基于地表水-地下水耦合模拟的黑河流域水资源配置研究

WATER RESOURCES REGULATION IN THE HEIHE RIVER BASIN BY USING AN INTEGRATED SURFACE WATER-GROUND WATER MODELING APPROACH

吴亚杰

11849077

硕士

环境工程领域工程

田勇

2020-05-25

2020-07-20

哈尔滨工业大学

深圳

黑河流域位于中国西北地区，是中国干旱区内陆河流域的典型代表之一，流域中游河西走廊灌溉农业区是我国重要商品粮食生产种植基地，同时它也成为黑河流域水资源的主要消耗区。黑河流域水资源总量十分紧缺，由于水资源开发利用程度过高而产生一系列生态环境问题。如何合理分配水资源使得经济效益最大成为水资源管理者关注的问题。近年来气候变化引起了大气变化和水循环时空变化，区域性极端干旱或洪水灾害事件可能更加频繁，这也使未来水资源管理面临更大挑战。本研究针对上述两方面问题，发展了基于地表水-地下水耦合模拟的黑河中下游水资源优化配置模型，构建了基于长短期记忆神经网络（LSTM）的黑河上游出山径流预测模型，并将两者相结合预测了黑河中下游水循环与水资源配置对未来气候变化的响应。本研究发展的水资源优化配置模型以农业用水经济效益最高为目标，通过在生态水文模型HEIFLOW框架内耦合高效的线性规划问题，使用单纯形算法求解该问题。利用优化配置模型开展了现状条件下2000年-2016年黑河中下游水资源配置模拟。优化结果表明，使用优化配置模型后，黑河中游灌区平均供水率得到提高，位于河道中下段的灌区获得更多的水资源，灌区间供水更加均衡。本研究提出了联合应用LSTM和主成分分析（PCA）方法来构建日径流预测模型，该方法仅使用气象驱动数据作为输入变量，可实现对日径流过程的长时间序列预测，通过使用PCA降维技术来简化神经网络结构，从而解决神经网络输入数据维度过高的问题。利用该方法对未来黑河上游六大子流域出山径流进行了长时间预测，预测结果显示2017年-2060年出山径流量以-1.0×106m3/yr的趋势缓慢下降。将LSTM模型预测结果作为水资源优化配置模型的黑河上游流量输入边界，对未来黑河中下游进行水资源优化配置模拟实验，其结果显示：未来黑河中下游的年均蒸散发以-0.07mm/yr的趋势减少，但中游灌区蒸散发有所增加；未来正义峡年径流量呈现递减趋势，递减率为-5.5×105m3/yr；未来靠近黑河上游的灌区将获得更多的水资源。

Heihe River Basin (HRB) is one of endorheic river basins in the arid area of China. Its middle-stream part is an important grain production base in China and consumes tons of water resources. There is a series of ecological problems in HRB as a result of overexploitation of water resources. Thus, it is of great importance for local authorities to reasonably manage water resource. Besides, climate change recently has led to such regional extreme catastrophes as droughts and floods, making future water resource management harder. To tackle these two challenges, this study developed optimal allocation model of water resource with conjunctive use of groundwater and surface water. The optimization objective is to maximize economic benefit of irrigation zone at middle HRB. The optimization model was solved by simplex algorithm. The simulation of water resources regulation in middle and lower HRB (ML-HRB) from 2000 to 2016 was carried out by the optimal allocation model. By making use of the optimization model, water supply rates in irrigation zones go rise. Additionally, this study combined Principal Components Analysis (PCA) and Long Short-term Memory (LSTM) to predict daily mountainous runoff in the upper HRB. PCA simplifies the structure of LSTM and improves the simulation performance. The LSTM model enables to predict long-term streamflow merely using only meteorological driving forces. According to the prediction results, the mountainous runoff from 2017-2060 drops by a rate of 1.0×106 m3/yr. Finally, optimal allocation model was used to research water cycle in middle and lower HRB under future climate change scenario. The simulation result shows that future evapotranspiration (ET) in ML-HRB will decrease by -0.07mm/yr, while future ET in irrigation zones will rise. Future streamflow through Zhengyixia will decline by 5.5×105m3/yr. The irrigation zones closer to upper HRB will get more water resource.

黑河流域 水资源优化配置 长短期记忆神经网络 气候变化 地表水地下水耦合模拟

Heihe River Basin optimal allocation of water resources lstm climate change coupling simulation of surface water and groundwater

中文

联合培养

南方科技大学

吴亚杰. 基于地表水-地下水耦合模拟的黑河流域水资源配置研究[D]. 深圳. 哈尔滨工业大学,2020.