Geo-information knowledge base system for drought pattern search

Abstract

This paper presents a novel idea of utilizing a first-order logic technique to create, search and match for drought patterns in the specific area of interest. Drought patterns in this work have been drawn from the regression analysis using weekly time period and vegetation health index (VHI) obtained from the National Oceanic and Atmospheric Administration (NOAA) satellite data. To show the devised search facility, we use drought situations in the northeast provinces of Thailand as demonstrative cases. Drought trend of each province can be inferred from the percentage of provincial area that has the value of VHI below 35; the lower VHI value, the more severe drought. According to NOAA, this VHI threshold indicates moderate-to-high drought level. The proposed method is a kind of geo-information knowledge base system that allows users to search for drought pattern in some specific area at a particular time of the year. Moreover, the system also reports other area in the same region of study that shows similar drought trend. The drought trend is recognized from a regression coefficient in which the positive coefficient is the sign of increasing drought level, whereas the negative value implies the decrease of drought level.