C.A. Reynolds, T. J. Jackson, and W.J. Rawls. 1999. Estimated Available
Water Content from the FAO Soil Map of the World, Global Soil Profile Databases,
and Pedo-transfer Functions
Principal Investigators:
C.A. Reynolds, T. J. Jackson, and W.J. Rawls
Agricultural Research Service
U.S. Dept. of Agriculture
Summary:
Spatial soil water-holding capacities were estimated for the Food and Agriculture
Organization (FAO) digital Soil Map of the World (SMW) by employing continuous
pedo-transfer functions (PTF) within global pedon databases and linking
these results to the SMW. The procedure first estimated representative
soil properties for the FAO soil units by statistical analyses and taxo-transfer
depth algorithms (FAO, 1996]. The representative soil properties
estimated for two-layers of depths (0-30 and 30-100 cm) included: particle-size
distribution; dominant soil texture; organic carbon content; coarse fragments;
bulk density, and porosity. After representative soil properties for the
FAO soil units were estimated, these values were substituted into three
different pedo-transfer functions (PTF) models by Rawls, et al [1982],
Saxton, et al [1986], and Batjes [1996]. The Saxton PTF model was finally
selected to calculate available water content because it only required
particle-size distribution data and results closely agreed with the Rawls
and Batjes PTF models that used both particle-size distribution and organic
matter data. Soil water-holding capacities were then estimated by multiplying
the available water content by the soil layer thickness and integrating
over an effective crop root depth of one meter or less (i.e., encountered
shallow impermeable layers). All soil property images are provided
for easy introduction into spatial water balance models. These raster
images have the same 5-minute spatial resolution of the original SMW to
preserve the integrity of the original data.
Primary References:
C.A. Reynolds, T. J. Jackson, and W.J. Rawls. 1999. Estimating Available
Water Content by Linking the FAO Soil Map of the World with Global Soil
Profile Databases and Pedo-transfer Functions. Proceedings of the AGU 1999
Spring Conference, Boston, MA. May31-June 4, 1999.
Reynolds, Jackson, and Rawls Estimated
Available Water Content
DATASET DESCRIPTION
Dataset Description
INTEGRATED DATASET
DataSet Citation:
C.A. Reynolds, T. J. Jackson, and W.J. Rawls. 1999. Estimated Available
Water Content from the FAO Soil Map of the World, Global Soil Profile Databases,
Pedo-transfer Functions. Data from the USDA Agricultural Research
Service. Published by the NOAA National Geophysical Data Center,
Boulder, CO. http://www.ngdc.noaa.gov/seg/fliers/se-2006.shtml
Projection:
Cartesian Orthonormal Geodetic (lat/long)
Spatial Representation:
Aggregate values for 1-degree grid cells (method varies)
Temporal Representation:
Static Modern Composite (circa 1990's)
Data Representation:
1-byte integers: quantitative values and background (no-data) type classes
for 1-degree cells.
Layers and Attributes:
Eleven independent multiple attribute layers
Dataset Description
DESIGN
Variables:
Bulk Density, Coarse Fragments, Depth, Organic Carbon, Porosity, Soil Fraction,
Texture, Water Holding Capacity
Origin:
Research dataset produced from the FAO Soil Map of the World, Global Soil
Profile Databases, and Pedo-transfer Functions.
Geographic Reference:
Unprojected geographic grid. Cartesian
Orthonormal Geodetic (lat/long)
Geographic Coverage: Global
Global
Maximum Latitude: +90 Degrees (N)
Minimum Latitude: -90 Degrees (S)
Maximum Longitude: +180 Degrees (E)
Minimum Longitude: -180 Degrees (W)
Geographic Sampling:
Integrated values for 1 degree grid cells
Time Period:
circa 1990's
Temporal Sampling:
Modern Composite
Dataset Description
SOURCE
Source Data Citation:
C.A. Reynolds, T. J. Jackson, and W.J. Rawls. 1999. Estimating Available
Water Content by Linking the FAO Soil Map of the World with Global Soil
Profile Databases and Pedo-transfer Functions. Proceedings of the AGU 1999
Spring Conference, Boston, MA. May 1-June 4, 1999.
Contributor:
C.A. Reynolds, T. J. Jackson, and W.J. Rawls
Agricultural Research Service
U.S. Dept. of Agriculture
BARC-West, Bldg 007, Rm 104
Beltsville, Maryland, 20708
Distributor:
U.S. Dept. of Agriculture
BARC-West, Bldg 007, Rm 104
Beltsville, Maryland, 20708
Tel: 301-504-7490
FAX: 301-504-8931
Date of Production:
1999
Lineage & Contacts:
-
Data Production:
C.A. Reynolds, T. J. Jackson, and W.J. Rawls
Agricultural Research Service
U.S. Dept. of Agriculture
BARC-West, Bldg 007, Rm 104
Beltsville, Maryland, 20708
-
Data Integration:
John J. Kineman and Joshua Klaus
NOAA National Geophysical Data Center
325 S. Broadway, E/GC1
Boulder, CO 80303 USA
fax: (303) 497-6513
Email: jkineman@ngdc.noaa.gov
Web: http://www.ngdc.noaa.gov/seg/eco.
Dataset Description
ADDITIONAL REFERENCES
Dataset Description
FILE LISTS
Reynolds, Jackson, and Rawls Estimated
Available Water Content
DATASET ELEMENT DESCRIPTIONS
Organic Matter
Description:
Percentage of soil organic matter (carbon) at two depths: 1) 0 - 30 cm
and 2) 30 - 100 cm depth.
Structure:
Vector polygon file in a Geodetic
(latitude/longitude) reference system: 1 degree resolution
Raster data file: 1-degree Cartesian
Geodetic (latitude/longitude) 4320x2160 grid
Series:
none
System Files:
Notes:
<text>
Porosity
Description:
Porosity of the soil measured as a percentage.
Structure:
Vector polygon file in a Geodetic
(latitude/longitude) reference system: 1 degree resolution
Raster data file: 1-degree Cartesian
Geodetic (latitude/longitude) 4320x2160 grid
Series:
none
System Files:
Notes:
<text>
Texture
Description:
Soil classifications from two sources. The first is from FAO and
divides soil into three classes: coarse, medium and fine. The
second data classification is from the USDA and divides the soil into 13
categories: sand, loamy sand, sandy loam, silt loam, silt, loam,
sandy clay loam, silty clay loam, clay loam, sandy clay, silty clay, clay
and salt flats.
