The influence of input data quality in determining areas suitable for crop growth at the global scale – a comparative analysis of two soil and climate datasets

Abstract

AbstractThe assessment of biophysical crop suitability requires datasets on soil and climate. In this study, we investigated the differences in topsoil properties for the dominant soil mapping units between two global soil datasets. We compared the ISRIC World Soil Information Center’s World Inventory of Soil Emissions Potential 5 by 5 arc min Soil Map of the World (ISRIC‐WISE 5by5 SMW ) with the Harmonized World Soil Database (HWSD) in 0.5 arc min. We also incorporated annual mean temperature and mean precipitation from two global climate datasets that were the WorldClim measurement‐based climate dataset and the Kiel Climate Model (KCM) modelled results of global climate from 1960 to 1990. We then applied a fuzzy logic approach using different combinations and resolutions of the datasets to determine the effects on the extent and distribution of suitable areas for 15 crops. We only used the spatially dominant soil class in the mapping units in the soil databases (resampled to the same resolution of 5 arc min), and we found that the estimates of topsoil properties (0–20 cm in ISRIC‐WISE and 0–30 cm in HWSD) of the seven analysed parameters were up to 40% lower in most of the HWSD than in the ISRIC‐WISE 5by5 SMW. Results from the KCM are 0.1 °C (1%) lower in mean global annual temperature and 20% higher in average global annual precipitation compared with the WorldClim data. The HWSD‐based runs resulted in 10% less crop‐suitable land than the ISRIC‐WISE 5by5 SMW‐based results. The KCM simulations predicted 1% less crop‐suitable land than the WorldClim model. Despite generalizations, our results demonstrate that discrepancies in crop suitable areas are largely due to the differences in the soil databases rather than to climate.