The process-based crop growth simulation model DSSAT is a powerful tool for modern agricultural system research. It can quantitatively describe the crop growth and development and yield formation processes and their relationship with climatic factors, soil environment, variety types and technical measures, and provide solutions for different conditions. Quantitative tools are provided for crop growth and yield prediction, cultivation management, environmental assessment, and future climate change assessment. However, when crop growth models develop from single-point research to regional-scale applications, the non-uniformity of the surface and near-surface environments that arises due to the increase in spatial scale leads to problems in the acquisition of some macroscopic data and the regionalization of parameters in the model. There are many difficulties, and model simulation results will also have great uncertainties. Remote sensing information can help crop growth models overcome these shortcomings to a large extent.
Remote sensing data from domestic satellites (such as HJ, GF, ZY), MODIS, Landsat, Sentinel-2 and other remote sensing data are effective means of monitoring crop growth status in a large area; crop growth models can use environmental factors to simulate the crop growth process and reveal the characteristics of crop growth and development. Cause and essence. Driven by the development of science and technology and the demand for agricultural applications, data assimilation methods combine remote sensing data with crop growth models to monitor crop growth and predict crop yields. This is one of the important contents and development trends of current agricultural information technology application research. The combination of the two can not only provide macroscopic monitoring information, but also dynamically reflect the crop growth and development process, which is conducive to realizing complementary advantages and enhancing application potential.
At present, there is no mature commercial software available for crop yield estimation based on data assimilation method coupling remote sensing and crop models. The "Practical Technical Application of Remote Sensing Data and Crop Model Assimilation" is specially held to help students master the basics of remote sensing and crop model assimilation. Knowledge, the differences and connections with traditional crop remote sensing monitoring methods, using a combination of "theoretical explanation + case practice + hands-on practice + discussion and interaction" to peel off the cocoons and analyze the application of data assimilation methods in crop growth monitoring and yield estimation in simple and easy-to-understand terms The experience and programming skills needed to solve the relevant scientific problems in agricultural production research.
This time it mainly involves the remote sensing data, PROSAIL model, DSSAT model, parameter sensitivity analysis, data assimilation algorithm, model coupling, accuracy verification and other major aspects in the assimilation modeling of remote sensing data and crop models. The setting of the outline mainly focuses on the above-mentioned links to design relevant basic theoretical knowledge and hands-on operation steps. Through the explanation and practical operation of each link, the purpose of this time can be achieved and the set goals can be achieved.
| Topic 1 : Basic theoretical knowledge of remote sensing |
| Remote sensing platforms (such as drones) and sensors, major domestic and foreign land satellites (such as Landsat, SPOT, HJ, GF) Basic principles of remote sensing, spectral response function, remote sensing data processing flow Application of remote sensing in terrestrial ecosystem monitoring |
| Topic 2 : Research progress at home and abroad on crop growth monitoring and yield estimation |
| A review of domestic and foreign research Research case analysis |
| Topic 3 : Fortran Programming Language |
| Software installation (using xp/win7/win8/win10 professional version notebook) Project file creation and basic syntax operations |
| Topic 4 : Basic principles of remote sensing inversion of crop parameters |
| Remote sensing inversion crop parameter types ² Biochemical components (Chlorophyll, nitrogen, dry matter, leaf moisture content, anthocyanins) ² Biophysical parameters (LAI, LAD, plant height, biomass) ² Physiological and ecological parameters （FPAR、ET） Crop parameter remote sensing inversion model ² Empirical model F linear model F-index model F-log model F … ² Physical model F radiative transfer model F geometric optical model F mixed model F computer simulation model ² Comparative analysis of different methods
|
| Topic 5 : PROSAIL Model |
| ² Input parameters: LAI/LAD/chlorophyll/anthocyanin/dry matter/carotenoids/moisture content/… ² Output parameter: Vegetation canopy reflectance
² Taking FORTRAN code as an example to simulate the reflectivity simulation process on a computer ² Simulate leaf reflectivity and transmittance ² Simulate vegetation canopy 400-2500 nm hyperspectral reflectance curve ² Simulate multispectral reflectance data from remote sensing sensors such as Landsat OLI and MODIS |
| Topic 6 : Parameter sensitivity analysis |
| ² Parameter selection to be optimized ² Local sensitivity analysis ² Global sensitivity analysis Introduction to F EFAST sensitivity analysis method F SIMLAB software operation process
|
| 专题七 遥感反演过程中的代价函数求解问题 |
| @ 代价函数构建 ² 反演方式 ² 反演参数 ² “病态”问题 ² 先验知识 ² 函数极值问题 @ 反演算法介绍 ² 优化技术 ² 查找表 ² 神经网络 ² 模拟退火 ² … @ 应用案例分析
|
| 专题八 基于查找表方法+PROSAIL模型的作物参数遥感反演 |
| 查找表原理 查找表实现 基于查找表和PROSAIL模型的作物参数遥感反演 |
| 专题九 基于优化算法+PROSAIL模型的作物参数遥感反演 |
| 代价/目标函数极值求解 测试函数极值求解 优化算法求解PROSAIL模型参数 待求解作物参数最优值提取 |
| 专题十 作物模型程序化表达与运行 |
| 模型分类 ² 经验模型 ² 半机理模型 ² 机理模型 模型选取原则 模型调试 模型标定 模型对比分析 应用案例分析 模型运行（以DSSAT作物模型为例、FORTRAN源码） ² 时间序列植被参数（如叶面积指数）演化模拟 ² 作物参数（如LAI）时间序列变化及产量模拟过程
|
| 专题十一 作物模型与遥感数据同化建模原理 |
| 作物模型与遥感观测耦合的必要性 ² 作物模型优缺点 ² 遥感观测优缺点 ² 耦合必要性 耦合方法 ² 驱动法 F 原理 F 程序实现过程 F 应用实例 ² 数据同化方法 F 发展历程 F 数据同化算法介绍
² 方法对比分析 作物模型参数敏感性分析 ² 待优化参数选择 ² 局部敏感性分析 ² 全局敏感性分析 作物模型与遥感数据同化 ² 同化遥感反演结果（如LAI遥感产品） ² 同化遥感观测反射率 |
| 专题十二 作物模型与遥感反演值同化建模的程序化实现（第一种方式） |
|
Fortrtan操作平台 遥感反演结果（如叶面积指数） 作物模型 变分算法 代价函数构建 迭代求解 输出：作物关键参数时间序列变化、产量估算结果、区域制图
|
| 专题十三 作物模型与遥感反射率同化建模的程序化实现（第二种方式） |
|
作物模型与遥感反射率同化建模框架 Fortrtan操作平台 遥感观测反射率 作物模型 植被冠层反射率模型 ² PROSAIL前向模型反射率模拟 耦合模型构建（作物模型+冠层反射率模型） 变分算法 代价函数构建 迭代求解 输出：作物关键参数时间序列变化、产量估算结果、区域制图 |
| 专题十四 互动、答疑、交流 |
原文连接：遥感数据与作物模型同化技术应用