Structure:
Vector polygon file in a Geodetic
(latitude/longitude) reference system: 1 degree resolution
Raster data file: 1-degree Cartesian
Geodetic (latitude/longitude) 4320x2160 grid
Series:
none
System Files:
Notes:
<text>
Water Holding Capacity
Description:
Amount of water that the soil can hold, measured in mm, at two depths of
soil.
Structure:
Vector polygon file in a Geodetic
(latitude/longitude) reference system: 1 degree resolution
Raster data file: 1-degree Cartesian
Geodetic (latitude/longitude) 4320x2160 grid
Series:
none
System Files:
Notes:
<text>
Fraction of Silt
Description:
Percentage of Silt at two different layers (0-30 cm) (30-100 cm)
Structure:
Vector polygon file in a Geodetic
(latitude/longitude) reference system: 1 degree resolution
Raster data file: 1-degree Cartesian
Geodetic (latitude/longitude) 4320x2160 grid
Series:
none
System Files:
Notes:
<text>
Fraction of Sand
Description:
Percentage of Sand in soil at two different layers (0-30 cm) (30-100 cm)
Structure:
Vector polygon file in a Geodetic
(latitude/longitude) reference system: 1 degree resolution
Raster data file: 1-degree Cartesian
Geodetic (latitude/longitude) 4320x2160 grid
Series:
none
System Files:
Notes:
<text>
Fraction of Clay
Description:
Percentage of Clay in soil at (0-30 cm) (30-100 cm)
Structure:
Vector polygon file in a Geodetic
(latitude/longitude) reference system: 1 degree resolution
Raster data file: 1-degree Cartesian
Geodetic (latitude/longitude) 4320x2160 grid
Series:
none
System Files:
Notes:
<text>
Bulk Density
Description:
Overall Density of soil at two levels (0-30cm and 30-100cm)
Structure:
Vector polygon file in a Geodetic
(latitude/longitude) reference system: 1 degree resolution
Raster data file: 1-degree Cartesian
Geodetic (latitude/longitude) 4320x2160 grid
Series:
none
System Files:
Notes:
<text>
Coarse Fragments
Description:
Percent of Coarse Fragments per soil unit.
Structure:
Vector polygon file in a Geodetic
(latitude/longitude) reference system: 1 degree resolution
Raster data file: 1-degree Cartesian
Geodetic (latitude/longitude) 4320x2160 grid
Series:
none
System Files:
File type |
Metadata |
Data |
Raster grid |
frag1.doc |
frag1.img |
Raster Series |
|
|
Vector Point |
|
|
Vector Line |
|
|
Vector Polygon |
|
|
Attribute Table |
coarsefrg.csv |
|
Color Palette |
|
|
Projection |
|
|
Notes:
<text>
Landscape Classes
Description:
Landscape classifications where no quantitative soil data exists. (i.e.
salt (113), inland water (114), rock debris (115), glaciers (116), shifting
dunes (117) ) These landscape types are collectively coded as "no
data" in the quantitative maps.
Structure:
Vector polygon file in a Geodetic
(latitude/longitude) reference system: 1 degree resolution
Raster data file: 1-degree Cartesian
Geodetic (latitude/longitude) 4320x2160 grid
Series:
none
System Files:
Notes:
<text>
Reynolds, Jackson, and Rawls Estimated
Available Water Content
TECHNICAL REPORTS
Estimating Soil Water-Holding Capacities by Linking
the FAO Soil Map of the World with Global Pedon Databases and Continuous
Pedo-transfer Functions
Technical Report
Estimating Soil Water-Holding Capacities by Linking the FAO Soil Map of
the World with Global Pedon Databases and Continuous Pedo-transfer Functions
C.A. Reynolds, T. J. Jackson, and W.J. Rawls
Agricultural Research Service
U.S. Dept. of Agriculture
BARC-West, Bldg 007, Rm 104
Beltsville, Maryland, 20708
Tel: 301-504-7490
FAX: 301-504-8931
e-mail: tjackson@hydrolab.arsusda.gov
reynolds@hydrolab.arsusda.gov
Reference:
When using a spatial database provided on this CD-ROM, please refer
to the following citation:
C.A. Reynolds, T. J. Jackson, and W.J. Rawls. 1999. Estimating Available
Water Content by
Linking the FAO Soil Map of the World with Global Soil Profile Databases
and Pedo-
transfer Functions. Proceedings of the AGU 1999 Spring Conference,
Boston, MA. May
31-June 4, 1999.
Distribution Liability:
This CD-ROM was prepared by an agency of the United States Government.
Neither the United
States Government nor any agency thereof, nor any of their employees,
make any warranty, expressed or implied, or assumes any legal liability
or responsibility for the accuracy, completeness, or misuse of the data,
or for damage, transmission of viruses or computer contamination through
the distribution of these data sets or for the usefulness of any information,
apparatus, product, or process disclosed in this report, or represents
that its use would not infringe privately owned rights. Reference therein
to any specific commercial product, process, or service by trade name,
trademark, manufacturer, or otherwise does not necessarily constitute or
imply its endorsement, recommendation, or favoring by the United States
Government or any agency thereof. Any views and opinions of authors expressed
herein do not necessarily state or reflect those of the United States Government
or any agency thereof.
File Formats:
The spatial soil database is provided in two file formats: IDRISI and
ARCVIEW.
These files can be easily exported into other common GIS software packages.
Important geographical information is provided in the IDRISI documentation
(*.doc) files:
file title : varies
data type : byte or real
file type : binary
columns : 4320
rows : 2160
ref. system : latlong
ref. units : deg
unit dist. : 1
min. X : -180
max. X : 180
min. Y : -90
max. Y : 90
pos'n error : unknown
resolution : 8.333334E-02
legend cats : varies
IDRISI or ARCVIEW software is not included with this CD-ROM, but IDRISI
software is available from http://www.clarklabs.org/ and ARCVIEW software
from http://www.esri.com/.
IDRISI is developed, distributed, and supported by the Clark Labs at
Clark University in Worcester, Massachusetts. The Clark Labs (formerly
The IDRISI Project) is a non-profit educational and research institution.
IDRISI is a user-friendly Geographic Information System (GIS), intended
to be affordable to all levels of users, and runs on IBM-PC compatible
platforms (DOS, Windows 3x, 95, and NT). No expensive graphics cards
or peripheral devices are required to make full use of the analytical power
of the software. IDRISI provides a wide spectrum of raster functionality
such as database query, spatial modeling, and image enhancement and classification.
Version 3 of IDRISI for Windows is a 32-bit application that is currently
under development and planned for release in 1999.
ARC/INFO and ARCVIEW software is developed, distributed, and supported
by Environmental Systems Research Institute (ESRI), Inc. at Redlands, CA.
