
data sets do not always follow the log-normal distribution (Zhang & Selinus 1998)asin
this study. The KolmogorovSmirnov (KS) test was employed to test the normality of
data sets, and if the data did not meet the requirements of a normal distribution, they were
transformed into a normal distribution using the Logarithmic or BoxCox transformation.
Figure 1. Location of the weather station sites. + and Оindicate the sites that temperature and
precipitation data sets were used, respectively.
4M. Emadi et al.
Downloaded by [Hacettepe University] at 22:41 14 July 2015
The BoxCox transformation is a likelihood-maximizing power transform that gives
skewed data a more normal distribution, thereby stabilizing variation (Box & Cox
1964; Li et al. 2013). It is performed by maximizing the likelihood function over a
power variable, λ. Equation (1) indicates the transformation function as follows:
y¼xλ1
λλÞ0
ln xðÞλ¼0
(1)
where yis the transformed value and xis the value to be transformed. λs estimation was
performed during the data sets preprocessing, using Develve software.
Geostatistics comprise a set of techniques and estimators which use the spatial
variability and correlation of a continuous spaced-distributed phenomenon to predict at
unsampled locations. They consist generally of two steps: a preliminary data exploratory
and structural analysis of the information in order to describe the spatial variability of the
variable and the spatial prediction at unsampled points. The spatial variability of normal-
ized MMTP and MMAT data sets were investigated using a semivariogram model.
Semivariance was calculated using Equation (2):
γhðÞ¼ 1
2NhðÞ
X
N
i¼1
ZxiðÞZxiþhðÞ½
2(2)
where γ(h) is the semivariogram at lag h,his the lag distance, Zis a random variable, Z
(xi) is random variable for a fixed location xi,N(h) is the number of pairs of values Z(xi)
and Z(xi +h) is a separated value by a vector h. A semivariogram model can be used to
indicate both structural and stochastic aspects of a variable. Common semivariogram
models include the Spherical, Exponential, and Gaussian models which fit the experi-
mental data (Pannatier 1993; Huisman et al. 2002).
Four commonly used interpolation schemes in estimating MMTP and MMAT were
ordinary kriging (OK), inverse distance weighting (IDW), conditional simulation (CS),
and cokriging (Co-K) with elevation as covariate.
IDW interpolation estimates are made based on values at nearby locations weighted
only by distance from the interpolation location. Greater weighting values are assigned to
values closer to the interpolated point (Goovaerts 1999; Bilgili 2013). OK is an inter-
polation method based on values at neighboring locations plus knowledge about the
underlying spatial relationships in a data set (Bilgili 2013; Emadi & Baghernejad 2014).
CS interpolation method is based on a form of stochastic simulation in which data values
are honored at their locations. This means that local details are not obscured by smoothing
like in kriging (Gamma Design Software 2004). OK and IDW smooth out local details of
spatial variation, especially as interpolated locations become more distant from measured
locations. Co-K is an important basic geostatistical method because it takes other poten-
tially important variables into consideration as covariates in the estimation. The elevation
data set used in this study included elevations of the climate stations plus gridded
(1 km × 1 km) elevations at non-station locales obtained from digital elevation models.
More detailed description of the four interpolation techniques are given in Isaaks and
Srivastava (1989), Webster and Oliver (2001), and Goovaerts (1999).
To compare the accuracy of four interpolation methods, the MMTP and MMAT data
sets were randomly divided into two subsets containing 85% and 15% of the total data as
training and testing, respectively. Estimation techniques were evaluated using the root
Archives of Agronomy and Soil Science 5
Downloaded by [Hacettepe University] at 22:41 14 July 2015
mean square error (RMSE), mean error (ME), and Lins concordance correlation coeffi-
cient (CCC) (Lin 1989) between the measured and predicted values of samples in the
testing data set as an indicator of estimation error. The CCC indicates the degree to which
pairs of the measured and estimated TMP and MMAT values fall on the 45° line through
the origin. RMSE, ME, and CCC are calculated with Equations (3)(5) as follows:
ME ¼1
nX
n
i¼1
yyiðÞ (3)
RMSE ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
nX
n
i¼1
yyiðÞ
2
s(4)
CCC ¼2ρσxσy
σ2
xþρ2
yþμxμy

2(5)
where yis the measured value; yi is the estimated value, ρis the Pearson correlation
coefficient between the measured and estimated values, σ
x
2
and σ
y
2
are the corresponding
variances of measured and estimated values, and μ
x
and μ
y
are the means for the measured
and estimated values.
Implementation of JNSM for assigning the SMR and STR classes
The required inputs of JNSM were 12 MMTP values, 12 MMAT values, offset between
mean annual air temperature and mean annual soil, available water capacity, elevation,
latitude, and longitude. The best interpolation technique for predicting the MMTP and
MMAT values were used to estimate the values on the 1 km × 1 km square grids across
the Mazandaran province. The excel spreadsheet data set were then populated with values
for each of the inputs required by the JNSM for each node, which spatially referenced data
exist. A CSV file was then created from an Excel file and imported into the JNSM with
the Batch Model Run. The JNSM was run individually for each node at 1 km interval. The
model outputs were in XML files. The XML2CSV application was used to consolidate the
XML files output from a JNSM batch run into a single spreadsheet file for the sake of
further analysis. The SMR and STR classes were then aggregated from the single
spreadsheet file and classified for a thematic map using ILWIS 3.2 software.
Figure 2 gives a flowchart showing the methodology applied in this study. All
conventional statistical analyses were conducted using the software package SPSS 13.0
for Windows (SPSS Inc., MatLab R, Armonk, NY, USA). The geostatistical computa-
tions were also performed using the GS+ software (Gamma Design Software LLC.,
Plain well, MI, USA). ILWIS 3.2 software (ITC, Enschede, The Netherlands) was used
for mapping.
Results and discussion
Exploratory data analysis
The summary statistics of MMTP and MMAT data sets are given in Tables 1 and 2,
respectively. The mean MMTP across the Mazandaran province was 64.2 mm, with the
6M. Emadi et al.
Downloaded by [Hacettepe University] at 22:41 14 July 2015
Figure 2. The flowchart of the methodology used in this study.
Archives of Agronomy and Soil Science 7
Downloaded by [Hacettepe University] at 22:41 14 July 2015
Table 1. The summary statistics for MMTP data (mm) collected from weather stations in Mazandaran province (n= 78).
Raw data
a
BoxCox transformed data
Months Min Max Mean SD CV (%) Skewness Kurtosis K-S p Skewness Kurtosis K-S p λ
Jan 15 132.2 68.57 27.3 39.8 0.29 0.41 0.37
Feb 19.8 130.1 69.3 23.8 34.3 0.15 0.42 0.49
Mar 25.3 112.8 68.57 20.4 29.7 0.01 0.62 0.50
Apr 28 110.6 55.18 17.8 32.3 1.04 0.84 0.02 0.01 0.28 0.59 0.4
May 20.5 90.6 47.44 17.0 35.8 0.59 0.40 0.00 0.01 0.59 0.347 0.14
Jun 11.3 76.6 34.32 16.2 47.3 0.67 0.48 0.00 0.01 0.86 0.11 0.1
Jul 8.5 79.6 35.76 18.6 52.1 0.72 0.27 0.02 0.03 0.81 0.27 0.18
Aug 6.1 80 36.53 19.8 54.1 0.23 0.93 0.16
Sep 6.8 171.4 68.44 48.5 70.9 0.53 0.65 0.03 0.16 1.08 0.07 0.4
Oct 12.8 250.8 97.15 70.0 72.0 0.84 0.28 0.00 0.07 0.88 0.32 0.23
Nov 15.7 227 100.13 55.8 55.7 0.67 0.32 0.00 0.05 0.54 0.23 0.39
Dec 22.5 186 89.81 39.7 44.2 0.47 0.39 0.07
Note:
a
Min, minimum; Max, maximum; SD, standard deviation; CV, coefficient of variation; K-S p, significance level of KolmogorovSmirnov test.
8M. Emadi et al.
Downloaded by [Hacettepe University] at 22:41 14 July 2015
Table 2. The summary statistics for MMAT data (°C) collected from weather stations in Mazandaran province (n= 53).
