基于遗传算法的二维土壤水与作物生长耦合模拟模型构建和参数优化

Developing and Parameter Optimization of Two-dimensional Soil Water Transport and Crop Growth Coupling Model Based on Genetic Algorithm

摘要:

为快速准确地估算滴灌条件下土壤-作物系统模型参数，基于二维土壤水与作物生长模拟模型（SWNCM-2D）耦合遗传算法（GA），建立了滴灌条件下土壤水力学参数与作物生长参数的优化模型，以土壤含水率和作物干物质量实测值与模拟值之间的标准均方根误差最小为优化目标，利用南疆地区棉花滴灌试验不同灌水量处理下的土壤含水率和作物生长动态及产量观测数据，优化求解土壤水力学参数与作物生长参数，并应用优化后的模型参数开展不同滴灌灌溉管理措施下的棉花产量与水分生产力预测。结果表明：耦合GA的SWNCM-2D模型参数优化结果较好，不同土层土壤含水率模拟值与实测值之间均方根误差（RMSE）、标准均方根误差（nRMSE）和一致性指数（d）分别为0.0095~0.0370cm3/cm3、5%~27%和0.6518~0.9642，干物质累积量和LAI的nRMSE分别为8%~17%和6.2%~23.0%，d均高于0.97。棉花皮棉产量随灌水量增大而增大，水分生产力随灌水量增大而减小；皮棉产量随灌水间隔增大而减小，水分生产力随灌水间隔增大先增大后减小；说明基于优化参数的全生育期土壤水分动态变化与作物生长过程的模拟较为准确。综合考虑棉花产量和水分生产力，推荐该地区适宜的灌溉制度为灌水间隔7d和灌水量120% ETc（作物需水量）组合。

Abstract:

To efficiently and accurately estimate the model parameters of the soilcrop system under drip irrigation conditions, a genetic algorithm (GA) was integrated with the twodimensional soil water and crop growth simulation model (SWNCM-2D) to establish an optimization model for soil hydraulic parameters and crop growth parameters under drip irrigation conditions. The objective was to minimize the standard root-mean-square error (RMSE) between the measured and simulated soil water contents as well as crop dry matter quality. By utilizing observed data on soil water content, crop growth dynamics, and yield under drip irrigation treatment in southern Xinjiang, the soil hydraulic parameters and crop growth parameters were optimized by using the SWNCM-2D model coupled with GA. These optimized model parameters were then utilized to predict cotton yield and water productivity under various drip irrigation management scenarios. The results demonstrated that parameter optimization using the SWNCM-2D model coupled with GA yielded favorable outcomes. The RMSE, nRMSE and d values between simulated and measured soil water contents in different layers ranged from 0.0095cm3/cm3 to 0.0370cm3/cm3, 5% to 27%, and 0.6518 to 0.9642 respectively;while nRMSE for dry matter accumulation and LAI were within the range of 8%~17% and 6.2%~23.0%, d all exceeded 0.97 threshold value.Cotton lint yield was increased with the increase of irrigation levels while water productivity was decreased accordingly;lint yield was decreased as irrigation intervals lengthened whereas water productivity was initially increased before declining with longer intervals.In conclusion, simulations based on optimized parameters provided accurate representation of dynamic changes in soil moisture throughout the entire growth period along with precise depiction of crop development processes.Considering cotton yield alongside water productivity, the recommended irrigation regime for this region was a watering interval of 7 days at a rate of 120% ETc.