ARC/INFO or the Spatial Analyst extension for ARCVIEW is required to display
the ARCVIEW_GRID images.
ABSTRACT
Spatial soil water-holding capacities were estimated
for the Food and Agriculture Organization (FAO) digital Soil Map of the
World (SMW) by employing continuous pedo-transfer functions (PTF) within
global pedon databases and linking these results to the SMW. The
procedure first estimated representative soil properties for the FAO soil
units by statistical analyses and taxo-transfer depth algorithms (FAO,
1996]. The representative soil properties estimated for two-layers
of depths (0-30 and 30-100 cm) included: particle-size distribution; dominant
soil texture; organic carbon content; coarse fragments; bulk density, and
porosity. After representative soil properties for the FAO soil units were
estimated, these values were substituted into three different pedo-transfer
functions (PTF) models by Rawls, et al [1982], Saxton, et al [1986], and
Batjes [1996]. The Saxton PTF model was finally selected to calculate available
water content because it only required particle-size distribution data
and results closely agreed with the Rawls and Batjes PTF models that used
both particle-size distribution and organic matter data. Soil water-holding
capacities were then estimated by multiplying the available water content
by the soil layer thickness and integrating over an effective crop root
depth of one meter or less (i.e., encountered shallow impermeable layers).
All soil property images are available on a CD-ROM for easy introduction
into spatial water balance models. These raster images have the same
5-minute spatial resolution of the original SMW to preserve the integrity
of the original data.
1. INTRODUCTION
Spatial hydrologic models that calculate water balances,
simulate climate, or estimate crop growth require available water-holding
capacity information. However, many digital soil databases do not provide
soil hydraulic information because soil hydraulic data are often missing
from soil profile databases or it is not physically possible to obtain
sufficient numbers of direct measurements across a watershed to adequately
reflect the spatial heterogeneity of soils. Due to lack of soil hydraulic
data, the simplest water balance models assume water-holding capacities
are spatially invariant, while more complex models estimate the spatial
variability of soils by linking soil survey maps with representative soil
profiles.
One indirect method to predict water retention properties
from physical soil data is to utilize pedo-transfer functions (PTF). PTF
models are typically regression equations derived from large soil profile
data sets that show high levels of statistical confidence between two or
more soil parameters. PTF models that predict available water content are
classically derived from particle-size distribution, organic matter content,
or bulk density data because these soil properties have a dominant role
in determining water-holding characteristics of soils. The statistical
nature of PTF models implies they should be derived from large profile
data sets with a diverse selection of soils so that they are adaptable
in other settings. PTF models are particularly useful when linking different
national pedon databases together because soil texture data is routinely
available from soil surveys and they are less expensive and time-consuming
to obtain than determining hydraulic properties by direct methods [Rawls,
et al, 1992, and Tietje and Tapkenhinrichs, 1993]. In addition, the
accuracy of PTF models are typically suitable when working at scales of
1:50,000 or smaller [Wösten and van Genucthen, 1988].
PTF models may be further divided into class and
continuous PTF sub-divisions. A class PTF predicts the hydraulic characteristics
by using soil texture classes while a continuous PTF predicts the hydraulic
characteristics using the actual measured percentages of clay, sand and
silt. Generally speaking, a class PTF is easier to use, but its accuracy
is limited because the approach provides one average hydraulic characteristic
for each textural class. In contrast, a continuous PTF requires the
exact textural composition of a soil, but has the positive effect that
the predicted hydraulic characteristics are likely to be more accurate
than class PTF estimates. [Wösten, et al, 1995].
The main objective of this study was to estimate
the water holding capacities of the FAO soil units by linking the Soil
Map of the World (SMW) with continuous PTF functions. Three different PTF
equations developed by Rawls, et al, [1982], Saxton, et al, [1986], and
Batjes [1996a] were compared with each other and the Saxton PTF model was
finally selected to estimate available water-holding capacity from the
SMW. The Rawls and Saxton PTF models, were derived from over 1300 soil
profiles extracted from North American soils, and were chosen for this
study because these models are versatile and require few parameters [Kern,
1995 and Tietje and Tapkenhinrichs,1993]. The Batjes model was compared
to these models because it was derived from a large global soil profile
data set with soil profiles extracted from several continents. In
addition, all three of the PTF models were selected because they can estimate
field capacity and permanent wilting point values with different matric
potential limits such as field capacities defined at –1/20, –1/10, and
–1/3 bars.
2. METHODOLOGY
Soil units are the most basic unit of the Soil Map of
World (SMW) legend which is based on the FAO soil classification system
[FAO-UNESCO, 1974]. Soil units are distinguished from one another by diagnostic
profile horizons having similar soil properties that evolved from comparable
pedo-genetic processes and climates. One to eight soil units comprise a
mapping unit, with the SMW containing nearly 5000 mapping units and a total
of 106 soil units under 26 major headings. The SMW demarcates the
boundaries of the mapping units, but soil properties and depth are only
indirectly related to soil units.
The FAO [1996] and Batjes [1997] made great efforts
to relate the FAO soil units to physical soil characteristics by statistically
analyzing global pedon databases to estimate soil texture, bulk density,
and organic matter content. The FAO [1996] and Batjes [1996a] also developed
empirical taxo-transfer rules (TTR) to estimate available water content
and depth, where the sub-strata was inferred by classification names [FAO,
1996] or no effort was made to differentiate the top- and sub-strata [Batjes,
1996a].
This study follows many of the same procedures as described by the
FAO [1996] and Batjes [1997 and 1996a] for deriving representative profile
and depth properties for the FAO soil units. However, the methodology
differs from previous approaches by estimating available water content
from continuous PTF models for both the top (0-30 cm) and sub strata (30-100cm)
of the soil units. The advantage of this approach, in contrast to
TTR models, is representative texture data is derived from statistical
analysis [FAO, 1996] for two layers of depth and PTF models can more easily
be utilized for developing spatial water-holding capacity images from other
digital databases. In addition, all digital images for this study
have 5-minute spatial resolution to preserve the spatial integrity of the
SMW, which differs from the ½-degree images developed by Batjes
[1997]. Finally, the TTR step of reclassifying available water into generalized
classes was eliminated [FAO, 1996, and Batjes, 1996a, and 1997], so that
the final images can be easily introduced into spatial water-balance models
or compared with the State Soil Geographic (STATSGO) database of the United
States [Reynolds and Jackson, 1999]. These retained numerical values do
not imply improved accuracy, but were kept for spatial modeling purposes.
