盐渍化农田排水沟水体溶存N_(2)O浓度预测

Predicting dissolved N2O concentrations from drainage ditches in salt-affected farmlands

本研究选取宁夏青铜峡灌区的典型盐渍化农田排水沟作为研究对象,基于采集的排水沟上覆水体关键水质参数溶解性有机碳(DOC)、水温(WT)、硝态氮(NO_(3)^(-)-N)和电导率(EC),构建了参数优化后的N_(2)O溶存浓度反向传播(BP)神经网络预测模型,并通过遗传算法(GA)和蚁群算法(ACO)对模型进行了优化,以提高预测精度和稳定性.结果表明,EC的增加显著促进了排水沟上覆水体中氧化亚氮(N_(2)O)的溶存浓度的提升,NO_(3)^(-)-N、EC与N_(2)O溶存浓度之间存在极显著的正相关关系,而WT、DOC则与N_(2)O溶存浓度表现出显著的负相关性.利用排水沟上覆水体N_(2)O溶存浓度实测数据进行验证,证实了所构建模型的有效性和可靠性,其中ACO-BP模型预测值和实测值的相关系数均大于0.70,最佳情况下R²达到了0.79,平均相对误差(MRE)仅为7.26%.

This study selected typical saline-alkali agricultural drainage ditches from the Qingtongxia Irrigation District in Ningxia as the subject of research.Based on key water quality parameters of the water body overlying the drainage ditches,including dissolved organic carbon(DOC),water temperature(WT),nitrate nitrogen(NO_(3)^(-)-N),and electrical conductivity(EC),a backpropagation(BP)neural network predictive model for the dissolved concentration of Nitrous oxide(N_(2)O)with optimized parameters was constructed.The model was further optimized using Genetic Algorithm(GA)and Ant Colony Optimization(ACO)to enhance the prediction accuracy and stability.The results indicate that an increase in EC significantly promotes the dissolved concentration of N_(2)O in the water body overlying the drainage ditches.There is an extremely significant positive correlation between NO_(3)^(-)-N and EC with the dissolved concentration of N_(2)O,while WT and DOC show a significant negative correlation with the dissolved concentration of N_(2)O.The effectiveness and reliability of the constructed model were verified using actual measured data of the dissolved concentration of N_(2)O in the water body overlying the drainage ditches,with the correlation coefficient of the predicted values and actual measured values of the ACO-BP model all exceeding 0.70.Under the best conditions,the coefficient of determination(R²)reached 0.79,and the Mean Relative Error(MRE)was only 7.26%.