融合多要素及其时滞效应的农业干旱综合监测方法研究

Agricultural drought monitoring is an important way to minimizing the loss of agricultural drought. However, many problems such as models with univariate, the factors and their feedbacks are unclear, and linear assumption of model factors still exist in current agricultural drought monitoring models, which making it difficult to accurately reflect agricultural drought information. This project is aimed at modelling and application of integrated agricultural drought monitoring model and take the Guanzhong Plain as the typical case area. First, we explore the mechanism of agricultural drought, and determine the key indicators of agricultural drought information and their feedbacks between these indicators. Then, an integrated agricultural drought monitoring model considering multi factors is established based on multi-source remote sensing and BP neural network. We also validated the results of model inversion from temporal and spatial dimension by station scale soil moisture data and typical drought episode. Next, we clarify the spatiotemporal variation of agricultural drought for whole growing season of winter wheat and summer maize over the Guanzhong Plain based on optimized integrated agricultural drought monitoring model. Finally, we quantitatively assess of agricultural drought risk variables through run length theory and joint probability distribution function method. Through this project, we hope to provide theoretical and method support for target disaster prevention and mitigation and improve the ability to adapt changing climate, and achieve the sustainable development of agriculture.

农业干旱监测是实现最大限度减轻农业旱灾损失的重要途径，然而目前农业干旱监测模型普遍存在模型要素单一、模型要素间反馈机理不清以及模型要素线性假设等不足，难以准确反映农业干旱信息。本项目拟以农业干旱综合监测模型构建及应用为研究目标，以关中平原为典型案例区，从农业干旱致灾机理探究出发，厘定农业干旱信息关键指示要素，明晰多要素间的滞后反馈关系，进而借助多源数据和神经网络方法，构建综合考虑多要素及其时滞效应的农业干旱监测模型，并采用站点尺度土壤相对湿度数据和典型干旱事件对模型反演结果进行时间和空间维度的验证。在此基础上，应用精度优化后的农业干旱综合监测模型，阐明关中平原冬小麦全生育期内的农业干旱时空演变规律，并通过游程理论和联合概率分布函数，揭示关中平原农业干旱超越概率和重现期等农业干旱风险变量特征，以期为国家精准防灾减灾、提高气候变化适应能力和实现农业可持续发展提供理论与方法支持。

农业干旱监测是实现最大限度减轻农业旱灾损失的重要途径，然而目前农业干旱监测模型普遍存在模型要素单一、模型要素间反馈机理不清以及模型要素线性假设等不足，难以准确反映农业干旱信息。本项目基于多源遥感数据和神经网络模型构建了综合考虑土壤-植被-大气连续体的综合干旱监测指数，并开展模型验证和应用工作，揭示了重点区域干旱多尺度特征；同时构建了基于陆地水储量的干旱监测指数，有效表征了陆地水储量亏缺的整体特征，并借助陆地水文模型分离了不同水资源组分，阐明了陆地水资源亏缺的贡献率及其影响因素。在植被-水分关系变化研究中，揭示了植被覆盖变化和植被水分利用效率变化规律及其主导因素，最后总结了干旱对作物产量影响的最新进展和主要挑战，探索了水-粮食-能源纽带关系的研究现状及未来趋势。本研究构建的综合干旱监测指数与实测土壤湿度具有较好的相关性，并通过典型干旱事件演变过程进行实证，表明该指数可以有效指示华北平原2014年夏季干旱时空变化过程。基于GRACE数据的干旱指数表明去趋势的结果能够更好地反映干旱的发生和发展过程，并发现黄河流域、淮河流域、海河流域和西南诸河流域等流域均在2013年之后呈现干旱加剧的趋势，主要表现为干旱历时和干旱强度的增加。欧亚大陆的陆地水储量整体呈现显著下降趋势，且检测出的五个下降趋势变化特点区域的降速介于-8.94 ~ -21.79 mm/yr之间。上述变化热点区域的陆地水储量下降主要归因于地下水的过度开采以及气候变暖导致的地表水的大量损失。影响因素分析表明，ENSO、AO和NAO是导致陆地水储量变化最重要的3个气候驱动因子。本项目综合利用野外观测数据、多源遥感数据和模型模拟数据，系统开展了基于综合干旱监测指数和像元尺度阈值的流域干旱评估，全面分析了水储量多时空尺度变化特征及与遥相关指数的关联，定量分析了干旱时空演变规律及其对植被覆盖和水分利用效率的影响，研究成果将有助于提升学界对综合干旱监测指数的认识，促进干旱影响及其适应策略的发展。

内容获取失败，请点击重试