The general procedure used to develop the spatial soil property images
and a global water-holding capacity image is illustrated in Figure 1 and
summarized as follows:
1. Estimate representative soil properties (0-30
cm and 30-100 cm depths) for each FAO soil unit by statistical analysis
of two global pedon databases [FAO, 1996].
2. Estimate available water content for the FAO
soil units by substituting representative soil texture and organic matter
data into pedo-transfer functions (PTF).
a. Rawls, et al [1982] and
Batjes, et al [1996] PTF models utilized both soil texture and organic
matter.
b. Saxton, et al [1986]
model utilized only soil texture data.
3. Estimate soil depth from taxo-transfer depth
algorithms [FAO, 1996] and estimate water-holding capacities by multiplying
available water content, rock fragment content, and depth.
4. Link generalized soil properties for each FAO
soil unit to the digital SMW with area composition rules [FAO, 1978] and
weighted-area average formula.
Each of the above steps is described in detail by the following sub-sections.
2.1 Digital Soil Databases
2.1a. FAO Soil Map of the World
The FAO-UNESCO Soil Map of the World (SMW) at 1:5 M
scale is undoubtedly the most comprehensive soil map with global coverage
[Sombroek, 1989, and Nachtergaele, 1996]. Development of the SMW was initiated
in 1961, with the first hardcopy map published in 1971 and the last map
of the 10 volume series completed in 1981 [FAO-UNESCO, 1971-1981].
The SMW was compiled from over 600 national soil maps and over 11,000
ancillary maps, from which most of the source soil maps and profiles were
provided by national soil organizations. Many of these source maps
varied widely in reliability, scales, and methodologies. As a method to
record the final reliability of SMW for each continent, the source maps
were ranked according to the level of soil survey in terms of systematic,
reconnaissance, and general information surveys. Based on these rankings,
it appears only one fifth of the worlds soil had been surveyed [Zobler,
1986].
The original SMW has nearly 5000 mapping units,
and the component soils of the mapping units are described on the original
map sheets. Defining boundaries of the mapping units entailed large amounts
of interpretation because economic limitations prevent a one-to-one ratio
of profile descriptions to component soil units. Gaps of missing data were
interpolated from ancillary sources such as aerial photos, remote sensing
images, climate, topography, geology, and vegetation maps.
When a mapping unit is not homogenous, the component
soil units are identified as one of the following types: dominant (covering
most of the mapping unit), associated (at least 20 percent of the mapping
unit area), or inclusion soils (less than 20 percent of the mapping unit
area). The exact area and distribution of the soil units within in a mapping
unit are not known, but soil unit area distribution can be estimated from
area composition rules that consider the number of soil units per mapping
unit [FAO, 1978]. Mapping units are classified into three slope classes
defined as flat (0-8%), mild (8-30%), and steep (>30%).
Soil units are further divided into three textural classes of coarse,
medium, and fine which are defined by their relative proportions of clay
(less than 2 ?m), silt (2-50 ?m), and sand (50-2000 ?m) content. Miscellaneous
soil units may also be classified as glacial, salt flats, dunes or shifting
sands, rock debris or desert detritus, or no data. The dominant soil
unit may also be classified into one of twelve phases, where soil phases
are not diagnostic for the separation of soil unit characteristics but
have significant impacts on land management uses. The twelve soil
phases are stony, lithic, petric, petrocalcic, petrogypsic, petroferric,
phreatic, fragipan, duripan, saline, sodic, and cerrado. Major climate
variants, such as permafrost and intermediate permafrost, were also denoted
on the mapping units.
In 1984, Environmental Systems Resources Institute
[ESRI, 1984] digitized the original SMW in vector form which had a bipolar
conic conformal projection for the western hemisphere and the Miller oblated
stereographic projection for the eastern hemisphere. The Global Resources
Information Database (GRID) project of the United Nations Environmental
Programme [UNEP/GRID, 1984] later re-projected the maps for both hemispheres
to a plate carree projection, and converted the SMW from a vector format
to a raster grid. The digital SMW provided the number of soil units comprising
each mapping unit and representative soil profiles for the FAO soil units
were provided by the FAO-UNESCO [1971-1981]. When converting the SMW to
digital form, UNEP/GRID [1984] noted that some FAO soil units were not
differentiated into textural groups, and several mapping units were listed
as one of the 26 general soil groups instead of being listed as one of
the more detailed 106 soil units. These classification oversights
make it a great challenge to infer representative soil properties from
such broad and generalized categories.
2.1b. Global Pedon Databases
During the late 1980s and early 1990s, several researchers
derived spatial soil properties by linking the UNEP/GRID digital SMW to
published profile descriptions from the FAO-UNESCO [1971-1981]. The linking
process involved grouping the profile descriptions, or pedons, into FAO
soil units and determining representative properties of these units at
different depths. Most of these early attempts to derive spatial soil properties
were intended as inputs for climate models, which degraded the original
5-minute resolution of the SMW to 1-degree [Zobler, 1986, Webb, et al,
1991 and 1993, Wilson and Henderson-Sellers, 1985, Sellers, et al, 1995]
and ½-degree grids [Batjes, 1996b, and Dunne and Willmott, 1996].
The FAO-UNESCO [1974] classification system also
forced researchers to devise a crude method for estimating additional soil
texture groups from the three broad FAO textural classes and to derive
representative soil properties of the FAO soil units from the original
and limited FAO-UNESCO [1971-1981] pedon database. These limitations and
errors were then spatially propagated when extrapolating representative
profile descriptions to generalized soil mapping units.
In recent years, global profile data sets have been
improved with the FAO-Soil Database System (SDB) containing over 1700 soil
profiles and the World Inventory of Soil Emission Potentials (WISE) database
of International Soil Reference and Information Centre (ISRIC) holding
over 4000 profiles [FAO, 1996, and Batjes, et al, 1997]. The FAO-SDB is
regarded as an improvement to the original FAO-UNESCO [1971-1981] data
set, while the WISE profile database is regarded as the most complete global
data set currently available. The soil profiles of the WISE data
set were contributed from national soil survey organizations, ISRIC’s-Soil
Information System (ISIS), FAO’s-SDB, and the United States Department
of Agriculture (USDA)-Natural Resources Conservation Service (NRCS) profile
data set.
2.2 Estimate Representative Soil Properties from Global Pedon Databases
This study followed the same procedures as described
by FAO [1996] and Batjes [1996a and 1997] to infer representative physical
soil properties for each soil unit by statistical analysis. These data
sets were processed by removing outliers and by standardizing horizon depths
into two layers (0-30 and 30-100 cm). The profiles were then separated
into FAO soil units and divided into profiles containing four soil textural
classes (coarse, medium, fine, and undifferentiated), totaling over 400
functional soil unit categories (106 soil units multiplied by 4 textural
classes). Statistical averages and medians for each FAO soil unit were
then determined to develop representative soil properties. Statistical
medians, instead of averages, were used for the WISE data set to reduce
the effect of outliers. For those soil unit categories with less than five
samples, expert opinion was consulted to estimate reasonable values.
