An improved typhoon monitoring model based on precipitable water vapor and pressure

The potential of monitoring the movement of typhoons using the precipitable water vapor (PWV) has been confirmed. However, monitoring the movement of typhoon algorithms is focused on PWV, making it difficult to describe the movement of a typhoon in detail minutely and resulting in insufficient accuracy. Hence, based on PWV and meteorological data, we propose an improved typhoon monitoring mode. First, the European Centre for Medium-Range Weather Forecasts Reanalysis 5-derived PWV (ERA5-PWV) and the Global Navigation Satellite System-derived PWV (GNSS-PWV) were compared with the reference radiosonde PWV (RS-PWV). Then, using the PWV and atmospheric parameters derived from ERA5, we discussed the anomalous variations of PWV, pressure (P), precipitation, and wind speed during different typhoons. Finally, we compiled a list of critical factors related to typhoon movement, PWV and P. We developed an improved multi-factor typhoon monitoring mode (IMTM) with different models (i.e., IMTM-I and IMTM-II) in different cases with a higher density of GNSS observation or only Numerical Weather Prediction (NWP) data. The IMTM was evaluated through the reference movement speeds of HATO and Mangkhut from the China Meteorological Observatory Typhoon Network (CMOTN). The results show that the root mean square (RMS) of the IMTM-I is 1.26 km/h based on ERA5-P and ERA5-PWV, and the absolute bias values are mostly within 2 km/h. Compared with the models considering the single factor ERA5-P/ERA5-PWV, the RMS of the IMTM-I is improved by 26.3% and 38.5%, respectively. The IMTM-II model manifests a residual of only 0.35 km/h. Compared with the single-factor model based on GNSS-PWV/P, the residual of the IMTM-II model is reduced by 90.8% and 84.1%, respectively. These results propose that the typhoon movement monitoring approach combining PWV and P has evident advantages over the single-factor model and is expected to supplement traditional typhoon monitoring.