Retrieval Research on pixel-level 3D Humidity fields by using GMS Multispectral Imagery

摘要

3D humidity field is implicated in radiance measurements of multi-spectral satellite imagery and cloud distribution observed different cloud type at vertical and horizontal direction. This paper first addresses the relationship between multispectral satellite information and probed relative humidity in each of standard isobaric surfaces using correlation analysis, least-square fitting and multivariate linear regression separately. In order to improve retrieval precision of lower layer humidity field, and at the same time, the objective analysis field of measured ground humidity data is introduced as a new parameter. Furthermore, based on spectral features of different bands as well as the difference day or night, the statistical retrieval models about humidity field analysis on each standard isobaric surface is established separately by using different spectrum composition, and the method is proposed retrieving 3D humidity fields of satellite imagery pixel resolution (0.1° lat and long) at all weather and all time by using multispectral satellite imagery. The contrast test between retrieval result and radiosonde data shows that the total deviation is about 15%. Generally, the retrieval precision in higher humidity area by using multispectral satellite imagery is superior to the one in lower humidity area and that in high and low layer is superior to that in middle layer. Because of the space-time disagreement between the satellite sounding and radiosonde observation, the retrieval error would be increased, which needs to be taken into account when the retrieved relative humidity field is analyzed and used.