Statistical averages from the FAO-SDB data set were
used for determining representative organic carbon, bulk density, and porosity,
where organic carbon data was measured by the Walkley-Black method, and
bulk density (?b) was measured according to the core-method [FAO, 1996].
Porosity (N) data was derived from bulk density by the following equation,
N = 1 – ?b /?s, where particle density (?s) was assumed as 2.65 gm/cm3.
Data for estimating the particle-size distribution
were obtained from the complete WISE database of over 4000 soil profiles
which is currently being reviewed and modified by expert opinion [IIASA,
et al, 1998]. However, the database file listed the statistical median
sand/clay/silt content for each FAO soil unit according to its respective
textural class of coarse, medium, fine, or undifferentiated. The percent
sand/clay/silt combinations were modified by proportionally adjusting the
sand/clay/silt fractions to 100, and replacing missing data (less than
five samples) by estimates from the FAO-SDB data set. The WISE and FAO-SDB
data sets often did not have undifferentiated data for the soil groups,
and these categories were approximated by substituting textural data for
soil units with the same textural occurrence [refer to Table 4 from Batjes,
et al, 1997].
A summary of the source files utilized to derive
the final two-layered representative soil unit properties is listed in
Table 1. The WISE median texture file [texture1.xls file from Table 1]
is available on the CD-ROM and the other files listed in Table 1 are available
from the FAO [1996] CD-ROM.
2.3 Estimate Available Water Content with Pedo-transfer Functions
Plant-available water-holding capacity information is
important data for studying the response of vegetation and hydrologic
systems, where plant-available water content is defined as the difference
of soil moisture content between field capacity (?f) and permanent wilting
point (?w). Field capacity and permanent wilting point are concepts
that define the upper and lower limits of soil-water available for plant
consumption. Neither field capacity of permanent wilting point is
a sharply defined quantity because they are related to many factors such
as soil profile characteristics, soil depth, crop type, plant growth stage,
and climate.
Field capacity is defined as the soil moisture content after a soil
has been thoroughly wetted to saturation and allowed to drain for 2-3 days
[Soil Survey Division Staff, 1993]. This definition makes the field capacity
concept more applicable to coarse-textured soils rather than to fine-textured
soils, because the rate of drainage is faster for coarse-soils due to larger
pores, whereas fine-textured soils may drain for weeks or months after
wetting due to smaller pore sizes and stronger matric potentials. Due to
these discrepancies, field capacity is frequently defined as the soil water
content corresponding to measured soil matric potentials, with the United
Kingdom, the Netherlands, and the United States accepting different values
of –1/20, -1/10, and
–1/3 bar, respectively.
Permanent wilting point is defined as the soil moisture content below
which plants wilt during the day and cannot recover overnight. Early experiments
with sunflowers found the permanent wilting point for a wide range of soils
closely correlated with a soil pressure potential of –15 bars. The
–15 bars value is now commonly used to estimate permanent wilting point,
but many drought-tolerant crops, such as wheat have the ability to survive
and extract water at levels well below this value.
In view of the definition limitations associated with field capacity
and permanent wilting point concepts, soil physicists tend to agree on
the lower limit of -15 bars for permanent wilting point, but disagree on
the soil matric potential for determining field capacity. For this study,
USDA standards of -1/3 and -15 bars were assumed as the respective upper
and lower limits for available water content [Soil Survey Division Staff,
1993]. However, all three PTF models used in this study can easily
estimate available water content for field capacity limits defined at -1/10
or –1/20 bars.
Rawls, et al, [1982] developed multiple linear regression equations
to estimate upper and lower available water limits by setting particle-size
distribution and organic matter as the independent variables. Kern [1995]
demonstrated that the Rawls equations, derived from more than 5300 horizons
from 32 states, apply for a diverse range of soils, unlike other PTF equations
derived from limited soil profile databases. The Rawls’ PTF regression
equations estimate the Brooks-Corey soil water retention parameters:
?f = 0.2576 - 0.0020Psand + 0.0336 Pclay + 0.0299Pom
(1)
?w = 0.026 + 0.0050Pclay + 0.0158Pom (2)
where, ?f is field capacity (percent volume) approximated by water retention
at -1/3 bar, ?w is the permanent wilting point (percent volume) approximated
by water retention at
-15 bars, and Psand, Pclay, and Pom are percentages of clay, sand,
and organic matter, respectively. Available water content, ??, is the absolute
difference between ?f and ?w.
Saxton, et al, [1986] later modified Rawls’ equations by eliminating
the organic matter term because the absolute difference of ?f and ?w is
relatively small for different concentrations of organic matter, as both
?f and ?w change at relatively the same rate [Kern, 1995]. The Saxton
model is:
? = (?/A)1/B (3)
where, ? = soil-water content, percent volume
A= 100.0 exp[-4.396 - 0.0715(Pclay) - 0.0004880(Psand)2 - 0.00004285(Psand)2Pclay]
B= -3.140 - 0.00222(Pclay)2 - 0.00003484(Psand)2 (Pclay)
?= water potential between 1500 and 10 kPa, or ?f =33 kPa and
?w =1500 kPa
Psand and Pclay = percentages of sand and clay, respectively
Both the Saxton and Rawls models are valid for sand and clay percentages
greater than 5 percent and clay content less than 60 percent.
Batjes, et al, [1996a] also developed step-wise multiple linear regression
equations by using measured available water data from the WISE database
having approximately 3000 soil profiles and over 15,000 horizons worldwide.
The Batjes equations are:
?f = 0.3624Pclay + 0.11705Psand + 1.6054Pom
(4)
?w = 0.4600Pclay + 0.3045Psand + 2.0703Pom
(5)
where, Psand, Pclay, and Pom are percentages of clay, sand, and organic
matter, respectively. The Batjes model is valid for percentages of
sand, clay and silt greater than 5 percent.
2.4 Estimate Depth and Calculate Water-holding Capacity for each FAO Soil
Unit
Water-holding capacity is a function of the available water content, rock
fragment content (>2 mm), and soil depth. Soils with abundant rock fragments
reduce water-holding capacities and increases infiltration rates, and accordingly
water-holding capacities are commonly adjusted by a rock fragment factor.
