遥感
环境科学
一致性(知识库)
云计算
均方误差
能量平衡
曲面(拓扑)
图像分辨率
气象学
钥匙(锁)
产品(数学)
缺少数据
随机森林
温度测量
曲面重建
参考数据
数据一致性
地球表面
数据挖掘
计算机科学
均方根
高分辨率
数据处理
空气温度
数据验证
近似误差
决定系数
水平衡
能量(信号处理)
点云
时间分辨率
摘要
Land Surface Temperature (LST) is a key parameter characterizing the energy balance at the Earth's surface. However, MODIS daily LST products are frequently disrupted by cloud cover, resulting in significant data gaps—particularly in cloudy and rainy regions—which severely limits their application. To address this issue, this study proposes an integrated reconstruction method for Guangxi that combines multi-source auxiliary data with a random forest algorithm. This approach not only effectively restores missing LST values under clouds to generate spatially continuous, seamless datasets but also simultaneously performs spatial downscaling, ultimately yielding high-precision LST products at 250m resolution. Site validation results demonstrate that the reconstructed and downscaled LST product exhibits high consistency with observed data, achieving a root mean square error (RMSE) of 0.87°C and a coefficient of determination (R2) of 0.8. This study provides valuable reference for developing high spatio-temporal resolution LST products.
科研通智能强力驱动
Strongly Powered by AbleSci AI