融合遥感和土壤双层水平衡模型的农业干旱监测方法研究

Drought monitoring with remote sensing data can reflect vegetative information with high spatial resolution. However, it is subject to the inversion model, the remote sensing data processing and quality. Meteorological drought monitoring can accurately reflect the drought condition based on observation data and physical models. Climatic data usually have very low spatial resolution (larger than 0.5°) and lack of vegetative information in drought. Our research focuses on producing calibrated 1km Palmer drought indices with input of multi-source remote sensing data in Huang huaihai Plain. The improved calibrating methodology is proposed by considering the calibration process at pixel level. The new drought index called Comprehensive Agriculture Drought Index(CADI) is proposed by blending commonly used remote sensing drought indices with calibrated Palmer drought indices based on multivariate statistics. The CADI also incorporates the crop phenology information so as to make it more applicable for agricultural drought monitoring.

遥感干旱监测具有较高的空间分辨率并能反映植被信息，但较依赖于遥感反演模型并受制于成像数据的处理和质量；气象干旱监测可以基于观测数据和物理模型来准确反映台站处的干旱情况，但却往往空间分辨率较粗（一般大于0.5°），而且缺乏植被响应的信息。本研究尝试融合遥感和Palmer双层水平衡模型，探索基于不同的空间域滤波器研究参数在不同空间尺度的变异规律，确定研究区最优的模型参数修正方案。获得黄淮海地区1km分辨率的修正Palmer干旱指数。研究还将结合常用的遥感干旱指数和Palmer干旱指数用线性、非线性和主成分分析等多元统计模型在考虑作物季相信息的基础上定义新的农业干旱指数，以实现黄淮海农业区有效的干旱监测。

传统的干旱监测方法有遥感监测方法和气象监测方法。遥感干旱监测可以反映作物随干旱变化的信息，却受制于遥感原始数据的质量和反演模型的准确度，而气象干旱监测往往基于水平衡模型来定义，可以直接反映田间水分亏缺信息，但水平衡模型很少考虑地表植被信息或者假定植被均质，因而缺少作物对干旱反馈的信息. 本研究尝试融合遥感和气象模型获取了多元干旱指数，研究中用Aquacrop模型模拟了水分对冬小麦长势的胁迫，用地理加权回归分析了不同地区夏玉米产量对遥感干旱指数的响应。研究基于作物的不同生育期，利用线性和非线性统计模型获取了综合农业干旱指数。该研究对于理解全球变化背景下干旱对农业的胁迫，减灾增产、农业管理等意义重大。