Validation of FY-4A AGRI layer precipitable water products using radiosonde data

The Advanced Geostationary Radiation Imager (AGRI) onboard the first of China's next-generation geostationary meteorological satellite series (FY-4A) can provide high temporal resolution layer precipitable water (LPW) products. The AGRI LPWs, including total precipitable water (TPW), low (PW_low), middle (PW_mid), and high (PW_high) levels precipitable water, were first validated using a year radiosonde data. The results show that AGRI LPWs generally agree well with radiosonde derived LPWs, and the accuracy of LPWs demonstrates obvious spatial and temporal patterns. All LPWs at 00 UTC were underestimated and the root means square error (RMSE) of TPW, PW_low, PW_mid, and PW_high at 00 UTC are 6.04, 1.79, 3.32, and 2.68 mm, while the correlation coefficients (R) are 0.953, 0.958, 0.925, and 0.887, respectively. In contrast, AGRI LPWs at 12 UTC have no obvious underor over-estimation, and shows a lower RMSE (1.49-4.4 mm) and a higher correlation (0.934-0.971). PW_high shows the largest mean absolute percentage error (MAPE) (>39%), while the MAPE of the other three LPWs is less than 29%. The RMSE of all four AGRI LPWs generally increase with water vapor, while R and MAPE decrease with water vapor. Thus, the AGRI LPWs over low latitudes with high water vapor content show larger RMSE while lower MAPE and R than mid and high latitudes. The accuracy of AGRI LPWs displays a seasonal pattern, with higher RMSE, lower R and MAPE in summer wet months, while lower RMSE, higher R and MAPE in winter dry months.