A simple regional downscaling approach for spatially distributing land use types for agricultural land

Abstract

Spatial explicit information of agricultural land use is a crucial requirement for many environmental models. In particular, for regional agro-ecosystem models that aim at agricultural management such as nutrient, soil and pest management for farming systems geo-referenced land use data are essential. In order to derive such spatial explicit agricultural land use data for Swiss agro-ecosystems, we developed a simple downscaling approach based on available geo-referenced farm census data and Swiss land use statistics. Allocation of farm census data to the raster cell of the Swiss land use statistics is achieved by using simple rules that take into account terrain attributes and the distance of agricultural land use types to the farm. The simple algorithm allocates home pastures, arable land and grassland to the farm census data. We tested the downscaling approach for the Canton Fribourg in western Switzerland for seven farm census periods (1980–2007). The agricultural land of about 770km2 was managed by about 5900farms in 1980 and by about 3400 farms in 2007. The modelled arable land agreed for 72% with the most recent land use statistics, while for grassland and home pasture only 41% and 37% of the grid cells were consistent. We presume that this discrepancy is partly caused by misclassification using aerial photographs in the land use statistics. Validation of the modelled agricultural land use types with recorded land use of soil monitoring sites revealed an agreement of 80% for the grassland sites and 59% for the arable sites. Hence, with the simple downscaling approach we already achieved fairly estimates of the spatial pattern of agricultural land use in the case study region. The possible application of such a downscaling approach is illustrated for the phosphorus inputs through animal manure into the agricultural soils of the Canton Fribourg.