Method for dynamic threshold method remote sensing data cloud identification supported by prior surface reflectance

Abstract

The invention discloses a method for dynamic threshold method remote sensing data cloud identification supported by the prior surface reflectance. The method comprises the steps that a, a prior surface reflectance database is established, national MODIS surface reflectance data are collected, and synthesis is conducted through a minimum value synthesis technology so that an MODIS high-accuracy surface reflectance data set can be formed; b, a dynamic threshold detection model is established on the basis of a 6S radiation transfer model, a change tendency of the apparent reflectance of the visible light and near-infrared band surface reflectance is simulated through the 6S radiation transfer model under different atmosphere and observation conditions, the minimum value and the maximum value of the change of the apparent reflectance of clear sky pixels are obtained under all the possible conditions, and a dynamic threshold cloud and cloud shadow detection model is established through nonlinear least square fit; c, data cloud and cloud shadow detection is conducted according to an HJ star CCD, the established dynamic threshold detection model is used for conducting an information extraction experiment of cloud identification and a cloud shadow. The method can effectively improve the cloud identification accuracy of satellite data with fewer wave bands.

Description

technical field [0001] The invention relates to a dynamic threshold method remote sensing data cloud recognition method supported by a priori surface reflectance, which is suitable for land observation satellites with fewer wave bands and the wave band settings are mainly concentrated in visible light and near-infrared wave bands. Background technique [0002] Clouds play an extremely important role in the energy balance and water cycle of the Earth-atmosphere system. The radiation budget balance of the earth-atmosphere system is mainly determined by the optical properties of clouds. At the same time, clouds also play an important role in the water cycle of the earth-atmosphere system. In order to estimate surface parameters, cloud parameters, and meteorological parameters based on the uplink radiation data obtained by satellites, it is first necessary to determine whether the observation pixel belongs to a cloud or a clear sky pixel, that is, to carry out cloud identificati...

AI Technical Summary

Problems solved by technology

At present, the most widely used threshold methods are ISCCP method (Rossow et al., 1989), APOLLO method (Saunders and Kriebel, 1988), CO 2 Thin slice method (Smith and Platt, 1978; Wylie and Menzel, 1989), etc. The above methods mainly use the difference in reflectivity and brightness temperature between clouds and typical surface objects, and use a fixed threshold method to realize the identification of clouds and the surface. For clouds, broken clouds, and cloud edges, the reflectance of a pixel is the result of the mixing of clouds and the surface, and the fixed threshold method is usually unable to identify the cloud coverage of this type of surface with high accuracy.

In addition, the threshold method requires satellite sensors to have a considerable number of channel settings, and the spectral range distribution is wide, involving visible light, near-infrared, mid-infrared and thermal infrared, but for some sensors, especially some high-resolution land observation satellites Sensors, such as my country's HJ-1CCD, often have fewer band settings, and the band distribution is mainly concentrated in a relatively narrow range of visible light and near-infrared. Traditional cloud identification methods are usually unable to effectively detect clouds for this type of sensor.

[0004] The traditional cloud pixel detection algorithm mainly relies on the reflectivity of visible light and near-infrared bands and the brightness temperature of thermal infrared bands. When this method is used for cloud and shadow recognition with less band data, it has two main disadvantages: one is due to Lack of brightness temperature support in the thermal infrared band, it is easy to cause misjudgment in high reflectivity areas and ice-covered areas; second, the existing research mainly uses the multi-spectral comprehensive threshold method to identify clouds. Due to the complexity of the surface type and the Due to the characteristics of "same object, different spectrum, same spectrum and different object" in the object spectrum, it is difficult to use a unified threshold to identify clouds and shadows.

In order to improve the cloud identification accuracy of mixed pixels and solve the difficulty of traditional threshold method for cloud identification of land observation satellites with fewer bands, it is necessary to propose a practical and effective cloud identification method

Images