摘要: |
日最低气温是判断作物是否遭受低温冷害的主要因子,开展日最低气温推算在作物低温冻害保险理赔服务中具有实际应用意义。本研究以贵州省为例,选取了2016年2月10、2023年1月31日(代表冬季)和2017年4月1日(代表春季)贵州省范围内的小时气温、遥感反演的夜间地表温度、地表高程、遥感水汽含量等多源数据,对数据进行预处理和质量控制后,采用相关分析法分析了不同数据与日最低气温的线性相关关系,利用相关性显著的地表温度、经度、纬度、水汽含量和高程数据构建了3个不同日期的日最低气温线性推算模型。经验证数据验证,使用线性显著参数构建的最优多元线性回归模型推算的2016年2月10、2017年4月1日和2023年1月31日的最低气温平均绝对误差(MAE)分别为1.224℃、0.894℃和1.727℃,均方根误差(RMSE)分别为1.518℃、1.201℃和2.132℃。表明采用夜间过境的卫星遥感地表温度对推算当日的日最低气温具有较好的效果,该工作对于低温农业保险应用具有一定的实用前景。 |
关键词: 遥感推算,日最低气温,地表温度,多元线性回归 |
DOI: |
投稿时间:2022-11-01修订日期:2023-07-20 |
基金项目:贵州山地茶叶保险气象指数研究及平台建设(黔农财[2017]20号);喀斯特石漠化植被生态质量变化对气候因子的响应(黔科合基础-ZK[2022]一般 273);基于高分卫星的火烧迹地自动提取及林火烈度模型研究(黔科合基础-ZK[2021]一般 193) |
|
Study on remote sensing estimation of daily minimum temperature over Guizhou from polar orbit satellite |
yangshijin,Tianpengju,hujiamin,liaoyao |
(Guizhou Meteorological Service Center;Guizhou Ecological Meteorology and Satellite Remote Sensing Center;Guizhou Institute of Mountain Environment and Climate) |
Abstract: |
The daily minimum temperature is the main factor to judge whether crops suffer from low temperature and cold damage. It is of practical significance to calculate the daily minimum temperature in crop low temperature insurance claim service. This study obtained multi-source data such as hourly temperature, remote sensing inversion surface temperature of the former night time, surface elevation and remote sensing water vapor content in Guizhou Province on February 10,2016, January 31,2023 (represents winter) and April 1,2017 (represents spring). After data preprocessing and quality control, the correlation analysis method was used to analyze the linear correlation between different data and daily minimum temperature, and 3 of the linear calculation model of daily minimum temperature was built by using the surface temperature, longitude, latitude, water vapor content and elevation data with significant correlation. The linear regression model constructed using the above five significant parameters is better than the linear regression model constructed using only the surface temperature, which is the most significant correlation factor. After comparison on validation datasets, the mean absolute error (MAE) of the best linear model are 1.224℃,0.894℃ and 1.727℃ for February 10,2016, April 1, 2017 and January 31,2023,respectively.While the root mean square error (RMSE) are 1.518℃ ,1.201℃ and 2.132℃. This work indicates better results can be obtained through surface temperature data of the former night, which has a certain application prospect for the low-temperature agricultural insurance. |
Key words: Remote sensing retrieval, Daily minimum temperature, Land surface temperature,Multiple linear regression |