Raw data
a
BoxCox transformed data
Months Min Max Mean SD CV (%) Skewness Kurtosis K-S p Skewness Kurtosis K-S p λ
Jan 6.7 8.8 5.07 4.41 86.8 1.464 0.89 0.00 0.48 0.35 0.06 1.94
Feb 6.1 9.6 5.09 4.12 80.9 1.426 0.83 0.00 0.16 0.24 0.08 2.35
Mar 2.7 11.9 7.16 3.43 47.9 1.289 0.85 0.00 0.48 0.67 0.07 2.97
Apr 3.5 16.2 11.67 3.03 25.9 0.975 0.16 0.00 0.24 0.71 0.21 3.56
May 7.0 20.1 16.27 3.13 19.2 1.06 0.59 0.01 0.25 0.92 0.25 4.31
Jun 6.0 24.5 20.31 3.53 17.4 1.61 3.83 0.00 0.29 0.88 0.09 4.79
Jul 6.3 28.6 23.85 3.7 15.5 2.30 8.51 0.00 0.29 0.64 0.068 5.25
Aug 6.9 27.1 23.13 3.58 15.5 2.11 6.87 0.00 0.38 0.94 0.05 5.82
Sep 4.4 26.4 21.12 4.26 20.1 2.12 6.74 0.00 0.16 1.08 0.45 4.55
Oct 2.1 21.7 17.37 3.32 19.1 1.12 0.71 0.00 0.07 0.88 0.23 4.62
Nov 0.2 16.6 12.34 3.8 30.8 1.43 1.48 0.00 0.05 0.54 0.39 4.06
Dec 3.9 11.3 7.68 4.18 54.3 1.43 0.88 0.00 0.04 0.55 0.46 3.15
Note:
a
Min, minimum; Max, maximum; SD, standard deviation; CV, coefficient of variation; K-S p, significance level of KolmogorovSmirnov test.
Archives of Agronomy and Soil Science 9
Downloaded by [Hacettepe University] at 22:41 14 July 2015
lowest value of 34.32 in June and the highest value of 100.13 mm in November (Table 1).
Higher coefficient of variance (CV) of MMTP were observed in October (72%) and lower
CV was found in winter, ranging from 29.7% (March) to 39.8% (January). The annual
mean air temperature across the Mazandaran province was 14.3°C, with the lowest
MMAT value of 5.07 in January and the highest value of 23.85 in July (Table 1).
Higher CV of MMAT were observed in winter varying from 86.8% (January) to 47.9%
(March), and lower CV was found in summer, ranging from 20.1% (September) to 15.6%
(July). By considering the variability classes proposed by Wilding (1985), it is confirmed
that the MMTP and MMAT showed moderate to high spatial variability across the
Mazandaran province.
The raw MMTP data for January, February, March, August, and December were
normally distributed (Table 1). All MMAT data sets did not pass the normality test and
need to be transformed. Transformation of these data sets showed that the normality of
distributions could be achieved. Logarithmic transformation did not pass the normality
test for skewed data. Therefore, the BoxCox transformation was used. The BoxCox
transformation can choose an appropriate power parameter (λ) to push the skewness
toward 0. The determined lambdas for non-normal distributed data sets are given in
Tables 1 and 2. In this study, application of BoxCox transformation was effective in
normalizing the data in addition to weakening the negative effect of outliers. The raw data
and the transformed data (shown in Tables 1 and 2) that followed a normal distribution
(KolmogorovSmirnov pvalue >0.05) were used for further analysis.
Spatial structure analysis
The optimal semivariogram models and best fitted model parameters for MMTP and
MMAT data sets based on the smallest RSS and highest R
2
values are given in Tables 3
and 4. No anisotropy was evident in the directional semivariograms of data sets. Thus,
isotropic semivariogram models were applied to the climatic data. All MMTP and MMAT
data sets differed in their spatial dependence (Tables 3 and 4). The semivariograms
revealed the spatial variability in the spatial structures of MMTP and MMAT data sets
across the Mazandaran Province. The spatial variability of MMTP data for January,
February, May, and August months were best described using exponential models,
Table 3. Semivariogram model parameters of the normalized MMTP data.
Months Model
a
Nugget Sill Nugget/Sill Classes Range (m) RSS R
2
Jan Exp 115 901.4 0.1276 Strong 46000 72865 0.89
Feb Exp 282 777.8 0.3625 Strong 101400 46466 0.79
Mar Gau 115.9 409.7 0.2828 Moderate 16200 32357 0.64
Apr Sph 0.00015 62900 0.2464 Strong 38400 5.811 0.68
May Exp 0.0015 0.0079 0.1891 Strong 41600 7.631 0.77
Jun Gau 0.000102 0.0046 0.2289 Strong 18400 3.31 0.75
Jul Sph .0027 0.0363 0.0743 Strong 46000 4.713 0.95
Aug Exp 35 483.2 0.0724 Strong 43900 8480 0.95
Sep Gau 0.26 3.61 0.0720 Strong 53300 0.463 0.97
Oct Gau 0.015 0.312 0.0481 Strong 58400 3.385 0.97
Nov Sph 0.001 2.28 0.0004 Strong 120300 0.456 0.94
Dec Sph 249 1885 0.1321 Strong 119400 45541 0.89
Note:
a
Sph, Gau, and Exp indicate spherical, Gaussian, and exponential models, respectively.
10 M. Emadi et al.
Downloaded by [Hacettepe University] at 22:41 14 July 2015
whereas the March, June, September, and October were best described using a Gaussian
model. A spherical model was the best fitted model for the MMTP of April, July,
November, and December. Meanwhile, the spatial variability of MMAT data for March,
August, and September months were best described using spherical model, whereas the
other months were best described using a Gaussian model.
Nugget semivariance (C
0
) is the combination of random errors and sources of varia-
tion at distances smaller than the shortest sampling interval (Goovaerts 1999). The spatial
heterogeneity of MMTP and MMAT data sets differs markedly due to the difference in
nugget C
0
. Spatial variability could be investigated using semivariograms, whereas the
ratio of nugget semivariance (representing random variation, which is undetectable at the
scale of sampling) to total semivariance (or the sill) could be used to classify spatial
dependence. To evaluate the spatial dependency of MMTP and MMAT data sets, a
criterion suggested by Cambardella et al. (1994) was used. Spatial class ratios were
categorized to define distinctive spatial dependency. If the spatial class ratio is <0.25,
the variable is considered strongly spatially dependent; if the ratio is >0.25 and <0.75, the
variable is considered moderately spatially dependent; and if the ratio is >0.75, the
variable is considered weakly spatially dependent (Cambardella et al. 1994).
The nugget/sill ratios for all MMTP and MMAT data sets expect for total precipitation
of March and mean air temperature of June were below 0.25, suggesting strongly spatial
dependence. The total precipitation of March and mean air temperature of June were
moderately spatially dependent, with the nugget/sill of 0.28 and 0.26, respectively. This
result demonstrates that the structural factor is dominant over nugget random component
for both MMTP and MMAT.
To find the distance of dependency of the spatially structured data, the range was
evaluated from the semivariogram results. The range of the semivariogram indicates the
effective distance between samples considered to be independent from each other. The
average range values for MMTP and MMAT were 59754 m and 96321 m, respectively.
When the distribution of soil properties is strongly or moderately spatially correlated, the
mean extent of these patches is given by the range of the semivariogram (Emadi et al.
2010). Range value varied from 16,200 m (March) to 120,300 m (November) for the
MMTP data sets. The total precipitation of November with a small nugget (0.001), a high
sill (2.28), and the largest range (120,300 m) indicated the greatest spatial variability
Table 4. Semivariogram model parameters of the normalized MMAT data.
Months Model
a
Nugget Sill Nugget/Sill Classes Range (m) RSS R
2
Jan Gau 9.0 × 10
1
6.72 × 10
3
0.0134 Strong 99419.7 7.14 × 10
6
0.98
Feb Gau 2.4 × 10
3
5.93 × 10
4
0.0405 Strong 97687.7 6.9 × 10
8
0.89
Mar Sph 4.37 × 10
5
2.94 × 10
6
0.1487 Strong 95436.0 2.7 × 10
12
0.81
Apr Gau 6.8 × 10
7
3.71 × 10
8
0.1835 Strong 100285.7 4.8 × 10
16
0.77
May Gau 3.22 × 10
10
1.86 × 10
11
0.1731 Strong 96475.2 7.8 × 10
21
0.85
Jun Gau 4.37 × 10
12
1.69 × 10
13
0.2603 Moderate 86110.8 5.45 × 10
25
0.84
Jul Gau 1.19 × 10
16
5.61 × 10
16
0.2127 Strong 94223.6 6.14 × 10
32
0.86
Aug Sph 2.66 × 10
14
1.20 × 10
15
0.2224 Strong 86780.8 2.9 × 10
29
0.87
Sep Sph 6.4 × 10
11
4.87 × 10
12
0.1315 Strong 104789.1 5.5 × 10
24
0.77
Oct Gau 4.11 × 10
11
2.31 × 10
12
0.1782 Strong 98726.9 8.3 × 10
23
0.88
Nov Gau 1.43 × 10
9
1.39 × 10
10
0.1028 Strong 99766.1 1.48 × 10
19
0.95
Dec Gau 9.5 × 10
5
1.31 × 10
7
0.0726 Strong 96821.6 1.83 × 10
13
0.94
Note:
a
Sph and Gau indicate spherical and Gaussian models, respectively.
