Method for estimating near-surface air temperature through remote sensing

Abstract

The invention provides a method for estimating near-surface air temperature by using remote sensing, which comprises the following steps: taking global ground observation data and reanalysis data as a data source, and equally dividing the global into a plurality of sub-regions according to latitude and land coverage types; establishing a database for each area by using ground observation data and reanalysis data in recent 10 years; establishing a linear or non-linear model, and recording the linear or non-linear model as Model-1; meanwhile, establishing a machine learning model with the same sample library, and recording the machine learning model as Model-2; carrying out weighted fusion on the two models to obtain a final model Model-fused of each sub-region; and for a certain area, based on the established Model-fused model, inputting parameters obtained by remote sensing to obtain a corresponding near-surface temperature. According to the invention, high-precision and time-space continuous near-surface temperature remote sensing estimation based on a conventional remote sensing product is realized, and the defects of low precision, coarse time-space resolution and even dependence on a specific sensor of an existing product are overcome.

Description

technical field [0001] The invention belongs to the technical field of meteorology, serves scientific research and industrial application relying on air temperature data, and particularly relates to a method for estimating near-surface air temperature by using remote sensing. Background technique [0002] The near-surface air temperature (2m) is generally obtained through instrument observation by meteorological stations, and has a wide range of uses in the fields of global change, meteorology, urban planning, and agriculture. However, the number of meteorological observation stations is limited and the distribution is not uniform, so related applications and research are severely limited. [0003] With the development of remote sensing technology, using remote sensing data to directly estimate near-surface air temperature has great application prospects. However, using remote sensing to estimate air temperature generally requires special band settings (for example, microwav...

AI Technical Summary

Problems solved by technology

[0005] On the premise that no specific satellite observation data is required, the present invention utilizes surface temperature data observed by conventional remote sensing and other auxiliary environmental parameters related to air temperature to establish a linear or nonlinear model between surface temperature and near-surface air temperature, and realizes the relationship between surface temperature and remote sensing surface temperature. Estimation of near-surface air temperature with the same temporal and spatial resolution, solving the problems of temporal and spatial discontinuity, insufficient accuracy and resolution of existing remote sensing air temperature data

Images