Unfortunately, the SMW does not contain any explicit information of coarse
fragments and soil depth, but coarse fragments were estimated from the
WISE data set [Batjes, 1997] and depth was inferred from the FAO [1996]
depth algorithm based upon taxonomic classification, phase, and slope.
The percent of coarse fragments per soil unit were statistically determined
from the WISE profile data set by Batjes [1997] who grouped each soil unit
into four general coarse fragment classes (0-5%, 6-15%, 16-40%, and greater
than 40%). Values of 5, 15, 40, and 50 percent were thus assumed
for each respective coarse fragment class. The FAO [1996] depth algorithm
is based on taxo-transfer rules which utilize classification name, soil
phase of the dominant unit, slope class, and relative soil unit area covered
per mapping unit. For example, the algorithm reduces soil depth for soils
located on steep slopes and for soils having Lithic, Petrocalcic, Petrogypsic,
Petroferric or Duripan phases. This study modified the FAO [1996] depth
algorithms slightly by assuming the maximum soil unit thickness was 250
cm or less in order to produce depth images of 8-bits instead of 16-bit
images. Water-holding capacities were also calculated for an effective
root zone of one-meter or less (i.e., encountered shallow impermeable layer)
because one meter is the effective rooting depth commonly assumed for most
crops.
In summary, the source files listed in Table 2 were used to estimate
soil water-holding capacities for each functional FAO soil unit category.
The files last three files listed in Table 2 are available on this CD-ROM
and the other files are available from the FAO [1996] CD-ROM.
2.4 Link Generalized Profile Descriptions to the SMW
The generalized soil properties for each soil unit were linked to the digital
SMW by mapping unit codes. This procedure first joined two tables
together, where the source table is the representative soil properties
for each soil unit (su_summ.txt from Table 2) and the destination table
is the soil unit percent area covered within each mapping unit (world764.dat
from Table 2). The average soil property per mapping unit was then estimated
by a weighted-area average formula,
Xmu = ?i=1 Ai * Xi
(6)
where i is the soil unit within the mapping unit, A is percent area
covered by the soil unit per mapping unit (World764.dat file from Table
2), and Xmu is the average soil property value of the mapping unit.
Calculations for total available water content (TAWC) are related to
depth and coarse fragments as:
TAWC = (1-fc)*[AWCts * d0-30 +d30-100AWCss]
(7)
where fc is percent of coarse fragments, AWCts and AWCss are the available
water content for the topsoil and subsoil layers, respectively, and d is
the depth from 0-30 and 30-100 cm. Results from equation (7) for
each soil unit were then linked to the binary SMW image [worldbin.img file
from Table 2] to generate the water-holding capacity image. In addition,
the final soil property tables, images, and metadata files from equation
(6) are stored under the respective soil property directory on the CD-ROM.
3. RESULTS
The scatter diagrams shown in Figures 2-4 compare available water content
for each FAO soil unit estimated by the Rawls, Saxton, and Batjes PTF models.
Figures 2 and 3 indicate the Batjes model agrees with the other two models,
even though the Rawls and Saxton models were derived from only North American
soils and Batjes model was derived from a worldwide distribution of soils.
Surprisingly, the Rawls and Saxton models in Figure 3, have the greatest
disagreement even though they were derived from similar profile databases.
Figures 3 and 4 indicate the Saxton model differs for sandy soils with
low water-holding capacities when compared to the other models, but an
advantage of the Saxton model is organic matter data is not required unlike
the Rawls and Batjes models. The Saxton model was therefore chosen for
calculating available water because it agreed with both models and did
not require organic matter data, a parameter regarded as having high variability
and uncertainty within most soil profile databases.
It should be noted that Histosols, Ferralsols, Andosols, and Vertisols
have unique soil properties, which may prevent accurate estimates of available
water from PTF models. Histosols contain large amounts of organic
matter and are typically excluded from the derivation of the PTF equations.
Accordingly, Histosols were barred from PTF calculations and assigned an
alternative water-holding capacity value of 249 mm/m for the production
of 8-bit images (i.e., values 250-254 assigned classes of sand dunes, glaciers,
rock debris, and inland water, respectively).
Ferrasols tend to have water-holding capacities of 80-90 mm/m [FAO,
1996 and Batjes, 1996a] which is lower than indicated by their soil texture.
All three PTF equations indeed over-estimated their storage capacity by
an average of 10-20 mm/m. Andosols are developed in volcanic ash
tuff and have water-holding capacities of approximately 190-200 mm/m [FAO,
1996 and Batjes, 1996a], which tends to be higher than indicated by their
soil texture. All three PTF equations in this study underestimated
the storage capacity for Andosols by approximately 40 mm/m. Vertisols
are fine-textured cracking clay soils dominated by montmorillonitic and
smectite clays, which swell when wet and crack when dry. High clay contents
make it difficult to wet and dry vertisols entirely, which leads to water
storage capacities higher than indicated by their soil texture. However,
all three PTF equations did not seem to under-estimate their water content
and instead closely agreed with published values of 130-135 mm/m [FAO,
1996 and Batjes, 1996a].
The water-holding capacities for each mapping unit as calculated by
the Saxton model were compared to the water-holding capacities calculated
by the FAO [1996] taxo-transfer rule (TTR) model. The FAO-TTR model used
qualitative descriptions of the FAO soil units to quantitatively estimate
available water content and depth for each mapping unit [results listed
in the smax1.asc file from FAO, 1996]. For this study, the water-holding
capacities from the FAO-TTR model were estimated by assuming values of
20, 40, 80, 125, 175, and 200 mm/m for the six FAO categories.
Figure 5 is the scatter diagram that compares the WHC estimates from
the Saxton PTF and FAO TTR models. The higher WHC values from the FAO TTR
model are due to the TTR model defining field capacity value at –1/20 bar,
instead of –1/3 bar as defined by the Saxton model, and no provision for
a coarse fragment reduction factor. In addition, the FAO-TTR model assumed
histosols, fluvisols, and gleysols were wetlands with no water-holding
capacities, and these are the values plotted along the abscissa axis in
Figure 5.
The WHC capacity images generated by the FAO-TTR and Saxton models
were also compared to readily available water-holding capacity data from
the Crop Production System Zone (CPSZ) project [van Velthuizen, et al,
1995 and FAO, 1998a]. The CPSZ data set divides northeastern Africa into
more than 1000 CPSZ zones and estimated the average readily available water-holding
capacity for each CPSZ zone. The vector file defining CPSZ units was placed
over the WHC images developed from the FAO-TTR and Saxton PTF models, and
average WHC data were extracted for each CPSZ unit.
