根系水分胁迫响应函数对土壤水及作物生长动态和产量模拟影响的研究

为探究根系水分胁迫响应函数对农田水分动态及产量模拟的影响,基于Richards方程和PS123作物生长模型分别进行了土壤水分动态和小麦产量的模拟,对比分析了VG（S型曲线）、MP（凹凸型曲线）及LS（S型曲线）3种水分胁迫响应函数。采用山西省霍泉站（3 a）及潇河站（2 a）的试验资料对模型中的土壤水力特征参数、水分胁迫响应函数参数进行反演,确定最优的参数值,得到了土壤含水率、蒸散量及籽粒产量的模拟结果。结果表明：（1） VG、MP及LS函数条件下的土壤水分及产量模拟结果均较好,回归估计标准误差（RMSE）值在0.021~0.036之间,模拟值与实测值间的R检验达到了极显著水平;对于霍泉站,VG、MP、LS函数条件下,土壤含水率模拟值与实测值的平均相对误差分别为6.37%、8.26%及7.18%,相关性系数值最小为0.7814,模拟产量的平均相对误差值分别为8.73%、8.40%及8.42%;对于潇河站,土壤含水率模拟值与实测值的平均相对误差结果普遍大于霍泉站,最大为12.47%。（2） 土壤水分动态模拟,S型曲线的水分胁迫响应函数较凹凸型曲线的水分胁迫响应函数表现出了更好的效果,其中使用VG函数模拟的平均相对误差较小,模拟的蒸散量更接近于实测值。（3） 产量模拟,3种水分胁迫响应函数差异不明显。综上,VG函数是一种精度较高的根系水分胁迫响应函数,且模型简洁、方便;改进的LS函数能够提高模拟精度,但模型的稳定性有待近一步研究。

Effect of root water stress response function on soil water,crop growth dynamics and yield simulation

To investigate the effect of the root moisture stress response function (RMSRF) on soil moisture dynamic and yield simulation, the soil moisture dynamic, and wheat yield were simulated based on the Richards equation and PS123 crop growth model. The three RMSRF of VG (S-curve), MP (convex curve), and LS (S-curve) were compared. The soil characteristic and RMSRF parameters in the model were inverted to determine the optimal parameter values using the measured data from Huoquan Station (three years) and Xiaohe Station (two years) in Shanxi Province, China to obtain the simulation results of soil moisture content, evapotranspiration, and grain yield. The analysis is as follows. (1) The soil moisture and yield simulations under VG, MP, and LS functions are good. The RMSE value is between 0.021 and 0.036; the r test between the simulated and measured values has reached an extremely significant level. For the Huoquan Station, under the three RMSRF of VG, MP, and LS conditions, the average relative errors of the simulated and measured values for soil moisture content are 6.37%, 8.26%, and 7.18%, respectively. The minimum certainty coefficient value is 0.7814. The average relative error values of the simulated wheat yield are 8.73%, 8.40%, and 8.42%, respectively. For the Xiaohe Station, the average relative error of the simulated and measured values for soil moisture content is greater than that of the Huoquan Station, with a maximum of 12.47%. (2) The simulation of soil moisture dynamic shows that RMSRF by the S-shaped curve shows better results than RMSRF by the concave-convex curve. The average relative error using the VG function is smaller, and the simulated evapotranspiration is closer to the actual measurement. (3) In terms of yield simulation results, there is no significant difference among the three RMSRF. In summary, the VG function is a high-precision RMSRF, and the model is simple and convenient. The suggested LS function can improve the simulation accuracy, but the stability of the model needs further research.