CN-122017850-A - Supercooling cloud detection method of microwave radiometer

CN122017850ACN 122017850 ACN122017850 ACN 122017850ACN-122017850-A

Abstract

A supercooling cloud detection method of a microwave radiometer comprises the steps of obtaining bright temperatures by a multichannel K-band radiometer, inverting temperature and humidity data by a physical model and AI, obtaining bright temperatures of different polarizations by a quadrupole W-band radiometer, generating DoLP and AoP images, distinguishing supercooling cloud and ice crystal cloud, and judging whether the supercooling cloud is the supercooling cloud or not by integrating the relations of the temperature and humidity data, the DoLP and the AoP.

Inventors

WANG GAN
ZHENG ZHIMING
HUANG SHIHUA
Zhu haijie
XIA XIAOQING
LI WEI
XU ZIXIANG
ZHANG ZHEHAN

Assignees

南京国睿防务系统有限公司
中国商用飞机有限责任公司上海飞机设计研究院

Dates

Publication Date: 20260512
Application Date: 20260416

Claims (7)

1. A method for detecting supercooled cloud of a microwave radiometer, comprising: acquiring bright temperature by using a multichannel K wave band radiometer, and inverting temperature and humidity data by using a physical model and AI; acquiring different polarization bright temperatures by using a four-polarization W-band radiometer, generating DoLP and AoP images, and distinguishing supercooled cloud and ice crystal cloud; And thirdly, integrating the relationship among temperature and humidity data, doLP and AoP, and judging whether the cloud is supercooled cloud.
2. The method of detecting supercooled cloud of a microwave radiometer of claim 1, wherein said step one comprises: Dividing an atmospheric layer in any range into a plurality of height layers, acquiring atmospheric sounding data, drawing a liquid water content profile and a temperature profile of the range according to each height layer, and using the liquid water content profile and the temperature profile as references for judging supercooling clouds for training a neural network; calculating the analog temperature value of the multichannel K wave band radiometer by adopting an MPM atmospheric radiation transmission model; The BP neural network is used as an inversion algorithm, and a functional relation between a brightness temperature value and liquid water content and temperature of the multi-channel K-band radiometer is established; and obtaining the temperature profile and the liquid water content profile of each height layer of the atmosphere layer in the range according to the actual sky brightness temperature measured by the K-band microwave radiometer.
3. The method for detecting supercooled cloud of a microwave radiometer according to claim 1, wherein said step two comprises: simulating to generate a DoLP image, namely a quadrupolar image of a mixture of supercooled clouds and ice crystal clouds, wherein the quadrupolar image comprises 0-degree polarization, 90-degree polarization, 45-degree polarization and 135-degree polarization, the flat cylinder is judged to be the ice crystal clouds, and the sphere is judged to be the supercooled clouds; scanning sky imaging by using a quadrupole W-band radiometer to obtain different polarization brightness temperatures, and calculating linear polarization degree DoLP and polarization angle AoP; Generating a DoLP local standard deviation image, dividing the DoLP image by adopting an Otsu threshold algorithm, and distinguishing supercooling cloud and ice crystal cloud by using a cylinder and a sphere; And generating AoP a local standard deviation image, segmenting AoP the image by adopting an automatic threshold algorithm, and verifying supercooling cloud and ice crystal cloud.
4. The method according to claim 1, wherein the third step comprises determining according to a temperature profile, determining according to a liquid water content profile if the brightness temperature of the region is within a predetermined range, determining according to a DoLP image and AoP image if the liquid water content of the region is within a predetermined range, and determining as the supercooled cloud if the divided portion of the region is approaching zero.
5. The method for detecting supercooled cloud of microwave radiometer of claim 3, wherein said calculating linear polarization degree DoLP and polarization angle AoP comprises using the formula Computing a first component of a Stokes vector Second component Third component Fourth component Wherein The bright temperature value representing the horizontal polarization, The bright temperature value representing the vertical polarization, The bright temperature value representing 45 deg. polarization, The bright temperature value representing 135 deg. polarization, The brightness temperature value of the left-hand circular polarization is shown, The right-hand circular polarization brightness temperature value is expressed by the formula And (3) with The linear polarization DoLP and the polarization angle AoP are calculated.
6. The method of claim 5, wherein segmenting the DoLP image using the Otsu thresholding algorithm comprises representing a gray scale threshold in the DoLP image by t, 0≤t < L, L being an upper limit, The probability of representing each gray level in the image, And Two weights classified by a threshold t are represented, And Representing the mean divided into two parts by the threshold t, Representing the inter-class variance, let 、、、 Traversing the threshold t, determining the maximum inter-class variance And dividing the DoLP image by setting the pixel point smaller than the threshold value to 0 and setting the pixel point larger than the threshold value to 1.
7. The method for detecting supercooled cloud of microwave radiometer of claim 5, wherein said employing an automatic thresholding algorithm to segment AoP the image includes setting a neighborhood window, representing the domain size by NxN, calculating AoP each pixel point in the image Neighborhood mean of (2) With local standard deviation Wherein Representing pixel points Corresponding local standard deviation.