Archives of Agronomy and Soil Science 11
Downloaded by [Hacettepe University] at 22:41 14 July 2015
relative to other MMTP data sets. Generally, the range values of MMTP data sets were
smaller than that of MMAT data sets. Mean air temperature of June and September had
maximum and minimum range values, 86,110 m and 104,789 m, respectively.
Interpolation of MMTP and MMAT
To determine the best accuracy of the four interpolation methods, ME, RMSE, and CCC
were used. Tables 5 and 6summarize the results of assessment of the accuracy of different
interpolation methods in estimating MMTP and MMAT. The accuracy of different inter-
polation methods varied within months for both MMTP and MMAT (Tables 5 and 6). The
results indicated that the CS method was not substantially suitable. Therefore, CS inter-
polator is worse than other interpolation methods for predicting MMTP and MMAT
values.
Table 5. The best interpolation method to predict MMTP accompanying with the quality of
assessment criterions.
a
Months Best interpolation method ME RMSE CCC
Jan OK 2.02357 26.72 0.66
Feb Co-K 4.89852 28.19 0.59
Mar IDW 1.6541 23.85 0.61
Apr OK 0.1441 1.02 0.68
May Co-K 0.0234 0.08 0.53
Jun Co-K 0.0141 0.06 0.61
Jul Co-K 0.0125 0.07 0.51
Aug Co-K 4.7954 18.89 0.55
Sep Co-K 0.1325 1.09 0.78
Oct Co-K 0.0298 0.27 0.65
Nov OK 0.1176 0.88 0.51
Dec OK 9.8401 36.88 0.49
Note:
a
ME, RMSE, and CCC were determined based on normalized or transformed data sets.
Table 6. The best interpolation method to predict MMAT accompanying with the quality of
assessment criterions.
a
Months Selected interpolation method ME RMSE CCC
Jan Co-K 0.12 5.41 0.57
Feb Co-K 1.2 7.81 0.55
Mar Co-K 81.5 321.5 0.59
Apr Co-K 5124 13258 0.44
May OK 90581 349681 0.62
Jun Co-K 216664 3063211 0.66
Jul OK 99238688 200583655 0.78
Aug Co-K 1762418 38333808 0.63
Sep Co-K 524421 2112694 0.59
Oct OK 409804 1098074 0.49
Nov Co-K 1102 38825 0.64
Dec Co-K 34 1503 0.62
Note:
a
ME, RMSE and CCC were determined based on normalized or transformed datasets.
12 M. Emadi et al.
Downloaded by [Hacettepe University] at 22:41 14 July 2015
OK performed best for interpolation of the mean total precipitation in January, April,
November, and December. However, the IDW technique interpolated the mean total
precipitation of March with the greatest accuracy. Co-K method surpassed OK, IDW,
and CS for interpolating the total precipitation of February, May, June, July, August,
September, and October. In all implementations of IDW, the power of one was the best
choice (over powers of two, three, and four), which is possibly due to the relatively low
skewness inherent in MMTP data. The greater the weighting power of IDW, the greater
RMSE of interpolation (data not shown).
OK and Co-K methods had the highest accuracy for MMAT compared with IDW and
CS interpolation methods. Ordinary Co-K also allows for local variability in the mean by
incorporating estimates of the local means of the primary and covariate variables centered
on the location being estimated. The Co-K interpolation method that used the elevation as
covariate outperformed the other interpolation methods for prediction of most MMAT, i.e.,
January, February, March, April, June, August, September, November, and December.
However, OK had greatest accuracy for interpolation of MMAT of May, July, and
October. The lowering in the accuracy of spatial prediction in kriging and Co-K compared
with IDW for some MMTP and MMAT data sets could be due to a high variation in the
air temperature within the distance of the station sites and also the distortion of linear
relationships of the air temperature with elevation. Both the mentioned variation and
distortion can be attributed to the strong sunshine in summer and strong nocturnal cooling
in winter (Ishida & Kawashima 1993).
Lloyd (2005) also concluded that OK provides the most accurate estimates of pre-
cipitation for January and February. The advantage of OK over other IDW and CS
methods is that the spatial variation structure is estimated through variogram and takes
spatial autocorrelation into consideration (Emadi et al. 2010; Aalto et al. 2013; Emadi &
Baghernejad 2014). Wu and Li (2013) showed that the large-scale surface trend describes
the relationship between the monthly temperature and the geographic (latitude and long-
itude) as well as topographic (elevation) variables in the United States.
Studies on predictive mapping of precipitation and temperature at high resolution like
this research were already performed using different interpolation techniques such as IDW
(Zhan et al. 2011; Kizza et al. 2012), OK (Lloyd 2005; Grimes & Pardo-Iguzquiza 2010;
Wagner et al. 2012; Aalto et al. 2013), and Co-K (Ishida & Kawashima 1993; Lloyd 2005;
Jantakat & Ongsomwang 2011).
Our results are not in agreement with findings of Aalto et al. (2013) who indicated that
spatial autocorrelation does not play an important role for MMAT presumably due to the
more stochastic nature of rainfall. They showed that the kriging with external drift method
was accurate and stable for interpolating monthly interpolations of climate data because of
the robustness of kriging. Sun et al. (2015) indicated that adding the kriging of regression
residuals can help improve the prediction performance. Carrera-Hernandez and Gaskin
(2007) used the elevation as the secondary variable improved the spatial variation of
temperature, even in cases of their low correlation with elevation. The advantages of using
terrain heights as auxiliary data are especially noticeable when sparse measured data are
accessible (Lloyd 2005).
The spatial distribution maps of MMTP and MMAT obtained by best interpolator
(Tables 5 and 6) across the Mazandaran province are shown in Figures 3 and 4. As shown
in Figure 2, the MMTP values are high in the upper northwest regions and in roughly
middle strip of the province on Alborz north facing slope. Overall, the predicted maps
show that MMTP decreases from the northwest to the east and northeast. The mountains
and the sea mostly influence the climatic conditions of the study area, leading to high
Archives of Agronomy and Soil Science 13
Downloaded by [Hacettepe University] at 22:41 14 July 2015
spatial variability. The wettest area is in the mountains at mead-west and in center strip
parts of the province with average precipitation exceeding 900 mm. Generally, the eastern
and southern parts of the province are drier than the western and northern north-facing
Figure 3. Spatial distribution maps of MMTP across the Mazandaran province.
14 M. Emadi et al.
Downloaded by [Hacettepe University] at 22:41 14 July 2015
Figure 4. Spatial distribution maps of MMAT across the Mazandaran province.
Archives of Agronomy and Soil Science 15
Downloaded by [Hacettepe University] at 22:41 14 July 2015
parts. The relative high CCC values of interpolators indicate that the MMTP and MMAT
mapping using selected methods (Tables 5 and 6) show a better agreement along the 45
degree line between the measured MMTP and MMAT values and the estimated values.
MMTP and MMAT values are interpolated at each node of a 1 km × 1 km grid across the
Mazandaran province using best interpolators for JNSM implementation.
Mapping SMR and STR
To determine the boundaries of the SMR and STR classes, the JNSM was applied to
the MMTP and MMAT predictive values at each node of 1 km × 1 km grids across the
Mazandaran province. For those data sets that the transformations were used for
normalization, the back transformation was performed. The maps of SMR and STR
classes are given in Figures 5 and 6. The obtained maps predict high diversity of SMR
classes including Aridic, Ustic, Udic, and Xeric moisture regimes. The predictive STR
Figure 5. Spatial distribution map of STR classes in Mazandaran province.
Figure 6. Spatial distribution map of SMR classes in Mazandaran province.
