Physically Consistent Synthesis of Tropical Cyclone Microwave Brightness Temperatures From Geostationary Infrared Observations

Abstract

Analyzing the evolution of Tropical Cyclones (TCs) is critical for understanding their structure and intensity, but it has been limited by observational constraints. Spaceborne passive microwave (PMW) observations can penetrate both non‐precipitating and precipitating clouds and provide information on the vertical distribution of hydrometeors at appropriate frequencies, but they suffer from the sparse temporal sampling of low‐Earth orbit satellites. In comparison, geostationary infrared (IR) radiometers provide continuous monitoring but are limited to cloud‐top information. To capture the continuous evolution of TC structure, we propose a pixel‐space diffusion transformer for synthesizing PMW brightness temperatures conditioned on geostationary IR observations. Beyond visual realism, the data‐driven model learns the nonlinear mapping from the IR to PMW and then quantifies the uncertainty inherent in the synthesis. Saliency map analyses reveal that the model captures distinct physical dependencies by linking high‐frequency scattering signatures to cloud‐top thermal features and by inferring low‐frequency emission signals from liquid water using a broader environmental context. Moreover, sensitivity experiments on the physics of concentric eyewall formation in TCs confirm that the synthesis responds consistently to TC dynamics. This model can compensate for sparse microwave observations, providing a new pathway for TC monitoring and forecasting.