ABSTRACT

This study presents a framework for spatially interpolating Cyclone Global Navigation Satellite System (CYGNSS) data over land. The method, Previously-Observed Behavior Interpolation (POBI), assumes that reflectivity data can be interpolated at unsampled locations by quantifying how past observations of reflectivity at the unsampled location varied with past observations of reflectivity at nearby locations. The POBI method is able to recreate truth reflectivity maps with a mean error of 0.17 dB (standard deviation 1.96 dB) and preserves the heterogeneous and discontinuous nature of reflectivity observations. This method is useful for applications that necessitate rapidly-updated data but have some tolerance for uncertainty in interpolated points.

摘要

本研究提出了一个框架，用于在陆地上对气旋全球导航卫星系统 (CYGNSS) 数据进行空间插值。该方法称为先前观察到的行为插值 (POBI)，它假设反射率数据可以通过量化过去对未采样位置反射率的观察结果与过去对附近位置反射率的观察结果的变化情况，在未采样位置进行插值。POBI 方法能够重新创建平均误差为 0.17 dB（标准偏差 1.96 dB）的真实反射率图，并保留反射率观测的异质性和不连续性。此方法适用于需要快速更新数据但对插值点的不确定性有一定容忍度的应用程序。