High-Resolution Satellite Observations Of Spatial And Temporal Variations Of Water Vapour In Typical Typhoon Environmental Fields

Typhoons often cause a variety of disasters such as high winds and heavy rainfall,and various secondary disasters such as floods,mudslides and landslides can cause serious damage to human life and properties,and under the current global warming background conditions sea surface temperatures are also rising,providing typhoons with more water vapour,which also makes the intensity of landing typhoon gradually increase,and the attenuation speed of typhoon after landing will slow down,and the duration on land will become longer.As a result of this phenomenon,typhoons have become increasingly harmful to human beings.Due to its geographical location,China is one of the countries most severely affected by typhoons,which is why typhoons have been the focus of research in atmospheric science in China.Water vapour is one of the most important factors affecting the whole life cycle of a typhoon,so the observations and study of atmospheric water vapor associated with typhoon genesis and evoluation are very important.Most of the existing studies on typhoon water vapour use reanalysis data.Reanalysis data is a combination of observations including satellite remote sensing data and numerical model,observations are assimilated to obtain highly accurate gridded atmospheric data with long time and spatial and temporal homogeneity.However,water vapour associated with typhoon is characterised by rapid spatial and temporal variations,therefore,the spatial and temporal resolutions of the current reanalysis data do not allow for depicting the small and mesoscale features in water vapour associated with the typhoon environment.Since geostationary weather satellite remote sensing data have high spatial and temporal resolutions,it is meaningful to use geostationary satellite high spatial and temporal resolution data to retrieve the atmospheric water vapour information with high spatial and temporal resolutions.There are many traditional methods such as one-dimensional variational(1DVAR)method for retrieving atmospheric water vapor information(profiles,layered precipitable water,total precipitable water),due to low latency requirement for real-time or near real-time applications,those methods can only retrieve water vapor information with low spatial and/or temporal resolutions.The random forest(RF)method has many advantages.This study attempts to use the RF method to retrieve water vapor information with high spatial and temporal resolutions,and compares it with the physical retrieval method(1DVAR method).Selected brightness temperature observations of the Advanced Himawari Imager(AHI)onboard the Himawari-8satellite during Typhoon Maria(201808)are used to retrieve the quantitative information of atmospheric water vapor in the environment.It is found that the retrieval method based on the RF algorithm has high computational efficiency,and the accuracy of the retrieved water vapor information is 35-45% higher than that of the 1DVAR method,which means that the full resolution water vapor retrieval can be achieved for potential applications through this method.Selecting a period of time for RF inversion results analysis,it was found that there were significant gradient changes in water vapor within an hour.These changes cannot be reflected in ERA5 data due to resolution limitations,but can be well displayed in high-resolution water vapor data inversion.In addition,a dataset corresponding to the resolution of current and future infrared imaging and observation instruments was constructed.It was found that the higher the spatial resolution,the more details of water vapor changes captured.This indicates that obtaining water vapor data with original resolution through high-resolution AHI observation data can effectively capture sub grid changes of water vapor.