Identification of the best algorithm for dust detection using MODIS data

Dust event is one of the atmospheric events of the world arid and semi-arid areas that had a significant increase in recent years and negative effects in different parts. In this study used MODIS data to identify and select the best algorithm for dust detection. For this purpose, three dust events of South West of Iran detected in 2012 using five different algorithms of dust detection including Ackerman BTD, Miller, dust index, TIIDI and DUST RGB methods, and methods compared. Studies show that methods of Ackerman BTD, Dust index, and Miller need to threshold regulation for each dust event; for this reason, suitable threshold was determined for each dust event using histogram method and dust identified. In addition, TIIDI method could separate dust phenomenon from other complications on the surface of the earth but as well could not identify dust on water. In DUST RGB method as well dust identified from other complication. In addition results of images classification and accuracy assessment showed that in all three dust events, DUST RGB method has maximum total accuracy among of other methods. Therefore, based on the results of matrix error and accuracy assessment, DUST RGB method was chosen as the best algorithm for dust detection.