An improved coastline inflection method for geolocation correction of Microwave Radiation Imager data

Abstract

ABSTRACT Geolocation accuracy is an important factor affecting the application of microwave remote sensing data. For effective application of microwave radiation data in quantitative remote sensing, geolocation errors should be corrected. The coastline inflection method (CIM) is widely used for geolocation correction of microwave radiometer data, which consists of the inflection point identification process and the error estimation process. The detected coastline that is composed of the brightness temperature inflection points can be identified in the first process, and the error between the detected coastline and the true coastline can be estimated in the second process. However, both processes will be affected when the coastline is complex or disturbed by some clouds, rain, etc. Therefore, we improve the CIM in its two processes. Firstly, a method based on the dynamic threshold is used to identify inflection points. Secondly, a new geolocation error estimation method combining the Iterative Closest Point (ICP) algorithm and the Kernel Correlation (KC) algorithm is proposed to estimate geolocation errors. By applying the proposed method, the FengYun-3C (FY-3C) Microwave Radiation Imager (MWRI) dataset selected in the specified regions from 1 April 2019 to 1 October 2019 is processed. For the data with complex coastline structure or some interference factors, compared with the CIM based on a surface fitting interpolation model and ICP algorithm, the number of inflection points is increased by about 70%, and the mean geolocation errors in cross- and along-track directions are reduced by about 24.12% and 21.61%, respectively. For the data with simple coastline structure and less interference, the two methods have similar effects. For all selected data, the mean geolocation errors in cross- and along-track directions is reduced from [0.5376, 0.5528] pixels to [0.1581, 0.1110] pixels. Mean geolocation accuracy is improved by approximately 75%.