SWAMP: A soil layer water supply model for simulating macroscopic crop water uptake under osmotic stress

•Model was calibrated to simulate salt accumulation in irrigated water table soils.•Under these conditions a reduction in water uptake (peas, maize) was simulated well.•The aggregated accuracy, correlation and pattern performance of the model was >75%.•No macro-pattern was observed, hence water uptake residuals contain no structure.•Model did not rely on salinity threshold and slope parameters during simulations. Models like SWAP, HYDRUS and SALTMED compute crop water uptake under osmotic stress with a dimensionless piecewise linear or S-shaped reduction function. Parameters for these functions, to reduce water uptake, corresponds normally to the Maas and Hoffman salinity threshold and slope values. Unfortunately, extensive crop- and site-specific calibration of the parameters is required. This is because these values, amongst other reasons, serve only as guidelines and express salt tolerance at a time and root-zone average soil salinity and not local total potential heads. In this paper an alternative model (Soil WAter Management Program, SWAMP), that does not rely on these parameters and functions were presented and evaluated. The algorithm used by SWAMP to simulate the water supply of a rooted soil layer and hence water uptake, under decreasing matric potentials was adapted to include the effect of decreasing osmotic potentials. Data from a lysimeter trial was used to evaluate SWAMP. The model was calibrated to represent the soil conditions of the trial, i.e. peas and maize were irrigated with EC's between 20 and 600mSm−1 and grown in sand to sandy loam soils with water tables of the same quality. Under these osmotic stress conditions, SWAMP was able to simulate weekly water uptake of both crops grown on both soils well, i.e. the aggregated accuracy, correlation and pattern performance (ISWAMP) were above 75%. No macro-pattern was observed. Thus, the water uptake residuals contain no structure that is not accounted for in the algorithm and parameters. No extensive calibration was necessary because the parameters for the algorithm were calculated from easily measured inputs. From the three most sensitive parameters, only the critical leaf water potential of a crop might be difficult to obtain. SWAMP contains default values for a number of crops. A model is, therefore, presented that simulate the change in osmotic stress with changing soil water content and that does not rely on the salinity threshold and slope parameters.