Scale decomposition of precipitation patterns and nowcasting in a high-resolution X-band radar network

Shortterm weather forecasting (nowcasting) performance can be improved by spatially filtering radar images of precipitation patterns and either predicting only those precipitation scales most representative of pattern motion or removing unpredictable scales after prediction. Previous research has shown that improvement in nowcasting precision can be obtained by first applying an elliptical spatial filtering procedure to observed WSR-88D Vertically Integrated Liquid water data to predict larger-scale features believed to better represent the motion of precipitation patterns. Another study used the wavelet transform to develop measures of predictability at each scale and designed adaptive wavelet filters to remove perishable scales from predicted continental-scale reflectivity data. This study investigated the adaptation of both approaches and Fourier filtering to examine the effects of scale filtering on nowcasting performance using a spectral-based nowcasting method and high-resolution Collaborative Adaptive Sensing of the Atmosphere radar reflectivity data.