Forest fire detection in 2001 using three-dimensional histogram

We have developed a fire detection method for the analysis of NOAA AVHRR data. The three-dimensional histogram technique is a method of displaying the brightness of the three different channels of AVHRR images. We assigned the X, Y and Z-axes of the three-dimensional space to the brightness of the three different channels respectively. In this study the three channels are CH1, CH3, CH5. We stored the fire information from 1996 to 2000 in a three-dimensional histogram. The target area is Far East Russia. The AVHRR images in 2001 are applied to the estimation of this method in order to detect the unknown fire points. This result is better than previously, especially the error detection of clouds as the fire decreases.