Spatial and temporal variation in soil quality are major determinants of the productive capacity of land. HarvestChoice uses a number of data sources and approaches in attempting to capture the multiple dimensions of soil quality and their impact on crop yields and the effectiveness of alternative soil and soil and water management interventions. Central to our analytical approach is the use of crop growth and cropping system simulation models to help address a range of simple to complex questions about the impacts of weather, soil, and farmer production and resource management decisions on productivity and sustainability. For example, how will crop yields respond to climate change in the breadbaskets of East Africa? How effective could small-scale irrigation, cereal-legume rotations, and better residue management strategies be in raising farm productivity across the Guinea Savannah of West Africa?
Sufficiently comprehensive and reliable soil data are difficult to obtain (a major rationale for investments such as the Africa Soil Iinformation System, AfSIS). Recognizing the existing shortcomings in conventional soil mapping data sources, HarvestChoice exploits three approaches to representing soil quality within an individual gridcell; the use of representative top-soil and sub-soil properties of each soil type occuring within a soil mapping unit (HWSD and FCC), detailed measurements of soil properties in soil profiles at a number of specific locations, e.g. soil pits, auger or soil cores (WISE) and, recognizing that local soil properties can vary widely, assuming a complete range of soil quality could potentially exist (HC27)
Soil Attributes Interpreted from Conventional Soil Maps
A HarvestChoice grid cell is assumed to contain only one soil mapping unit, but each soil mapping unit may contain several individual soil types. The HWSD and FCC datasets contain (HWSD) or are interpretations of (FCC) representative top-soil and sub-soil attribute values for each soil type found in a soil mapping unit.
- Harmonized World Soil Database (HWSD) HWSD Version 1.2 is a 30 arc-second raster database containing over 16,000 different soil mapping units. For sub-Saharan Africa, HWSD characterizes more than 7,000 soil mapping units made up from 132 soil types. On average, each soil mapping unit comprises at least 3 different soil types. The HWSD was constructed by harmonizing individual regional- and country-scale soil map databases using a globally consistent method. However, there are limitations in the range of properties reported compared to the data needs of cropping system models such as DSSAT and APSIM. Thus, while HWSD can provide information on key soil variables for descriptive purposes, it is generally inadequate for representing soil profiles with the level of detail demanded by crop models.
- Soil Fertility Capability Classification (FCC) The FCC methodology (Sanchez et al. 2003) delivers the most widely-used set of aggregate qualitative indicators of soil fertility for assessing the global distribution of soil constraints to crop growth. HarvestChoice facilitated developing an updated version of FCC using HWSD, and this new dataset (Palm et al. 2007) will play a role in HarvestChoice's forthcoming assessment of the region-wide distribution of agricultural production constraints in sub-Saharan Africa.
Point soil sampling and soil profile characterization
ISRIC-WISE - Global Soil Profile Data (version 3.1) is a global soil profile database that provides point-basis information of individual soil profiles, including soil chemical and physical characteristics. It provides information on site-specific soil profiles at very detailed levels. Furthermore, HarvestChoice converted 3,404 individual SSA soil profiles from WISE into DSSAT-compatible format. Crop modelers can now use this large collection of soil profiles as a pool from which to choose those that best-match specific soil types identified in soil map databases (such as HWSD). The primary limitation of the WISE database is that the number and distribution of soil sample sites are not fully representative of the footprint of agriculture across SSA.
Spectrum of Generic Soil Profiles
HC27 Generic Soil Profiles To overcome the limitations of HWSD and WISE for crop modeling applications, HarvestChoice developed a pragmatic approach that characterizes a complete range of 27 generic soil profiles based on three criteria to which crop productivity is considered most sensitive: texture, rooting depth, and soil organic carbon content. By classifying three levels for each category and setting their boundary conditions, we generated 27 soil profiles ranging from deep, loamy soils rich in organic matter to shallow, sandy, infertile soils, in formats compatible with DSSAT and APSIM cropping system models. This dataset, called HC27, does not replace actual soil measurements, but provides a means of generating a complete range of potential of crop growth outcomes at a given location, depending on which of the 27 generic soils most closely match actual soil types. Additionally, by mashing-up the mapping unit-based HWSD and site-specific WISE on 5 arc-minute grids using 5 key soil variables (soil type, texture, organic carbon content, pH, and water availability), we created a look-up table to identify the best-matching soil profiles and their area share in each grid cell in sub-Saharan Africa. When no appropriate soil profile was found from WISE, we used the HC27 Generic Soil Profiles to find a suitable proxy.