16 M. Emadi et al.
Downloaded by [Hacettepe University] at 22:41 14 July 2015
classes comprise the Mesic, Thermic, and Cryic temperature regimes. The areas with
Aridic, Ustic, Udic, and Xeric moisture regimes were spread to 2.04%, 27.37%, 7.2%,
and 63.39% of the study area, respectively (Ta ble 7). As presented in Ta ble 7,
approximately 0.91%, 66.31%, and 32.78% of the study area were classified as
Mesic, Thermic, and Cryic temperature regimes, respectively. The STR changes from
Thermic in the north to Mesic in the Alborz heights in south. There are very small
areas with Cryic STR class.
A considerable area (27.37%) classified as Ustic in the predictive SMR map is
already classified as Udic in two conventional soil surveys of the Mazandaran
province (Izadpanah et al. 1976; Rameshni & Banaei 1984)andSWRImap(Soil
and Water Research Institute 2000). These areas are found in approximately central
part of the study area and may be due to negative summer water balance associated
with higher evapotranspiration and relatively lower summer precipitation in the period
of record (Winzeler et al. 2012). Moreover, the subclass of Ustic SMR in the current
study is the Wet Tempustic that had very close condition with the Udic SMR class.
Winzeler et al. (2012) also forecasted an increase in the Ustic moisture regime and a
decrease in the Udic moisture regime in 2080 compared to the current status in US.
The Newhall model as a standard application of the SMR classification has provided a
useful, broad, quick, and cheap method to differentiate diverse large areas of land
(Bofante et al. 2011). Modern environmental challenges (e.g., the effects of climate
change, environmental pollution, etc.), however, cannot be adequately solved by such
a static approach and necessitate instead a dynamic characterization of the environ-
ment (Bofante et al. 2011). Bofante et al. (2010) stated the Newhall model over-
estimate the dry condition in the soil moisture control section during the year but
allows the separation of different pedoclimatic settings. Winzeler et al. (2012)demon-
strated that map production through direct application of a soil moisture model to
geospatial data layers can lead to more consistent model output than a historical hand-
delineated map made using expert knowledge. Bofante et al. (2011) explored several
strategies for improving soil climate estimates within soil taxonomy schemes and
recommended simulation modeling as one viable approach. Other recommendations
included a greater reliance on physical measurement and possible modifications of
taxonomic definitions based directly on matric potential measurements taken over time
rather than soil moisture control sections.
Table 7. Resulting overall areas for the classes and subclasses of soil climate based on JNSM.
Soil moisture regime classes Subclasses Area (km
2
) Relative abundance (%)
Udic Typic Udic 98.7 0.42
Dry Tempudic 1609.4 6.77
Ustic Wet Tempustic 6502.8 27.37
Xeric Typic Xeric 12551 52.83
Dry Xeric 2509.1 10.56
Aridic Weak Aridic 485.1 2.04
Soil temperature regime classes
Mesic 7788.3 32.78
Thermic 15750.6 66.31
Cryic 217.1 0.91
Archives of Agronomy and Soil Science 17
Downloaded by [Hacettepe University] at 22:41 14 July 2015
Conclusions
Soil climate characteristics have been used to differentiate taxa from the order down to the
suborder, great group, subgroup, family, and soil series levels in US Soil Taxonomy. The
main objective of this work was to generate the surface of MMTP and MMAT and to
delineate the SMR and STR classes based on predicted MMTP and MMAT values using
JNSM across the Mazandaran province, northern Iran. When necessary, BoxCox trans-
formation was applied to achieve normality followed by geostatistical analyses. The OK
and Co-K predictions were more robust than IDW and CS for interpolation of both
MMTP and MMAT with smaller ME and RMSE and higher CCC for all months except
for March that the IDW with weighing power of one had greatest accuracy. Moreover,
results indicate that, in most cases, the use of elevation data to inform estimation of
MMTP and MMAT in Mazandaran province is beneficial. These interpolation methods
allowed us to develop a gridded database for areas where no weather stations exist. The
range values of MMTP data sets were smaller than that of MMAT data sets. OK
performed best for interpolation of the MMTP of January, April, November, and
December. However, the IDW technique interpolated the mean total precipitation of
March with the greatest accuracy. Co-K method surpassed OK, IDW, and CS for inter-
polating the total precipitation of February, May, June, July, August, September, and
October. OK and Co-K methods had the highest accuracy for MMAT compared with IDW
and CS interpolation methods. The predictive maps show high diversity of SMR classes
including Aridic, Ustic, Udic, and Xeric moisture regimes. The predictive STR classes
comprise the Mesic, Thermic, and Cryic regimes. Because JNSM assumes precipitation
excess exits the soil as runoff or as deep percolation, resulting predictive maps of SMR
and STR classes are only valid for well-drained soils without perched or permanent water
tables. Furthermore, soil surveys need to be accomplished across the entire Mazandaran
province to delineate the soils with Aquic SMRs. The Aquic SMR regime is commonly
reported in low-lying plains and coastal deltas of Mazandaran province. Therefore, the
application of current SMR predictive map in these areas needs more detailed considera-
tion of soils. Overall, results indicated that geostatistical approaches can potentially
provide and accurate the mapping of SMR and STR classes in areas with sparse or
nonexistent weather station data. These maps also could be considered in the future to
better analyze the climatic inference on soil processes and behavior.
Acknowledgement
The authors are grateful to two anonymous reviewers and editor-in-chief, Dr. Galina Machulla, for
their critical comments and suggestions that improved the quality of the manuscript.
Disclosure statement
No potential conflict of interest was reported by the authors.
References
Aalto J, Pirinen P, Heikkinen J, Venäläinen A. 2013. Spatial interpolation of monthly climate data
for Finland: comparing the performance of kriging and generalized additive models. Theor Appl
Climatol. 112:99111.
Bahmaniar MA, Abtahi A, Banai MH. 1999. Soil landscape relationships between different physio-
graphic units in Tejan catchment (northern Iran). Iran Agric Res. 18:107124.
Benavides R, Montes F, Rubio A, Osoro K. 2007. Geostatistical modelling of air temperature in a
mountainous region of Northern Spain. Agric Forest Meteo. 146:173188.
18 M. Emadi et al.
Downloaded by [Hacettepe University] at 22:41 14 July 2015
Bilgili AV. 2013. Spatial assessment of soil salinity in the Harran Plain using multiple kriging
techniques. Environ Monit Assess. 185:777795.
Billaux P. 1978. Estimation du regime hydriquedes sols au moyen des données climatiques. Le
method graphique: Son utilization dansle cadre de la Taxonomie Americaine des sols. (In
French, with English abstract.) Cah ORSTOM Ser Pedol. 16:317318.
Bofante A, Basile A, Manna P, Terribile F. 2010. Soil moisture regime classification of Soil
Taxonomy through physically based soil water balance modeling. Paper presented at: Soil
Solutions for a Changing World. 19th World Congress of Soil Science; Brisbane, Australia.
Bofante A, Basile A, Manna P, Terribile F. 2011. Use of physically based models to evaluate USDA
soil moisture classes. Soil Sci Soc Am J. 75:181191.
Bostan PA, Heuvelink GBM, Akyurek SZ. 2012. Comparison of regression and kriging techniques
for mapping the average annual precipitation of Turkey. Int J Appl Earth Obs Geoinf.
19:115126.
Box GEP, Cox DR. 1964. An analysis of transformations. Stat Soc Ser. 26:211252.
Cambardella CA, Moorman TB, Novak JM, Parkin TB, Karalan DL, Turco RF, Konopka AE.
1994. Field scale variability of soil properties in central Iowa soils. Soil Sci Soc Am J.
58:15011511.
Cambule AH, Rossiter DG, Stoorvogel JJ, Smaling EMA. 2014. Soil organic carbon stocks in the
Limpopo National Park, Mozambique: amount, spatial distribution and uncertainty. Geoderma.
213:4656.
Carrera-Hernandez JJ, Gaskin SJ. 2007. Spatio temporal analysis of daily precipitation and tem-
perature in the Basin of Mexico. J Hydrol. 336:231249.
Costantini EAC, Castelli F, Raimondi S, Lorenzoni P. 2002. Assessing soil moisture regimes with
traditional and new methods. Soil Sci Soc Am J. 66:18891896.
Emadi M, Baghernejad M. 2014. Comparison of spatial interpolation techniques for mapping soil
pH and salinity in agricultural coastal areas, northern Iran. Arch Agron Soil Sci. 60:13151327.
Emadi M, Baghernejad M, Bahmanyar MA, Morovvat A. 2012. Changes in soil inorganic phos-
phorous pools along a precipitation gradient in northern Iran. Int J Forest Soil Erosion.
2:143147.
