Sustainable Approach for Forest Fire Prediction

Abstract

With the rising global temperatures and already depleting forest cover there is one phenomenon which gets fiercer due to the first issue and worsens the second issue, respectively, and it is called forest fire. With already rapidly decreasing forests it is very important to curb, predict and mitigate the cases of forest fire which 90% of the times occurs due to a human error. Forests are of prime importance to the sustainability of the planet and hence desperately need to be conserved. The paper demonstrates the successful usage of various Data Mining models to predict the burned area in case of a forest fire by considering four low cost meteorological variables, temperature, rain, relative humidity and wind. A rigorous comparative study is presented of the performance of each Data Mining model for the given task and its reviewal reveals that Gene Expression Programming is the best at predicting forest fire area with four weather variables.