publised in Remote Sensing; doi:10.3390/rs11151741

Abstract

Total precipitable water (TPW), a column of water vapor content in the atmosphere, provides information on the spatial distribution of moisture. The high-resolution TPW, together with atmospheric stability indices such as convective available potential energy (CAPE), is a good indicator of severe weather phenomena in the pre-convective atmospheric condition. With the advent of high performing imaging instrument onboard geostationary satellites such as Advanced Himawari Imager (AHI) onboard Himawari-8 of Japan and Advanced Meteorological Imager (AMI) onboard GeoKompsat-2 of Korea, it is expected to provide unprecedented spatio-temporal resolution data (e.g., AMI plans to provide 2 km resolution data at every 2 minutes over the northeast part of East Asia). To derive TPW from such high resolution data in a timely fashion, an efficient algorithm is highly required. Here, machine learning approaches—random forest (RF), extreme gradient boosting (XGB), deep neural network (DNN)—are assessed for TPW retrievals from AHI over the clear sky in Northeast Asia area. For the training dataset, the 9 Infrared brightness temperatures (Tb) of AHI (Tb8 to 16 centered at 6.2, 6.9, 7.3, 8.6, 9.6, 10.4, 11.2, 12.4, and 13.3 μm, respectively), 6 dual channel differences (DCD) and observation conditions such as time, latitude, longitude, and satellite zenith angle for two years (September 2016 to August 2018) are used. The corresponding TPW is prepared by the integrating water vapor profiles from the European Centre for Medium-Range Weather Forecasts re-analysis data (ERA-Interim). The algorithm performances are assessed using the reference data including ERA-Interim and radiosonde observation (RAOB) data. The results show that the DNN model performs better than RF and XGB with a correlation coefficient of 0.95, a mean bias of -0.16 mm, and a root mean square error (RMSE) of 3.74 mm when compared to the ERA-Interim reference data. Similarly, DNN results in a correlation coefficient of 0.97, a mean bias of -0.05 mm, and a RMSE of 4.33 mm when compared to RAOB. Contributing variables by model for TPW retrievals and the spatial distribution of the estimated TPW are carefully examined and discussed.