Emadi M, Baghernejad M, Pakparvar M, Kowsar SA. 2010. An approach for land suitability
evaluation using geostatistics, remote sensing, and geographic information system in arid and
semiarid ecosystems. Environ Monit Assess. 164:501511.
Emery X, Ortiz JM. 2007. Weighted sample variograms as a tool to better assess the spatial
variability of soil properties. Geoderma. 140:8189.
Esfandiarpour I, Salehi MH, Karimi AR, Kamali A. 2013. Correlation between soil taxonomy and
world reference base for soil resources in classifying calcareous soils: a case study of arid and
semi-arid regions of Iran. Geoderma. 197198:126136.
Gamma Design Software. 2004. GS +: geostatistics for the environmental sciences. Plainwell (MI):
Gamma Design Software, LLC; p. 160.
Goovaerts P. 1999. Geostatistics in soil science: state-of-the-art and perspectives. Geoderma.
89:145.
Grimes DIF, Pardo-Iguzquiza E. 2010. Geostatistical analysis of rainfall. Geogr Anal. 42:136160.
Harris P, Fotheringham AS, Crespo R, Charlton M. 2010. The use of geographically weighted
regression for spatial prediction: an evaluation of models using simulated data sets. Math
Geosci. 42:657680.
Huisman JA, Snepvangers JJJC, Bouten W, Heuvelink GBM. 2002. Mapping spatial variation in
surface soil water content: comparison of ground penetrating radar and time domain reflecto-
metry. J Hydrol. 269:194207.
ICOMMOTR. 1991. Circular letter no. 2. Lincoln (NE): National Soil Survey Center.
Isaaks EH, Srivastava RM. 1989. An introduction to applied geostatisics. New York (NY): Oxford
University Press.
Ishida T, Kawashima S. 1993. Use of cokriging to estimate surface air temperature from elevation.
Theor Appl Climat. 47:147157.
IUSS Working Group WRB. 2014. World reference base for soil resources. 3rd ed. Rome: Food and
Agriculture Organization of the United Nations.
Izadpanah B, Riyahi A, Irvani-Mohajeri SA. 1976. Semi-detailed soil survey and land classification
of Mazandaran province (Neka-Siyahrud area). Tehran: Soil and Water Institute, Ministry of
Agricultures and Natural Resources.
Archives of Agronomy and Soil Science 19
Downloaded by [Hacettepe University] at 22:41 14 July 2015
Jafari A, Khademi H, Finke P, Wauw J, Ayoubi S. 2014. Spatial prediction of soil great groups by
boosted regression trees using a limited point dataset in an arid region, southeastern Iran.
Geoderma. 232234:148163.
Jantakat Y, Ongsomwang S. 2011. Assessing the effect of incorporating topographical data with
geostatistical interpolation for monthly rainfall and temperature in Ping basin, Thailand.
Suranaree J Sci Technol. 18:123139.
[jNSM] Java Newhall Simulation Model [Internet]. 2012. Newhall Simulation Model v 1.5.1; [cited
2012 Feb 22]. Available from: http://www.nrcs.usda.gov
Khormali F, Ghergherechi S, Kehl M, Ayoubi S. 2012. Soil formation in loess-derived soils along a
subhumid to humid climate gradient, northeastern Iran. Geoderma. 179180:113122.
Khormali F, Kehl M. 2011. Micromorphology and development of loess-derived surface and buried
soils along a precipitation gradient in northern Iran. Quat Int. 234:109123.
Kizza M, Westerberg I, Rodhe A, Ntale HK. 2012. Estimating areal rainfall over Lake Victoria and
its basin using ground-based and satellite data. J Hydrol. 464465:401411.
Lapen DR, Hayhoe HN. 2003. Spatial analysis of seasonal and annual temperature and precipitation
normals in Southern Ontario, Canada. J Great Lakes Res. 29:529544.
Li H, Shen W, Zou C, Jiang J, Fu L, She G. 2013. Spatio-temporal variability of soil moisture and
its effect on vegetation in a desertified aeolian riparian ecotone on the Tibetan Plateau, China. J
Hydrol. 479:215225.
Lin LI. 1989. A concordance correlation coefficient to evaluate reproducibility. Int Biometric Soc.
45:255268.
Lloyd CD. 2005. Assessing the effect of integrating elevation data into the estimation of monthly
precipitation in Great Britain. J Hydrol. 308:128150.
McGrath D, Zhang C, Carton OT. 2004. Geostatistical analyses and hazard assessment on soil lead
in silvermines area, Ireland. Environ Pollut. 127:239248.
Newhall F, Berdanier CR. 1996. Calculation of soil moisture regimes from the climatic record.
Lincoln (NE): Natural Resources Conservation Service.
Pahlavan-Rad MR, Toomanian N, Khormali F, Brungard CW, Komaki CB, Bogaert P. 2014.
Updating soil survey maps using random forest and conditioned Latin hypercube sampling in
the loess derived soils of northern Iran. Geoderma. 232234:97106.
Pannatier Y. 1993. Statistics of spatial processes: theory and applications. New York (NY): Springer-
Verlag. Chapter 2, MS-Windows programs for exploratory variography and variogram modelling
in 2D; p. 165170.
Rameshni K, Banaei MS. 1984. Reconnaissance and semi-detailed soil map of west Mazandaran
province (Siyahrud-Tukehrud area), northern Iran. 12 sheets: 1:50,000; 70 × 100 cm; black and
white color. Tehran: Soil and Water Institute.
[SSS] Soil Survey Staff. 2014. Keys to soil taxonomy. 12th ed. Washington (DC): USDA-Natural
Resources Conservation Service.
Sun W, Zhu Y, Huang S, Guo C. 2015. Mapping the mean annual precipitation of China using local
interpolation techniques. Theor Appl Climatol. 119:171180.
SWRI-Soil and Water Research Institute Iran. 2000. Soil resources and use potentiality map of Iran
(1:1000000), Tehran.
Symeonakis E, Bonifacio R, Drake N. 2009. A comparison of rainfall estimation techniques for sub-
Saharan Africa. Int J Appl Earth Obs Geoinf. 11:1526.
Taghizadeh-Mehrjardi R, Sarmadian F, Minasny B, Triantafilis J, Omid M. 2014. Digital mapping of
soil classes using decision tree and auxiliary data in the Ardakan region, Iran. Arid Land Res
Mang. 28:147168.
[USDA] United States Department of Agriculture. 1999. Soil taxonomy: a basic system of soil
classification for making and interpreting soil surveys. Washington (DC): USDA, U.S. Gov.
Print. Office.
Van Wambeke A. 1986. Newhall Simulation Model: a BASIC program for the IBM PC (DOS 2.0 or
later). New York (NY): Cornell University.
Wagner PD, Fiener P, Wilken F, Kumar S, Schneider K. 2012. Comparison and evaluation of spatial
interpolation schemes for daily rainfall in data scarce regions. J Hydrol. 464465:388400.
Waltman SW, Miller DA, Bills B, Waltman WJ. 2011. JAVA Newhall simulation model Update to
a traditional soil climate simulation model. Presented at the Fundamental for Life: Soil, Crop
and Environmental Sciences. Soil Science Society of America annual meetings; San Antonio,
TX, USA.
20 M. Emadi et al.
Downloaded by [Hacettepe University] at 22:41 14 July 2015
Webster R, Oliver MA. 2001. Geostatistics for environmental scientists. Brisbane: Wiley.
Wilding LP. 1985. Spatial variability: its documentation, accommodation and implication to soil
surveys. In: Nielsen DR, Bouma J, editors. Soil spatial variability. Wageningen: Pudoc; p. 166
194.
Winzeler HE, Waltman SW, Waltman WJ, Libohova Z, Beaudette D. 2012. A methodology for
examining changes in soil climate geography through time: U.S. soil moisture regimes for the
period 19712000. Soil Sci Soc Am J. 77:213225.
Wu T, Li Y. 2013. Spatial interpolation of temperature in the United States using residual kriging.
Appl Geog. 44:112120.
Zhan MJ, Yin JM, Zhang YZ. 2011. Analysis on characteristic of precipitation in Poyang Lake basin
from 1959 to 2008. Procedia Environ Sci. 10:15261533.
Zhang CS, Selinus O. 1998. Statistics and GIS in environmental geochemistry some problems and
solutions. J Geochem Explor. 64:339354.
