Precipitation type identification method based on precipitation particle multi-characteristic parameters

Abstract

The invention discloses a rainfall type identification method using rainfall particle multi-characteristic parameters. The method comprises the following steps of: establishing a feature vector required by a precipitation particle recognition model by utilizing feature parameters such as precipitation particle geometric information, gray scale information and speed information measured by an observation instrument; according to the actual measurement particle image information, dividing the main rainfall particle types into ice crystal small particles, dendritic snowflakes, columnar snowflakes, aggregated snowflakes and raindrops; and using a support vector machine method to establish a rainfall particle type recognition model, and achieving fine recognition and classification of rainfalltypes. According to the invention, fine classification and identification of precipitation particles can be effectively realized. The measurement performance of the imaging raindrop spectrometer can be further exerted, and the fineness degree of rainfall type description is improved.

Description

technical field [0001] The invention relates to the technical field of meteorological automatic measurement and radar remote sensing precipitation ground correction, in particular to a precipitation type identification method based on multiple characteristic parameters of precipitation particles. Background technique [0002] Using precipitation particles to distinguish precipitation types is a challenging and important task in atmospheric science research. With the continuous development of numerical weather prediction, the construction, improvement, and calibration of its models are more sensitive to precipitation types, and the commonly used coarse classification methods such as rain, snow, and hail gradually cannot meet the relevant needs. Conventional raindrop spectrometers can measure the diameter and velocity of precipitation particles, and then distinguish precipitation types based on the correspondence between diameter and velocity, but cannot be further subdivided....

AI Technical Summary

Problems solved by technology

With the continuous development of numerical weather prediction, the construction, improvement, and calibration of its models are more sensitive to precipitation types, and the commonly used rough classification methods such as rain, snow, and hail are gradually unable to meet the relevant needs

The conventional raindrop spectrometer can measure the diameter and velocity of precipitation particles, and then distinguish the precipitation type according to the corresponding relationship between diameter and velocity, but it cannot be further subdivided

Images