Daily 0.1° All-Weather Microwave Land Surface Emissivity (MLSE) Dataset for GMI

This dataset provides daily, all-weather microwave land surface emissivity (MLSE) estimates derived from observations of the Global Precipitation Measurement Microwave Imager (GMI). The product is provided on a 0.1° spatial grid with a daily temporal resolution over the study domain of 70°E–135°E and 15°N–55°N. The dataset contains MLSE estimates for nine GMI frequency–polarization channels: 10.65 GHz V/H, 18.7 GHz V/H, 23.8 GHz V, 36.5 GHz V/H, and 89.0 GHz V/H. Clear-sky reference emissivities were first retrieved by inverting the microwave radiative transfer equation using the RTTOV model. GMI Level 1C brightness temperatures, ERA5-Land skin temperature, and atmospheric temperature and humidity profiles from radiosonde observations were spatially and temporally collocated to generate physically based training samples. An all-weather machine-learning model was then developed using a Regressor Chain with Extremely Randomized Trees as the base estimator. Predictor variables include GMI brightness temperatures, Sentinel-1 SAR backscatter, NDVI, snow cover, land-cover type, elevation, geographic location, and day of year. The resulting product provides spatially continuous MLSE estimates under both clear and cloudy conditions. Its daily frequency and 0.1° grid preserve more short-term variability and finer land-surface heterogeneity than conventional monthly emissivity products. The dataset can support microwave retrievals of land and atmospheric variables, radiative-transfer simulations, and studies of land-surface microwave emission.