A SURVEY OF SEVERAL AGROCLIMATIC SOIL-WATER BALANCE MODELS WITH REFERENCE TO THEIR SPATIAL APPLICATION

This survey evaluates the applicability of the principal categories of soil water models for studying the variablility of crop water balance at regional scales. For each model attention is given to the definiti on of the sphere of validity, the underlying hypotheses and the data r equirements. For most applications of such models at the regional scal e, model simplification increases with the extent of the application a rea. This review suggests that this option is not theoretically justif ied. Simplified approaches - either analogue or empirical - with few d ata requirements, have limited validity and, because their physical si gnificance is low, have state parameters with poorly-defined means of measurement or cumbersome calibration procedures. Mechanistic approach es need more data, but allow for a larger range of boundary conditions . Thus, they might be better adapted to spatial applications, where th e variation in local conditions is often large. The problem is, theref ore, whether to use a physically-based model, where parameterisation i s approximated, or a more easily parameterised, simplified model which is often working beyond its limits. To answer this question more rese arch is necessary on the quantification of the sources of error of mod el predictions.