Figures 6 and 7 are scatter diagrams that compare the average
WHC estimates extracted by CPSZ unit from the FAO-TTR and Saxton models
to the WHC estimates from the CPSZ database. These figures indicate that
both the FAO-TTR and Saxton PTF models estimated larger water-holding capacities
than the CPSZ database, because the field capacity and permanent wilting
points for the CPSZ database were defined differently. In addition,
the scatter is less in Figure 6 than in Figure 7 which indicates the WHC
estimates from Saxton PTF model more closely agree with the CPSZ database
than with the FAO-TTR model.
In summary, Figures 5 and 6 indicate that the WHC image developed
from the Saxton PTF model agreed with both WHC databases developed by the
FAO [1996] and van Velthuizen, et al, [1995], respectively. In addition,
the WHC image developed from the Saxton PTF model agreed with the State
Soil Geographic Database (STATSGO) at a 1:1 M scale [Reynolds and Jackson,
1999]. These results demonstrate that continuous PTF models have
the same ability as TTR models to estimate water-holding capacities for
the SMW.
4. DISCUSSION
Developing the SMW and compiling the global profile databases was an enormous
task involving many national and international organizations, and these
databases change constantly as more soil information is collected and digital
storage capabilities improve. Two current international cooperative projects
are designed specifically to update the SMW and global pedon databases;
the International Geosphere-Biosphere Programme- Data Information System
(IGBP-DIS) and the Soil and Terrain (SOTER). The IGBP-DIS project
aims to aid global modelers by expanding the existing profile attribute
information and thereby improve the statistical median estimates of soil
properties. Additional taxo-transfer and pedo-transfer functions
are also being developed and tested for inferring auxiliary soil properties
such soil organic carbon, soil nitrogen, water-holding capacity, soil thermal
properties, rooting depth, hydraulic conductance, and water regime [Scholes,
et al., 1995].
ISRIC is currently implementing the SOTER project which intends to
convert the different national taxonomy classifications to the revised
FAO [1990] classification system and produce a digital SOTER product at
a 1:5 M scale by the year 2002 [FAO, 1995, Natchergaele, 1996, and ISRIC,
1997]. Simultaneously, ISRIC is compiling national soil maps for developing
a global SOTER product at 1:1 M scale, but this product will not be released
until after the 1:5 M SOTER product is finished [ISRIC, 1997].
During the past two decades, many countries have compiled a new wealth
of soil information, producing digital GIS maps with attribute tables at
scales of 1:1 M and better. When modeling at the national or river basin
level, most of these recent national databases are recommended over the
SMW, because these finer resolution national databases are typically based
on more recent information and utilized standard soil sampling and classification
procedures.
Many digital soil databases with finer resolutions may not include
soil hydraulic properties due to lack of data, difference in data collection
procedures, or difference in soil hydraulic property definitions.
In these cases, employing PTF models as described can be useful for estimating
soil water-holding capacities or other hydraulic properties. For example,
the new release of the SOTER products for northeastern Africa at 1:1 M
scale and Central and Southern America at 1:5 M scale include representative
soil texture data [FAO, 1998a and 1998b], but do not include soil water-holding
capacities or soil hydraulic properties data. However, introducing continuous
PTF models into these digital products or other digital soil databases
can easily assist hydrologic modelers in predicting missing soil hydraulic
properties.
5. CONCLUSIONS
This study slightly modified the FAO [1996] method to calculate water holding-capacities
for the FAO soil units by introducing continuous PTF models. The
spatial soil property modifications for the SMW included: using a larger
global pedon database to statistically derive the representative particle-size
distributions; modifying the FAO depth algorithm to calculate an assumed
root depth of 1-meter or less; predicting available water content from
the Saxton PTF model; and calculating the water-holding capacity by multiplying
the available water content to the soil unit thickness for two-layers of
depth. The average soil property for each FAO mapping unit was then
computed by a weighted-area average formula and spatially linked to the
SMW.
Representative particle-size distribution data from the WISE pedon
database were substituted into the three continuous PTF models developed
by Rawls, et al, [1982], Saxton, et al, [1986], and Batjes [1996a] and
their results compared with one another. All three PTF models closely agreed
with each other, and they can easily estimate available water for field
capacity limits defined at –1/3, -1/10, or –1/20 bars. The Saxton PTF model
was finally selected for calculating available water-holding capacity between
–1/3 and –15 bars, because it did not require organic matter content data
and it agreed with the other two PTF models.
Available-water-holding capacity was then estimated for each FAO soil
unit by multiplying available water content estimates from the Saxton PTF
model, depth estimates from the FAO [1996] algorithm, and coarse fragment
content estimates from Batjes [1997]. The final available water-holding
capacity image was then compared to two other spatial WHC images by van
Velthuizen, et al, [1995] and FAO [1996]. The results indicated that continuous
PTF models are suitable for map scales of 1:5 M, but the method for determining
representative soil unit properties from global pedon databases is the
most critical input.
Compiling the SMW at a 1:5 M scale in the 1960s and 1970s was a prodigious
international task due to the large volumes of data, different national
soil classification systems, and inherent soil variability. Much of the
SMW is considered inaccurate today because it was compiled from source
maps of varied reliability; used an outdated classification system with
three broad textural classes; and was based on limited soil profile information.
These limitations were then propagated spatially when extrapolating soil
property information from detailed profile descriptions to generalized
soil mapping units.
Until more detailed global soil information becomes available, it is
recommended that spatial models at the national or sub-national level should
utilize national soil maps at 1:1 M scale or better, if available. These
national digital soil databases, such as STATSGO, should be more accurate
than the SMW because they are typically based on more recent and finer
resolution data sets that were derived from updated soil sampling procedures
and classification systems. However, the SMW linked to soil properties
derived from global pedon databases is still recommended for spatial models
at the global or continental scale model because of its comprehensive coverage.
5. REFERENCES
Batjes, N. H., A world dataset of derived soil properties by FAO-UNESCO
soil unit for global modeling, Soil Use Manage., 13:9-16, 1997.
Batjes, N.H., G. Fisher, F.O. Nachtergaele, V.S. Stolbovoy, and H.T.
van Velthuizen, Soil data derived from WISE for use in global and regional
AEZ studies (Version 1.0), Interim Report IR-97025/May, International Institute
for Applied Systems Analysis (IIASA), Austria, 1997.
Batjes, N.H., Development of a world data set of soil water retention
properties using pedotransfer rules, Geoderma, 71:31-52, 1996a.
Batjes, N.H., Documentation to ISRCI-WISE Global Data Set of Derived
Soil Properties on a 1/2o by 1/2o Grid (Version 1.0), Working Paper and
Preprint 96/05, International Soil Reference and Information Centre (ISRIC),
Wageningen, 1996b.