Description

Supercooling cloud detection method of microwave radiometer Technical Field The invention belongs to the technical field of supercooling cloud detection by a microwave radiometer, and particularly relates to a multi-channel radar detection, neural network inversion, temperature and liquid water profile application and DoLP and AoP image processing technology. Background The material radiates electromagnetic waves at a certain temperature, and the microwave radiometer can capture electromagnetic wave energy radiated, scattered and reflected by objects in the field of view of the antenna and convert the electromagnetic wave energy into blackbody temperature. The microwave radiometer mainly comprises an antenna, a receiver and a recording display device, and can deeply detect the position below the surface layer of an object to acquire information which is difficult to acquire by infrared light and visible light. The microwave radiometer has certain cloud layer penetrating capacity, is less interfered by meteorological factors such as fog, rain, snow and the like, can work in all weather, and has wide application in various fields such as atmosphere, ocean, vegetation, soil and the like. The cloud layer containing supercooled water is supercooled cloud, when an aircraft passes through the supercooled cloud, supercooled water can be frozen to form an ice layer by contacting with the aircraft body, and the aerodynamic performance is destroyed to cause flight accidents, so that aviation safety is threatened. Supercooled clouds lack condensation nuclei, and water is a small spherical liquid water condensate that radiates in an isotropic mode, characteristic of polarization. The water of the ice crystal cloud is condensed into flat dendrites with obvious polarization characteristics. The method for detecting the supercooling cloud distribution can ensure flight safety, and the prior art comprises three types, namely, a weather satellite with a microwave radiometer inverts cloud parameters, supercooling cloud can be identified in a large range, but resolution precision of middle and low cloud layers is not high, a sampling aircraft with the microwave radiometer directly acquires micro-physical parameters such as cloud particle phase state, supercooling water content and the like through cloud penetration, but the cost is high, an airspace is limited, the aircraft has icing risk and is difficult to operate normally, and ground detection adopts cloud radar, microwave radiometers and weather station cooperative networking, so that the method has the advantage of all weather and full-period low cost and becomes a current mainstream monitoring means. The ground detection depends on a microwave inversion algorithm, and has insufficient sensitivity to the mixed phase state of supercooled cloud and ice crystal cloud in cloud layers, and phase misjudgment is easy to cause. How to accurately identify the water-ice mixed phase state is a problem to be solved in the process of improving ground detection results and constructing an aviation early warning system. Disclosure of Invention In order to solve the technical problem that the ground detection supercooled cloud cannot effectively distinguish supercooled water in a mixed phase state from ice crystals, the supercooled cloud detection method of the microwave radiometer is adopted, and the technical effects of full time and low power consumption are achieved. And step one, acquiring bright temperature by using a multichannel K wave band radiometer, and inverting temperature and humidity data by using a physical model and AI. Firstly, dividing the atmospheric layer in any range into a plurality of height layers, acquiring atmospheric sounding data, drawing a liquid water content profile and a temperature profile of the range according to each height layer, and using the liquid water content profile and the temperature profile as references for judging supercooling clouds for training a neural network. And then, calculating the analog temperature value of the multichannel K-band radiometer by adopting an MPM atmospheric radiation transmission model. And then, using a BP neural network as an inversion algorithm to establish a functional relation between the brightness temperature value, the liquid water content and the temperature of the multi-channel K-band radiometer. And finally, obtaining the temperature profile and the liquid water content profile of each height layer of the atmosphere layer in the range according to the actual sky brightness temperature measured by the K-band microwave radiometer. And secondly, acquiring different polarization brightness temperatures by using a four-polarization W-band radiometer, generating DoLP and AoP images, and distinguishing supercooled cloud and ice crystal cloud. First, a DoLP image, i.e., a quadrupolar image of a mixture of supercooled clouds and ice clouds, comprising 0 ° polarization, 90 ° polarization, 45 ° polarization, 135 °