A model for estimating soil moisture deficits under cereal crops in Britain:2. Performance

Abstract

SUMMARYThe new model, described in a previous paper (Francis & Pidgeon, 1982), for estimating soil moisture deficits under cereal crops in Britain, was evaluated against experimental neutron probe data for eight test crops. The test crops covered a wide range within Britain of growing-season weather and soil types for both winter and spring cereals. Comparisons were also made with estimates of three other models, those of Ritchie, Rhoades & Richardson (1976), Jensen, Wright & Pratt (1971) and Smith & Douglas (1975).The new model gives the best fit for six of the eight test crops, with root-mean-square errors (RMSE), taken over the whole season, of less than 10 mm of deficit and having small random bias. RMSE seldom exceeds 10 mm for any section of any season. By contrast, the other three models have whole season RMSE values in the range 10–40 mm. The model of Ritchie et al. (1976) generally overpredicted deficits early in the season and thereafter performed erratically; those of Jensen et al. (1971) and Smith & Douglas (1975) generally underestimated deficits. For two test crops where the new model performed poorly, serious inadequacies in crop growth and/or meteorological input data were identified as the reasons.A sensitivity analysis for the new model emphasized the importance of the soil variables that were incorporated and the need for rudimentary crop growth information to supplement default assumptions.It is concluded that the new model, having average RMSE values similar to expected neutron probe measurement errors, offers significant advantages over other models in estimation of soil moisture deficits. The new model could further provide a basis for extension to other crops or to description of the distribution with depth of soil moisture deficit.