Dunne, K.A. and C.J. Willmott, Global distribution of plant-extractable
water capacity of soil, Int. J. Clim., 16, 841-859, 1996.
ESRI, UNEP/FAO World and Africa GIS Database, Final Report, Environmental
Systems Research Institute (ESRI), Redlands, California, 1984.
FAO-UNESCO, Soil Map of the World, Ten Volumes, Food and Agriculture
Organization, Rome, 1971-1981.
FAO-UNESCO, Soil Map of the World, Legend, Food and Agriculture Organization,
Rome, 1974.
FAO, Report on the Agro-Ecological Zones Project, Vol 1: Methodology
and Results for Africa, Food and Agriculture Organization, Rome, 1978.
FAO, Revised Legend: Soil Map of the World, World Soil Resources Report
60, Food and Agriculture Organization, Rome, 1990.
FAO, Global and National Soils and Terrain Digital Databases (SOTER),
World Resources Report #74, Rev. 1, Food and Agriculture Organization,
Rome, 1995.
FAO, The Digitized Soil Map of the World Including Derived Soil Properties,
CD-ROM, Food and Agriculture Organization, Rome, 1996.
FAO, The Soil and Terrain Database for northeastern Africa and Crop
Production System Zones of the IGAD subregion, Land and Water Digital Media
Series No. 2, Food and Agriculture Organization, Rome, 1998a.
FAO, Soil and Terrain Database for Latin America and the Carribean
- 1:5 Million scale, Land and Water Digital Media Series No. 5, Food and
Agriculture Organization, Rome, 1998b.
IIASA, FAO, and ISRIC, WISE database, International Institute for Applied
Systems Analysis (IIASA), Food and Agriculture Organization (FAO), and
the International Soil Reference and Information Centre (ISRIC), unpublished,
1998.
ISRIC, SOTER Newsletter, Number 10, International Soils Reference and
Information Centre, Wageningen, March, 1997.
Kern, J.S., Evaluation of soil water retention models based on basic
physical properties, Soil Sci. Soc. of Am. J., 59:1134-1141, 1995.
Nachtergaele, F. O., From the Soil Map of the World to the Global Soil
and Terrain Database, AGLS Working Paper, Food and Agriculture Organization,
Rome, 1996.
Rawls, W.L., D. L. Brakensiek, and K.E. Saxton, Estimation of soil
properties, Tans. ASAE, 25:1316-1320, 1982.
Rawls, W. J., L.R. Ahuja, and D.L. Brakensiek, On estimating the hydraulic
properties of unsaturated soils, In: Proc. Int. Workshop on Indirect Methods
for Estimating the Hydraulic Properties of Unsaturated Soils, M.Th. van
Genuchten (ed.), pp. 329-340, University of California, Riverside, California,
1992.
Reynolds, C. A. and T. J. Jackson, Comparing spatial soil properties
between the state soil geographic database (STATSGO) and the FAO soil map
of the world, Submitted to Soil Sci. Soc. of Am. J., 1999.
Saxton, K.E., W. L. Rawls, J.S. Rosenberger, and R. I. Papendick, Estimating
generalized soil-water characteristics from texture, Soil Sci. Soc.
Am. J., 50:1031-1036, 1986.
Scholes, R. J., D. Skole, and J.S. Ingram, A Global Database of Soil
Properties: A Proposal for Implementation, IGBP-DIS Working Paper Number
10, January, Toulouse, France, 1995.
Sellers, P.J., B.W. Meeson, J. Close, J.Collatz, P.Corprew, D. Dazlich,
F.G. Hall, Y. Kerr, R. Koster, S. Los, K. Mithcell, J. McManus, D. Meyers,
K.-J. Sun, and P.Try, An Overview of the ISLSCP Initiative I Global Data
Sets for Land-Atmosphere Models, 1987-1988, Volumes 1-5, CD-ROM,
NASA, 1995.
Soil Survey Division Staff, Soil Survey Manual. United States Department
of Agriculture, Soil Conservation Service, Handbook No. 18, US Government
Printing Office, Washington D.C., 1993.
Sombroek, W.G., Geographic quantification of soils and changes in their
properties. In: Soils and the Greenhouse Effect, A.F. Bouwman (ed.), John
Wiley and Sons, New York, 1989.
Tietje, O. and M. Tapkenhinrichs, Evaluation of pedo-transfer functions,
Soil Sci. Soc. Am. J., 57:1088-1095, 1993.
UNEP/GRID, GRID Global Soils Database Documentation, In: Global Ecosystems
Database, Version 1.0: Disc A, Dataset Documentation, J.J. Kineman (ed.),
1992, United States Department of Commerce, National Oceanic and Atmospheric
Administration, National Geophysical Data Center, Boulder, Colorado, 1984.
USDA-NRCS, State Soil Geographic (STATSGO) Database, United States
Department of Agriculture (USDA). National Resources Conservation Service
(NRCS). Fort Worth, Texas, 1991.
van Genuchten, M.Th. and F.J. Leij, On estimating the hydraulic properties
of unsaturated soils, In: Proc. Int. Workshop on Indirect Methods for Estimating
the Hydraulic Properties of Unsaturated Soils, M.Th. van Genuchten (ed.),
pp. 1-14, University of California, Riverside, California, 1992.
van Velthuizen, H., L. Verelst, and P. Santacroce, Crop Production
System Zones of the IGADD Sub-Region, Agrometeorology Working Paper Series
No. 10 , Food and Agriculture Organization, Rome, Italy, 1995.
Webb, R.S. and C. E. Rosenzweig, Specifying land surface characteristics
in general circulation models: soil profile data set and derived water-holding
capacities, Global Biogeochemical Cycles, 7:1:97-108, March, 1993.
Webb, R.S., C.E. Rosenweig, and E.R. Levine, A global data set of soil
particle size properties, NASA Technical Memorandum #4286, September, 1991.
Wilson, M.F. and A. Hendersen-Sellers, A global archive of land cover
and soils data for use in general circulation models. J. Clim., 5:119-143,
1985.
Zobler, L., A world soil file for global climate modeling, NASA Technical
Memorandum #87802, Washington, D.C., 1986.
Wösten, J.H.M., P.A. Finke, and M.J.W. Jansen, Comparison of class
and continuous pedotransfer functions to generate hydraulic properties,
Geoderma, 66:227-237, 1995.
Wösten, J.H.M., and M.Th. van Genuchten, Using Texture and other
soil properties to predict the unsaturated soil hydraulic functions,
Soil Sci Soc. Am J., 52:1762-1770, 1988.