Estimation of inter-annual winter crop area variation and spatial distribution with low resolution NDVI data by using neural networks trained on high resolution images

The current work aimed at testing a methodology which can be applied to low spatial resolution satellite data to assess inter-annual crop area variations on a regional scale. The methodology is based on the assumption that within mixed pixels such variations are reflected by changes in the related multitemporal Normalised Difference Vegetation Index (NDVI) profiles. This implies that low resolution NDVI images with high temporal frequency can be used to update land cover estimates derived from higher resolution cartography. More particularly, changes in the shape of annual NDVI profiles can be detected by a Neural Network trained by using high resolution images for a subset of the study years. By taking into account the respective proportions of the remaining land covers within a given low resolution pixel, the accuracy of the net can be further increased. The proposed methodology was applied in a study region in central Italy to estimate area changes of winter crops from low resolution NDVI profiles. The accuracy of such estimates was assessed by comparison to official agricultural statistics using a bootstrap approach. The method showed promise for estimating crop area variation on a regional scale and proved to have a significantly higher forecast capability than other methods used previously for the same study area.