Neural network based methods for cloud classification on AVHRR images

Abstract

Arti cial neural networks trained on spectral and textural features extracted from Advanced Very High Resolution Radiometer (AVHRR) images have been used to develop an automated cloud classi cation system. Selection of the optimum combination of features was achieved by using statistical methods presented in earlier work by Gu et al. and by running large numbers of neural network simulations on test datasets. The performance of these methods surpasses that of other approaches such as the use of Gabor lters for texture segmentation and the maximum likelihood classi er. A particular architecture for an operational classi cation system is presented based on a two-stage multiple network con guration which is shown to segment complex images to a high degree of accuracy and achieves an overall accuracy on an independent, representative test set of 91%.

A Universal Dynamic Threshold Cloud Detection Algorithm (UDTCDA) supported by a prior surface reflectance database

“…However, because the principle of such a method is unclear, training and validation samples are required to cover most conditions of land surface and cloud type. Because the conditions are not specified during training, this method is less accurate [ Karlsson , ; Clark and Boyce , ; Walder and Maclaren , ]. The object‐oriented method is designed to segment the images into meaningful “objects,” which can be described as a set of features, and realizes the “object” classification by the established relations or differences between the object and class structure.…”

Ground nephogram classification based on textural feature extraction and neural networks

“…Ground-based image analysis of clouds has not been studied in detail in the past. One of the reasons why automated ground-based cloud recognition has been ignored is because it proves to be a more difficult task than satellite image analysis task for cloud identification [1][2][3][4][5][6]. But the ground-based observation is also an essential method to get cloud parameters.…”

The Potential for Self-Organizing Maps to Identify Model Error Structures

“…SOMs have been widely applied to characterize synoptic patterns based on sea level pressure or geopotential height (e.g., Hewitson and Crane 2002;Cavazos et al 2002;Michaelides et al 2007;Johnson et al 2008;Schuenemann and Cassano 2009). SOMs have also been used in remote sensing to identify cloud type (e.g., Tian et al 1999;Walder and MacLaren 2000). SOMs have also been used in remote sensing to identify cloud type (e.g., Tian et al 1999;Walder and MacLaren 2000).…”