Archives of Agronomy and Soil Science 21
Downloaded by [Hacettepe University] at 22:41 14 July 2015
... Mazandaran province, northern Iran, is located on the southern coast of the Caspian Sea. There is a descending precipitation gradient from the west to east across the region, leading to a diversity of soil moisture regime (SMR) and soil temperature regime (STR) classes [57]. Due to the changes in SOC contents in northern Iran caused by the human activities and natural attributes (landslide, flooding, depression) [58,59], the existence of a high-quality SOC prediction map with known uncertainty in the Mazandaran province is crucial. ...
... There is a gradient of the decreasing precipitation from the west (around 1400 mm) to the east direction (around 450 mm) leading to the diversity of soil moisture regime (SMR) and soil temperature regime (STR) classes across the province. The xeric SMR class covers the largest area in the province, followed by the udic and aquic classes, while thermic (66%) is the most abundant STR followed by mesic (33%) and cryic (1%) [57]. The variation of elevation ranges from the Caspian Sea coastal areas with elevations <−5 m to more than 3000 m above sea level in the highlands of the Alborz Mountain range. ...
... The SOC mean value was the highest in the udic SMR class with mean values of 3.85% followed by the aquic (2.45%) and xeric (2.10%), respectively. The high precipitation for soils with the udic SMR class [57] led to the high aboveground biomass production inputs. The greater SOC contents in soils having the aquic SMR class compared to the xeric SMR could be related to anaerobic (reducing) conditions decreasing the rates of organic matter decomposition [115]. ...
Preprint
Full-text available
Estimation of the soil organic carbon content is of utmost importance in understanding the chemical, physical, and biological functions of the soil. This study proposes machine learning algorithms of support vector machines, artificial neural networks, regression tree, random forest, extreme gradient boosting, and conventional deep neural network for advancing prediction models of SOC. Models are trained with 1879 composite surface soil samples, and 105 auxiliary data as predictors. The genetic algorithm is used as a feature selection approach to identify effective variables. The results indicate that precipitation is the most important predictor driving 15 percent of SOC spatial variability followed by the normalized difference vegetation index, day temperature index of moderate resolution imaging spectroradiometer, multiresolution valley bottom flatness and land use, respectively. Based on 10 fold cross validation, the DNN model reported as a superior algorithm with the lowest prediction error and uncertainty. In terms of accuracy, DNN yielded a mean absolute error of 59 percent, a root mean squared error of 75 percent, a coefficient of determination of 0.65, and Lins concordance correlation coefficient of 0.83. The SOC content was the highest in udic soil moisture regime class with mean values of 4 percent, followed by the aquic and xeric classes, respectively. Soils in dense forestlands had the highest SOC contents, whereas soils of younger geological age and alluvial fans had lower SOC. The proposed DNN is a promising algorithm for handling large numbers of auxiliary data at a province scale, and due to its flexible structure and the ability to extract more information from the auxiliary data surrounding the sampled observations, it had high accuracy for the prediction of the SOC baseline map and minimal uncertainty.
... Collection and processing of sample data Based on the grid distribution point method, more than 1 100 sampling points were distributed in this study. Spatial prediction and veri cation method Geostatistical methods are widely used to predict the spatial distribution of soil properties [30][31][32] . In this paper, we chose the Ordinary Kriging (OK) method to predict the spatial distribution of soil pH 33 . ...
Preprint
Full-text available
Spatial variation of soil pH is important for the evaluation of environmental quality. A reasonable number of sampling points has an important meaning for accurate quantitative expression on spatial distribution of soil pH and resource savings. Based on the grid distribution point method, 797, 700, 594, 499, 398, 299, 200, 149, 100, 75 and 50 sampling points, which were randomly se-lected from 908 sampling points, constituted 12 sample sets. Semi-variance structure analysis was carried out for different point sets, and ordinary Kriging was used for spatial prediction and ac-curacy verification, and the influence of different sampling points on spatial variation of soil pH was discussed. The results showed that the pH value in Kenli County was generally between 7.8 and 8.1, and the soil was alkaline. Semi-variance models fitted by different point sets could well reflect the spatial structure characteristics of soil pH. With the decrease of the number of sampling points, the Sill value of sample set increased, and the spatial autocorrelation gradually weakened. Considering the prediction accuracy, spatial distribution and investigation cost, the number of sampling points greater than or equal to 150 could satisfy the spatial variation expression of soil pH at the county level in the Yellow River Delta. This was equivalent to taking at least 107 sam-pling points per 1 000 km ² . The results in this study are applicable to areas with similar environ-mental and soil conditions as the Yellow River Delta, and have certain reference significance for them.
... Scholars have been using geostatistics to predict the values of soil properties due to their consistency (Liu et al. 2014;Behera and Shukla 2015) in minimizing the blunders and costs (Behera and Shukla 2015). Geostatistical technology has become the most efficient tool to precisely predict the spatial variability of soil fertility (Silva et al. 2003;Emadi et al. 2016;Mohamed et al. 2018). Interpolation methods accurately predict soil attributes (Pandey and Pandey 2010;Yao et al. 2013). ...
Article
In the study, two geostatistical methods—Ordinary Kriging (OK) and Least Distance Weighting (IDW)—were utilized to predict the spatial variability and distribution of soil properties for improved nutrient management in the Fogera plain, Northwest Ethiopia. In an area of 5646 ha, 60 composite soil samples were collected at a 0–20-cm soil depth and analyzed for soil pH, organic carbon (OC), total nitrogen (TN), available phosphorus (av.P), exchangeable calcium (Ca2+), potassium (K+), sodium (Na+), and magnesium (Mg2+), electrical conductivity (EC), and cation exchange capacity (CEC) using standard analytical procedures. The data were then incorporated into a GIS database and semi-variogram, and geostatistical analyses were performed with ArcGIS software version 10.5. Descriptive statistical treatments were applied using IBM Statistical Packages for Social Sciences (SPSS) software version 24. The performance of interpolation methods was assessed using the mean error (ME), root mean square error (RMSE), Nash–Sutcliffe efficiency (NSE), and coefficient of determination (R2) extracted from cross-validation of predicted maps. Analysis of different semi-variogram models depicts different degrees of spatial dependence. An exponential model is detected for the soil pH, CEC, OC, and TN; a spherical one for EC and Ca2+; a moderate one for soil K+; and a weak spatial dependence for soil av. P. This result demonstrates a high spatial continuity and dependence between adjacent soil samples. The inverse distance weighting (IDW) and ordinary kriging (OK) models well described the variation of all soil fertility parameters except organic carbon (OC) and available phosphorus (av. P), which had low NSE (≤ 50%) for both IDW and OK methods. Consequently, the generated maps revealed that the spatial variability of soil properties was adequate to predict the values of soil fertility indicators in the non-sampled locations within the study area and similar regions. The geostatistical-based soil fertility maps will be helpful for farmers, researchers, and policymakers to improve soil management methods, optimize fertilization strategies, and enhance crop productivity. By means of the methodological approach applied, we have succeeded in demonstrating the strong spatial dependence of the studied soils; however, a particular attention to the implementation of site-specific soil management practices in the Fogera plain is essential.
... In addition to field assessments using easily observable site features, indicator vegetation, and easily identified soil properties [7], many studies focused on the model predictions of SNRs and SMRs. The models are often based on interpolation schemes [8] and statistics-based schemes [9,10] using varying model predictors from field-based plant indicators [7], model-based clay content [9], model-based soil drainage [11], remote sensing data [12], and map-based soil texture [10], with varying class number, map resolution, and model accuracies. However, there is a lack of model studies that estimated SNRs and SMRs with high resolution (i.e., ≤10 m) and high accuracy using easily accessible model predictors. ...
Article
Full-text available
Citation: Zhao, Z.; Yang, Q.; Ding, X.; Xing, Z. Model Prediction of the Soil Moisture Regime and Soil Nutrient Regime Based on DEM-Derived
... Compared with studies on the local scale, it has more research value which paves the way for future works on the red beds on a regional scale, such as regional evaluations on soil resource potentials, environmental monitoring, optimization of land use policy, health evaluation of soil and ecosystem, locating the remediation areas of soil. Similar literature has been conducted in other countries, such as a study focusing on mapping the regional spatial distribution of soil moisture content in Italy (Emadi et al., 2015). Therefore, the selection of interpolation parameters and the comparison of interpolation accuracy in this study could also pave the path for future research on the spatial distribution of regional soil moisture. ...
