Wading through a swamp of complete confusion: how to choose a method for estimating soil water retention parameters for crop models

A minimum input for water dynamics simulation in crop models are soil water content at field capacity (DUL, drained upper limit), wilting point (LL, lower limit) and, often, saturation (SAT). Eight methods for estimating these water-retention parameters were compared using the following procedure: [1] Stepping through the texture triangle in increments of 1% clay and 1% sand, LL, DUL and SAT were calculated for all possible texture combinations (from 0% sand to 100% sand, giving >5000 cases, though not all could be used for all methods); results were grouped by soil type in the USDA classification system. [2] The estimated LL and DUL were compared with field-measured data from across the USA. [3] (Imaginary) soils with a homogenous profile of each of these texture combinations were defined and the DSSAT crop model was run with 11 years of weather data to estimate soybean yield. The discrepancy between estimation methods for water-retention parameters was so big that it is hard to make recommendations on which method to use for which soil. Yet, an analysis with a set of field-measured data showed that the Saxton method performed best for LS, SL, L and SIL soils, with a RMSE <0.018. Using these data as input to the CROPGRO-Soybean model (which is part of DSSAT) showed a worrisome variability among methods in simulated crop yield. The dataset of both field-measured and lab-measured values of LL and DUL showed very different estimates, shedding doubt on the value of lab-measured waterretention data for parameterizing a crop model. Several methods showed inaccuracies in their equation structure.