Soil erosion science: Reflections on the limitations of current approaches ☆

Abstract To date erosion scientists have failed to address — or have addressed inadequately — some of the ‘big questions’ of our discipline. For example, where is erosion occurring? Why is it happening, and who is to blame? How serious is it? Who does it affect? What should be the response? Can we prevent it? What are the costs of erosion? Our inability or reluctance to answer such questions damages our credibility and is based on weaknesses in commonly-used approaches and the spatial and temporal scales at which much research is carried out. We have difficulty in the recognition, description and quantification of erosion, and limited information on the magnitude and frequency of events that cause erosion. In particular there has been a neglect of extreme events which are known to contribute substantially to total erosion. The inadequacy and frequent misuse of existing data leaves us open to the charge of exaggeration of the erosion problem (a la Lomborg). Models need to be developed for many purposes and at many scales. Existing models have proved to be of limited value, in the real as opposed to the academic world, both because of problems with the reliability of their results, and difficulties (with associated costs) of acquiring suitable data. However, there are some positive signs: models are now being developed for purposes including addressing questions of off-site impacts and land-use policy. Cheap, reliable and technically simple methods of erosion assessment at the field scale are needed. At the global scale, an up-date of GLASOD based on a scientific approach is urgent so that we are at least able to identify erosion ‘hotspots’. In terms of explanation of erosion, the greatest need is for a full recognition of the importance of socio-economic drivers. The accession of new countries to the EU with different economic and land-use histories emphasises this need. Too often we have left people, especially the farmers, out of the picture. Our approach could be characterised as ‘data-rich and people-poor’.