Article
Soil moisture is an important indicator for monitoring land degradation and plays an important role in soil biogeochemistry. Mapping the spatial distribution of soil moisture provides fundamental information for soil management and agricultural production. In this study, 225 sampling points in the study area, Nanxiong basin, were investigated to map the spatial distribution of soil moisture in the typical ecological degradation red beds area. Four interpolation methods including inverse distance weighting (IDW), ordinary kriging (OK), radial basis function (RBF) and empirical bayesian kriging (EBF) were used to estimate the continuous soil moisture distribution. The results showed that the soil moisture of the study area ranged from 8.12% to 32.82% with an average of 18.42% and a median of 18.20%. The annual average temperature and soil bulk density had significantly negative correlations with soil moisture, of-0.53 and − 0.31, respectively. The regional soil moisture had a moderate variation with the variation coefficient of 21.66% and strong spatial dependence with the nugget-to-sill ratios of 32.23%. The IDW method obtained a more accurate estimation on the spatial variability of soil moisture in the Nanxiong basin. The spatial distribution of soil moisture in the study area was drawn by different interpolation methods. The results show that the red bed degradation in the eastern part of the Nanxiong basin is more severe than in other areas, in which the geomorphic characteristics of red bed desert have appeared with the lowest soil moisture in this study. In addition, the spatial heterogeneity of soil moisture is also pertinent to human disturbance and land use. The results of spatial soil moisture distribution are of great significance for monitoring land degradation and agricultural drought in the red bed area.
... Within the province, the rainfall in the eastern part is about 400 mm and increases along a westward gradient to 1400 mm; hence, the soil moisture regime is classified as xeric, ustic, and udic along that same gradient. The predominant soil temperature regime class is classified as being thermic and followed by mesic and cryic [37]. The most important soil environmental factor, which influences the soil development and variability at the provincial scale, is climate, followed by the relief, parent material, and vegetation, which control local-scale soil variation [38,39]. ...
Article
Full-text available
Soil texture and particle size fractions (PSFs) are a critical characteristic of soil that influences most physical, chemical, and biological properties of soil; furthermore, reliable spatial predictions of PSFs are crucial for agro-ecological modeling. Here, series of hybridized artificial neural network (ANN) models with bio-inspired metaheuristic optimization algorithms such as a genetic algorithm (GA-ANN), particle swarm optimization (PSO-ANN), bat (BAT-ANN), and monarch butterfly optimization (MBO-ANN) algorithms, were built for predicting PSFs for the Mazandaran Province of northern Iran. In total, 1595 composite surficial soil samples were collected, and 64 environmental covariates derived from terrain, climatic, remotely sensed, and categorical datasets were used as predictors. Models were tested using a repeated 10-fold nested cross-validation approach. The results indicate that the hybridized ANN methods were far superior to the reference approach using ANN with a backpropagation training algorithm (BP-ANN). Furthermore, the MBO-ANN approach was consistently determined to be the best approach and yielded the lowest error and uncertainty. The MBO-ANN model improved the predictions in terms of RMSE by 20% for clay, 10% for silt, and 24% for sand when compared to BP-ANN. The physiographical units, soil types, geology maps, rainfall, and temperature were the most important predictors of PSFs, followed by the terrain and remotely sensed data. This study demonstrates the effectiveness of bio-inspired algorithms for improving ANN models. The outputs of this study will support and inform sustainable soil management practices, agro-ecological modeling, and hydrological modeling for the Mazandaran Province of Iran. Citation: Taghizadeh-Mehrjardi, R.; Emadi, M.; Cherati, A.; Heung, B.; Mosavi, A.; Scholten, T. Bio-Inspired Hybridization of Artificial Neural Networks: An Application for Mapping the Spatial Distribution of Soil Texture Fractions. Remote Sens.
... Within the province, the rainfall in the eastern part is about 400 mm and increases along a westward gradient to 1400 mm; hence, the soil moisture regime is classified as xeric, ustic, and udic along that same gradient. The predominant soil temperature regime class is classified as being thermic and followed by mesic and cryic [37]. The most important soil environmental factor, which influences the soil development and variability at the provincial scale, is climate, followed by the relief, parent material, and vegetation, which control local-scale soil variation [38,39]. ...
Article
Full-text available
Spatial variation of soil pH is important for the evaluation of environmental quality. A reasonable number of sampling points has an important meaning for accurate quantitative expression on spatial distribution of soil pH and resource savings. Based on the grid distribution point method, 908, 797, 700, 594, 499, 398, 299, 200, 149, 100, 75 and 50 sampling points, which were randomly selected from 908 sampling points, constituted 12 sample sets. Semi-variance structure analysis was carried out for different point sets, and ordinary Kriging was used for spatial prediction and accuracy verification, and the influence of different sampling points on spatial variation of soil pH was discussed. The results show that the pH value in Kenli County (China) was generally between 7.8 and 8.1, and the soil was alkaline. Semi-variance models fitted by different point sets could reflect the spatial structure characteristics of soil pH with accuracy. With a decrease in the number of sampling points, the Sill value of sample set increased, and the spatial autocorrelation gradually weakened. Considering the prediction accuracy, spatial distribution and investigation cost, a number of sampling points greater than or equal to 150 could satisfy the spatial variation expression of soil pH at the county level in the Yellow River Delta. This is equivalent to taking at least 107 sampling points per 1000 km2. The results in this study are applicable to areas with similar environmental and soil conditions as the Yellow River Delta, and have reference significance for these areas.
Article
Full-text available
This study aimed to evaluate spatial variability of selected soil parameters as a smart agricultural technology guide to precise fertilizer application. A farm designated as Field 3 which is under Arabica coffee within a bigger Soil Mapping Unit (SMU) was selected for a more detailed soil observation at a scale of 1:5000. Soil samples were taken at depths of 0 to 15 and 15 to 30 cm across 20 sample locations in grids and selected properties analysed in the laboratory. Kriging interpolation method was used to estimate the accuracy of interpolation through cross-validation of the top soil parameters. In 0 to 15 and 15 to 30 cm depth, soil reaction, percentage organic carbon and percent nitrogen showed low variability of 5.1% and 5.8%, 10.4% and 12.7%, 14.5% and 17.6% respectively. Phosphorus was deficient in both depths and showed moderate variability of 36.2% and 42.3% in 0 to 15 and 15 to 30 cm respectively. Calcium and Magnesium ranged from sufficient to rich and showed moderate and low variability in top and bottom depths, respectively. All micronutrients were sufficient in the soil. The soils were classified as Mollic Nitisols. Results showed that soil parameters varied spatially within the field therefore, there is need for variable input application depending on the levels of these elements and purchasing of fertilizer blends that are suitable for nutrient deficiencies. Precision agriculture is highly recommended in the field to capitalize on soil heterogeneity.
Article
Soil salt accumulation and salt leaching are very important issues for a long-term irrigation-dominated oasis with shallow groundwater in arid land. The critical ratio of drainage to irrigation (CRDI) was defined and determined to help the guide drainage and irrigation practices and maintain sustainable oasis agricultural development. This study aimed to identify the soil salinity evolution under long-term irrigation and to determine the threshold of irrigation to drainage with land-use change and groundwater depth by regression, water and salt balance, and geostatistical analysis. The results showed that the soil salinity in 0–20 cm soil depth has decreased greatly in the Weigan Oasis over the past 60 years, with an annual average decrease of 0.68 g·kg⁻¹ in the Weigan Oasis. The soil salt content generally increases from inside to outside of the region and varies from the upperstream to downstream. In the past 30 years, land-use change has caused a sharp reduction in the salinity in the region, and the process of irrigation and drainage has made a decrease in the total salt of the irrigated district of 207.38 × 10⁵ t. Meanwhile, shallow groundwater depth and salinity are the main factors influencing soil salinization in this fluvial oasis. A CRDI is determined to be 8.20% in the Weigan Oasis. In conclusion, our study provides scientific support for the prevention and control of regional soil salinization, as well as a managerial basis for the rational allocation of water resources in long-term irrigation-dominated oasis.
