用分布式水循环模型与机器学习预测内蒙古河套灌区节水潜力

为探究内蒙古河套灌区真实节水潜力，该研究构建河套灌区分布式水循环模型与基于机器学习的盐分模型，设置节水方案集，定量分析各方案下的灌区引、耗水量、地下水埋深、积盐量变化等。结果表明：1）水面蒸发的纳什系数均不低于0.654，相对误差绝对值不高于分别为4.82%，相关关系为0.88，排水过程纳什系数均不低于0.600，相对误差绝对值不高于分别为5.11%，相关关系为0.82，地下水埋深的纳什系数均不低于0.628，相对误差绝对值不高于分别为5.12%，相关关系为0.86，满足灌区水循环满足精度要求。本文选择采用土壤盐分模型，得到土壤积盐量与实测值的纳什系数均不低于0.76，满足精度要求。2）渠道砌衬方案S1、田间节水调控方案S2、种植结构调整方案S3的耗水节水量分别为2.93亿、3.02亿和2.54亿m³。S1+S2+S3组合方案灌区耗水节水量最多，为9.11亿m³，S2+S3方案组合次之。3）渠系水利用系数提高，将引起地下水水位下降，不利于排盐，S1方案下地下水埋深大于3 m的面积比例较基准方案增加了7.59%，不利于灌区排盐。田间工程措施使得相应的农田入渗量减少，地下水位下降，有利于灌区脱盐，S2方案下地下水入渗补给量较基准方案减少2.57亿m³，灌区地下水位下降较为明显，S2方案有利于灌区脱盐。S3方案下地下水入渗补给量略微减少，地下水位变化不大，有利于灌区脱盐。不同方案组合，S1+S2、S1+S2+S3方案下对地下水埋深影响较大，尤其是S1+S2+S3方案在灌区西北部、山前、乌拉特前旗、乌梁素海东部的形成连片埋深高值区，影响区域生育期农田作物与林草地植被生长。S1S2方案下不利于灌区脱盐，自然植被生育期平均埋深超过2.5 m的比例较基准方案增加了5.46%。在综合考虑生态环境的约束下，推荐耗水节水量最大的方案S2+S3，即灌区适宜的耗水节水潜力为5.69亿m³。该方案下虽然也会引起地下水位略有下降、进乌梁素海排入水量略微减少，但最有有利于灌区排盐。研究可为引黄灌区节水方案制定与灌溉管理提供技术支撑。

Prediction of water-saving potential in Inner Mongolia Loop using distributed water cycle model and machine learning

This study aims to explore the real water-saving potential in the Hetao Irrigation Area of Inner Mongolia of China. The distributed water cycle and salinity models were constructed using machine learning. A series of water-saving schemes were established to quantitatively analyze the water intake, water consumption, groundwater depth, and salt accumulation. The results were as follows: (1) The distributed water cycle model was used to calibrate and verify some indicators, such as evaporation, runoff, and groundwater depth in the water cycle process. The Nash coefficients of water surface evaporation, drainage processes, and groundwater depth were no less than 0.654, 0.600, and 0.628, respectively, where the absolute relative errors were no higher than 4.82%, 5.11%, and 5.12%, respectively, and the correlation values were 0.88, 0.82, and 0.86, respectively, fully meeting the accuracy requirements. Three algorithms of machine learning were selected to construct the soil salinity model. The Nash coefficients of soil salt accumulation were no less than 0.76, compared with the measurement. (2) Canal and field water-saving measures were optimized to screen the crop structure adjustment. Seven water-saving schemes and combinations were constructed for the main water-saving measures. In Scheme S1, the water utilization coefficient of the canal increased to 0.60, with a water-saving amount of 293 million cubic meters. Scheme S2 was used to implement the field water-saving regulation, particularly with a water-saving amount of 302 million cubic meters. In Scheme S3, the crop structure was adjusted without the reduction in the amount of water that diverted from the Yellow River, where the water-saving amount of 254 million cubic meters was obtained. Among different scheme combinations, Scheme S1+S2+S3 shared the highest water-saving amount of 911 million cubic meters, followed by Schemes S2+S3 and S1+S3 with a water-saving amount of 569 and 557 million cubic meters, respectively. (3) An increase in the canal water utilization coefficient led to a decline in groundwater levels, which was unfavorable for salt drainage. In Scheme S1, the proportion of areas with a groundwater depth greater than 3 meters increased by 7.59%, compared with the baseline scheme, whereas, the proportion with a groundwater depth between 2.5 and 3.0 meters increased by 4.44%, which was not conducive to the salt drainage. The field engineering measures reduced the infiltration rate of farmland, leading to a decline in the groundwater levels, which was beneficial for the salt drainage. In Scheme S2, the groundwater recharge in the irrigation area decreased by 257 million cubic meters, compared with the baseline scheme, indicating a more significant decline in groundwater level. Scheme S2 was favorable for the salt drainage. In Scheme S3, the groundwater recharge slightly decreased with the relatively stable groundwater level, which was also favorable for salt drainage. Among them, the S1+S2 and S1+S2+S3 combination shared a greater impact on the groundwater depth, especially S1+S2+S3 in the northwest part of the irrigation area. A high-value area of continuous buried depth was formed to dominate the growth of crops and vegetation during the growing season. There was a 5.46% increase in the proportion of naturally vegetated areas with an average buried depth exceeding 2.5 meters in Scheme S1+S2, compared with the baseline scheme. Therefore, the scheme S2+S3 was recommended with the largest water-saving amount, where the suitable water-saving potential of the irrigation area was 569 million cubic meters, considering the constraints of the ecological environment. Consequently, this scheme can be the most favorable for salt drainage in the irrigation area.