Accuracy of coarse-scale land quality maps as a function of the upscaling procedure used for soil data

A thoroughly investigated study area in Northern Germany was used to determine the accuracy of coarse-scale land quality maps (1:200,000 and 1:1,000,000), and to evaluate possible improvements by applying different upscaling procedures on fine-scale soil data (1:5000). The land quality chosen for this investigation was the mean annual groundwater recharge rate, obtained by a deterministic simulation model of the soil water balance. Three versions of soil maps were compared and evaluated: the existing soil maps versus new soil maps, created by selecting region-specific soil profiles without changing the boundaries of the existing map units or created by direct aggregation of map units. In the study area, characterized mainly by glacial deposits, the taxonomic criterion of the upscaling procedure is irrelevant for the accuracy of land quality maps at the 1:1,000,000 scale. The results can be used to divide the total error into two parts: at the 1:200,000 and 1:1,000,000 scale, about 75% of the error is caused by soil spatial variability and cannot be avoided at these scales; the remaining 25% indicate the degree of uncertainty that is caused by the mapping or upscaling procedure and can be eliminated to produce “best-possible maps”.