基于作物模型与遥感数据同化的农田土壤速效养分反演方法研究

Soil nutrition status is the primary factor of farmland productivity, information on soil properties is important for precise field management. Precise fertilization based on field nutritients mapping can increase the economic benefit, reduce the environment pollution and guarantee the sustainable development of agriculture. Yet currently there are several limitations on supporting precise fertilization with remote sensing derived soil nutrition status map: 1) the available nutritiensts information is hard to get due to low content; 2) the simulation result is unstable which is affected by crop type, soil type, environment and time of RS data; 3) usually hyper-spectral image is required, which is more difficult to acquire and process; 4) only surface soil nutrition information can be acquired and 5) the accuracy can easily be depressed by factors as crop coverage and soil water content. Thus taking precise fertilization as serving target, this study will take the crop growing characteristics under soil nutrition stress as breakthrough point, first selecting WOFOST model to simulate crop growth and building two different assimilation methods to connect RS and crop model, then taking yield and LAI as indicators to simulation the influence of soil nutrition stress, and develop a methodology for mapping soil available nutrition status based on lookup table (LUT) and equation set. This study attempts to provide a method for fast, low-cost and guaranteed mapping of soil available nutrition status base on crop model and remote sensing.

土壤养分状况是决定农田潜在生产力的主要因素，土壤性状及肥力状况信息可以为精准农田管理提供相应依据，其关键是准确获取土壤速效养分的空间差异信息。项目以农田精准施肥为目标，针对目前农田土壤养分遥感监测中存在的速效养分信息获取难、反演结果不稳定、对遥感数据要求高、只能获取土壤表层养分状况以及受作物覆盖与水分条件影响大等问题，以土壤养分缺失所引发的作物表现特征为切入点，分析整理不同养分供给缺失情况下作物的生长特征，以叶面积指数和作物单产为指标，以WOFOST作物模型为作物生长模拟工具，以查找表和构建方程组为主要实现手段，通过数据同化构建实现农田土壤速效养分遥感估算方法，实现耕层土壤速效养分的遥感获取，以满足农田精准施肥对土壤速效养分快速、低成本的保障性获取需求。

土壤养分状况是决定农田潜在生产力的主要因素，土壤性状及肥力状况信息可以为精准农田管理提供相应依据，其关键是准确获取土壤速效养分的空间差异信息。项目以农田精准施肥为目标，针对目前农田土壤养分遥感监测中存在的速效养分信息获取难、反演结果不稳定、对遥感数据要求高、只能获取土壤表层养分状况以及受作物覆盖与水分条件影响大等问题，以土壤养分缺失所引发的作物表现特征为切入点，分析整理不同养分供给缺失情况下作物的生长特征，以叶面积指数和作物单产为指标，以WOFOST作物模型为作物生长模拟工具，以查找表和构建方程组为主要实现手段，通过数据同化构建实现农田土壤速效养分遥感估算方法，实现耕层土壤速效养分的遥感获取，以满足农田精准施肥对土壤速效养分快速、低成本的保障性获取需求。研究主要成果包括WOFOST作物模型的参数敏感性分析与模型标定、速效养分反演的可行性分析、适用于养分的遥感数据同化新算法、基于WOFOST模型和时间序列遥感数据同化的速效养分反演新方法、土壤速效养分反演算法的优化等。从科学意义来看，本项目改变以往遥感估算陆表参数时总是从光谱特征出发的思路，并克服统计模型的局限性，尝试对于光谱特征不明显的速效养分，采取遥感技术和作物生长机理模型相结合实现估算的方法，这种方式拓展了遥感应用的思路。