基于作物冠层红外温度的多尺度干旱监测与灌溉决策

Precision irrigation management of the large farm and contiguous agriculture field would be trend and urgent demand of social development in China, under the background and situation of water shortage increasing and labor cost gradually rising. Research of drought monitoring and irrigation decision-making in real-time is very hot issue in application theory of hydrology & water resources and irrigation water management, utilizing the data of the crop canopy infrared temperature, surface temperature, soil water moisture and field meteorology at different scale in irrigation district. . The project will be carried out in a typical irrigation district in North China, where the main agricultural crop includes maize, sunflower and wheat, and so on. The field experiments and data collection would be finished at 3 scales, which involve irrigation experiments in plots near an irrigation ditch, scanning by Unmanned Aerial Vehicles around the branch canal and inversing data from the Satellite Remote Sensing. Through the data assimilation and fusion, the daily data in high resolution could be obtained, to build the areal drought monitoring model based on the crop canopy infrared temperature. The calibration and validation of the Model could be implemented by the real-time ground supervision and investigation. The field data collection pattern would be set up by the Geostatistics method, and its verification and test through the simulation of the Model. Then, the economic and least amount and layout scheme of the field monitoring should be gotten with the data optimizing. The scale effect and up-scaling method of the irrigation decision-making would be discussed and gained. Meanwhile, the second Model of areal drought monitoring and irrigation decision-making in real-time could be built. It would be run up to test the applicability.. The research results could make theoretical basis and technology support for assessing agricultural field water consumption, limited water resources deploying reasonably and precision irrigation management in irrigation district.

在水问题日益严重、劳动力成本逐步上升、大型农场和连片农田精量灌溉管理成为社会发展的趋势和要求的背景下，利用不同尺度下作物冠层红外温度、农田表面温度、土壤墒情、气象要素等数据进行区域干旱实时监测和灌溉决策研究，在水文水资源应用基础理论和灌溉用水管理实践上均具有十分重要的意义。以我国北方灌区主要农作物为研究对象开展田间试验和区域数据调研，通过典型农渠试验点、支渠中尺度数据采集和灌域尺度的遥感热红外温度反演和数据同化与融合，构建基于作物冠层温度的区域旱情监测模型，并利用地面监测和调查数据进行模型率定和校核；与基于地质统计学的田间数据采集模式进行校核和验证，确定不同尺度下监测点的合理数目和布设方案；分析作物适宜灌溉决策指标的尺度扩展和应用方法，形成区域作物旱情实时监测及灌溉综合决策模型。研究成果可为正确评估农田耗水、合理调配有限水资源、灌区精量高效的灌溉决策与管理提供理论依据和技术支撑。

项目主要关注于基于作物冠层红外温度的多尺度干旱监测与灌溉决策研究。经过4年执行期的研究，取得了以下几方面主要研究成果：.（1）通过地面田间数据的连续采集和卫星过境遥感图片的地面温度反演对比分析，可知：1）在下垫面植被均匀、土壤水分空间变异性较小的区域，卫星图片遥感反演地表温度与地面作物冠层温度监测吻合较高，地面监测点数据可以代表临近5个像元（90m×90m）。2）不同计算地表比辐射率来反演地面温度的方法适用于不同作物类型，玉米、春小麦用简单的Sobrino法为宜，葵花地利用覃志豪方法较好，冬小麦-夏玉米连作区差别不大。.（2）通过数据融合在灌区蒸散发空间降尺度的验证及应用可知，1）不同作物融合蒸散发与水量平衡蒸散发变化过程较吻合，在区域农田耗水总量验证中，融合蒸散发与水量平衡蒸散发相一致，两者决定系数达到了0.635。2）融合结果与Landsat 蒸散发在空间纹理信息和空间差异性上一致，融合结果良好。3）ESTARFM融合算法在农田耗水空间降尺度得到较好的应用，可有效区分不同作物耗水之间的差异。.（3）S-I模型特征参数在玉米和向日葵田间进行了率定和验证，可知：1）利用S-I简化模型对玉米和向日葵田间进行作物ETd的估算，在13:00时结果最好。2）模型特征参数a、b值13:00时玉米a、b值皆为负值，而向日葵a为正数、b为负数；玉米和向日葵田块叶面积指数LAI对b值大小的影响，呈相反的趋势，而风速的影响则为一致。3）模型参数a、b可在河套灌区直接应用，在其他地区可作为参考。.（4）根据融合算法确定的区域地表温度，估算灌区玉米和向日葵的作物ET，并与水量平衡方程结果进行了对比，实现从农田精量观测到灌区作物需耗水估算。可知：1）通过大气校正法反演Landsat7&8地表温度，结合多时相MODIS地表温度数据，利用ESTARFM融合算法实现了地表温度的空间降尺度，构建了高时空分辨率地表温度数据集。2）2015年和2016年黄济分灌域、清惠分灌域（考虑分灌域内井灌区的灌水量）、乌拉河分灌域和杨家河分灌域水量平衡蒸散量与模型蒸散量的相对误差（RE）均小于±10%，说明S-I模型估算解放闸灌域区域作物蒸散量具有较好的精度。