Performance evaluation of AquaCrop and DSSAT-SUBSTOR-Potato models in simulating potato growth, yield and water productivity under various drip fertigation regimes

Water and fertilizer are two important factors affecting crop growth and yield. Crop growth models are powerful tools for making irrigation and fertilization scheduling and predicting crop yield. The performance of crop models on potato under different irrigation amounts along with different N, P and K rates has been rarely evaluated, especially under drip fertigation. In this study, the accuracy of AquaCrop and DSSAT-SUBSTOR-Potato models in simulating potato growth, yield and water productivity under different drip fertigation regimes was compared. For model calibration and validation, a two-year field experiment with three irrigation levels (W1, 60% ETC; W2, 80% ETC and W3, 100% ETC, where ETC was the crop evapotranspiration) and four fertigation (N-P2O5-K2O) rates (F1: 100–40–150 kg ha−1, F2: 150–60–225 kg ha−1, F3: 200–80–300 kg ha−1 and F4: 250–100–375 kg ha−1) was carried out in a sandy region of northwest China in 2018 and 2019. The statistical indicators showed that the AquaCrop model gave satisfactory predictions of canopy cover (R2 ≥ 0.92, NRMSE ≤ 14.7%, d ≥ 0.96 and RMSE ≤ 10.8%), total dry matter (R2 ≥ 0.92, NRMSE ≤ 25.0%, d ≥ 0.94 and RMSE ≤ 2.2 t ha−1), final dry tuber yield (R2 ≥ 0.57, NRMSE ≤ 12.21%, d ≥ 0.82 and RMSE ≤ 0.83 t ha−1) and fresh tuber yield (R2 ≥ 0.53, NRMSE ≤ 12.02%, d ≥ 0.81 and RMSE ≤ 4.91 t ha−1) in the four fertigation treatments (F1-F4) under W3 and W2 during the two potato growing seasons. The final total dry matter and tuber yield under W1 were over-estimated. The simulated water productivity and soil water content also agreed well with the observations (R2 ≥ 0.91, NRMSE ≤ 5.51%, d ≥ 0.94 and RMSE ≤ 0.48 kg m−3, and R2 ≥ 0.40, NRMSE ≤ 14.8%, d ≥ 0.77 and RMSE ≤ 11.5 mm). The simulation accuracy of DSSAT-SUBSTOR-Potato model was lower than that of AquaCrop model, which only showed a goodness of fit between simulated total dry matter, soil water content, tuber yield, water productivity and observations under high irrigation amount with high fertilization rate (W3F4). Since the observed tuber yield increased with the increase of irrigation level (60–100% ETC) under the same fertigation rate, 120%ETC and 40%ETC were further simulated using the AquaCrop model to test the changes in tuber yield with further increased or decreased irrigation levels. The scenarios simulation showed that the tuber yield was only increased by 0.19%−1.52% under 120%ETC compared with that under 100% ETC, but 18.0% more irrigation water was consumed. Comprehensively considering water resources and tuber yield, the irrigation level of 100% ETC (W3) along with fertigation (N-P2O5-K2O) rate of 200–80–300 kg ha−1 (F3) was recommended for the sustainable potato production in the study region. The results of this study can provide guidance for the application of crop models in drip-fertigated potato.