Article
Full-text available
Increasingly, the geographically weighted regression (GWR) model is being used for spatial prediction rather than for inference. Our study compares GWR as a predictor to (a)its global counterpart of multiple linear regression (MLR); (b)traditional geostatistical models such as ordinary kriging (OK) and universal kriging (UK), with MLR as a mean component; and (c)hybrids, where kriging models are specified with GWR as a mean component. For this purpose, we test the performance of each model on data simulated with differing levels of spatial heterogeneity (with respect to data relationships in the mean process) and spatial autocorrelation (in the residual process). Our results demonstrate that kriging (in a UK form) should be the preferred predictor, reflecting its optimal statistical properties. However the GWR-kriging hybrids perform with merit and, as such, a predictor of this form may provide a worthy alternative to UK for particular (non-stationary relationship) situations when UK models cannot be reliably calibrated. GWR predictors tend to perform more poorly than their more complex GWR-kriging counterparts, but both GWR-based models are useful in that they provide extra information on the spatial processes generating the data that are being predicted. KeywordsRelationship nonstationarity–Relationship heterogeneity–GWR–Kriging–Spatial interpolation
Article
Full-text available
Soil climate is the record of temporal patterns of soil moisture and temperature and is an important component of the structure of U. S. Soil Taxonomy. The U. S. Soil Survey has used the Newhall simulation model (NSM) for estimating soil climate from atmospheric climate records at weather stations since the 1970s. The current soil climate map of the United States was published in 1994 by using NSM runs from selected weather stations along with knowledge-based hand-drawn mapping procedures. We developed a revised soil climate mapping methodology using the NSM and digital soil mapping techniques. The new methodology is called grid element Newhall simulation model (GEN), where a coordinate system is used to divide geographic space into a grid, and each element or grid-cell serves as a reference area for querying and organizing model input and for organizing and displaying model output. The GEN was used to make a soil moisture map of the conterminous United States (GEN-CONUS). The GEN-CONUS and the 1994 map were compared to each other and to two sets of weather station data from years 1961 to 1990 and years 1971 to 2000 (NCDC). Agreement between GEN-CONUS and the 1994 map was 75.6%. GEN-CONUS had higher agreement than the 1994 map with NSM output from NCDC data for 1961-1990 and 1971-2000 (k = 0.845 and 0.777). The GEN methodology was also used to generate a map of projected soil climate in the year 2080 for part of the Southern Rocky Mountains, predicting expansion of the Ustic and contraction of the Udic moisture regimes.
Article
Full-text available
Temperature is one of the most important factors influencing every aspect of life. In response to the increasing greenhouse effect in recent years, the demand for understanding the spatial variability of temperature in the U.S. has risen dramatically. To meet this need, we developed a statistical model for constructing a gridded temperature dataset over the mainland United States. Based on the data collected from 922 meteorological stations in the U.S., temperatures at over 5000 unknown locations were predicted in January and July, 2010. This study utilized variables of latitude and longitude (model 1), and latitude, longitude and elevation (model 2) as inputs in a residual kriging method to interpolate the average monthly temperature. We also estimated temperatures at the same locations with the kriging function of ArcGIS and compared the performances of our models with that of ArcGIS. We found that, by adding an elevation factor, our model (model 2) had a better predicting performance than that of ArcGIS kriging function in both January and July. However, only estimation in July was not different from the observation. This suggests that our kriging model is capable of capturing the spatial variability of temperature, but it is sensitive to season. The successful interpolation of July temperature indicates that the accuracy of interpolation can be improved by adding appropriate variables. Seasonal models developed in future research can be valuable tools for meteorological and climatological research.
Article
Full-text available
This paper aims to assess the effect of incorporating topographical data with geostatistical interpolation for monthly rainfall and temperature in Ping Basin, Thailand. The spatial interpolation techniques based on 11 semivariogram models of 4 main sub-types of cokriging with 3 topographical variables: elevation, longitude, and latitude have been applied in this study. The best interpolation models from cokriging technique on mean monthly rainfall and mean monthly temperature are selected by Akaike Information Criterion (AIC) based on partial sill, range and nugget that the best monthly models of kriging technique is operated in same mentioned selection. In addition, an assessment of the effective results of the cokriging interpolation models is performed by 2 approaches: i) comparing the errors of the best results from other interpolations excluding topographic data with the least MAE, MRE and RMSE value and ii) comparing the accuracy of results from Multiple Linear Regression (MLR) with the coefficient of determination (r 2). It was found that cokriging models of mean monthly rainfall and mean monthly temperature have more effectiveness than other interpolations excluding topographic data and MLR including topographic data. Therefore, this study can use the best results of sub-type and semivariogram model from cokriging including topographic variables for mean monthly rainfall and mean monthly temperature surface interpolation.
Article
This study was undertaken to develop our knowledge relating to the soil inorganic phosphorous (P) pools along a precipitation gradient in calcareous forest soils. The precipitation gradient at Mazandaran province, northern Iran, reflected in the three soil moisture regimes (SMRs) e.g. xeric, ustic and udic SMRs that differed in their annual precipitation (550, 780, and 1250 mm y-1 , respectively). The representative soil profiles on the same slope position were selected for the three studied SMRs and the soil samples were collected from the genetic horizons to determine some soil characteristics and soil inorganic P (Pi) pools namely, Ca 2-P, Ca 8-P, Ca 10-P, Al-P, Fe-P, and Occluded-P. Increasing available water balance along the precipitation gradient induced greater weathering intensity across this climosequence leading to the pH and EC decline, organic matter accumulation, carbonate loss, and clay formation. Olsen-P and Ca 2-P which are considered as readily available for plants, constitutes a small fraction of total soil Pi for all soils studied and its amounts decreased with depth within all sites. The distribution of soil P fractions across the gradient suggested decreasing Ca-bound P and relatively increasing amounts of occluded P with increasing rainfall. The total Pi was decreased appreciably at udic SMR compared with xeric SMR. The maximum total Pi and Ca-bound P occurred in the upper part of the secondary calcium carbonate horizons when such a zone was present in soils. With increasing precipitation the measured amounts of the Ca 8-P, as pedogenic Ca, and Ca 10-P, as primary mineral P (mainly apatite P), decreased about 20 and 2.5 times from the xeric to udic SMR, respectively. Overall, the results herein strengthen our understanding of Pi transformations during pedogenesis along precipitation gradient in northern Iran and provide important insight into P pools distributing within soil solum at different observable SMRs in these regions.
Article
Digital soil mapping (DSM) involves acquisition of field soil observations and matching them with environmental variables that can explain the distribution of soils. The harmonization of these data sets, through computer-based methods, are increasingly being found to be as reliable as traditional soil mapping practices, but without the prohibitive costs. Therefore, the present research developed decision tree models for spatial prediction of soil classes in a 720 km 2 area located in an arid region of central Iran, where traditional soil survey methods are difficult to undertake. Using the conditioned Latin hypercube sampling method, the locations of 187 soil profiles were selected, which were then described, sampled, analyzed, and allocated to six Great Groups according to the USDA Soil Taxonomy system. Auxiliary data representing the soil forming factors were derived from a digital elevation model (DEM), Landsat 7 ETM þ images, and a map of geomorphology. The accuracy of the decision tree models was evaluated using overall, user, and producer accuracy based on an independent validation data set. Our results showed some auxiliary variables had more influence on the prediction of soil classes which included: topographic wetness index, geomorphological map, multiresolution index of valley bottom flatness, elevation, and principal components of Landsat 7 ETM þ images. Furthermore, the results have confirmed the DSM model successfully predicted Great Groups with overall accuracy up to 67.5%. Our results suggest that the developed methodology could be used to predict soil classes in the arid region of Iran.
Article
Soil Taxonomy differs from other international soil classification systems because it includes the estimation of the soil moisture regime (SMR) and is primarily based on the estimation of dry days in a soil moisture control section (SMCS). Alternatively, the International Committee on Soil Moisture and Temperature Regimes (ICOMMOTR) has proposed a new approach based on biweekly soil water potential values determined at a fixed depth. Currently, SMR estimation is performed through simple models (standard approaches), such as the Newhall and Billaux models, that are not physically based. The aims of this work were to test the appropriateness of both standard Soil Taxonomy and the ICOMMOTR approaches through the use of a hydrologic model based on the Richards equation (SWAP), and to evaluate possible SMR classification alternatives. The SWAP-derived SMR classifications were compared with those derived by standard approaches, as well as the ICOMMOTR proposals, under eight pedoclimatic conditions (southern and northern Italy) where the SWAP model was calibrated and validated. The standard approaches clearly overestimated the dry conditions in the SMCS. Being mainly climate based, however, they were able to separate different pedoclimatic settings. In contrast, the physically based approach showed realistic results in terms of SMR estimation but did not differentiate pedoclimatic settings for either Soil Taxonomy or ICOMMOTR approaches. Three possible alternatives to SMR classification were investigated: (i) applying the ICOMMOTR classification approach supported by the use of hydrologic Richards-based models; (ii) changing the rules of Soil Taxonomy for classifying SMR and SMCS to better fit "real" soil hydrologic behavior; and (iii) continuing to apply Newhall and Billaux models but clearly clarifying their climate-driven criteria.