Gene expression programming for forest fire risk modeling in western Himalayas

Western Himalayas are mainly prone to chir pine forest fires, which are predominantly governed by climatic factors. Forest fire is one of the main reasons for forest degradation and has a hazardous impact on the environment, economy, and human health. Therefore, the present investigation aimed to develop forest fire risk models based on climatic parameters using gene expression programming (GEP) for Solan district of Himachal Pradesh. Climatic parameters viz., maximum temperature (T_x), minimum temperature (T_n), mean temperature (T_a), soil temperature (T_s), maximum relative humidity (RHx), minimum relative humidity (RH_n), mean relative humidity (RH_a), rainfall (RF), sunshine hours (SS) and wind speed (WS), for the past fifteen years was randomly divided into a training set (75%) and validation set (25%). Training data was used to construct eight models, which had different combinations of ten weather parameters, and the models were validated using validation data. Several statistical criteria, viz., coefficient of determination (R²), Pearson’s correlation coefficient (r), and statistical errors were used for the evaluation of the performance of Models. Model 2, Model 5, and Model 8 showed better performance in both the training and validation stage; however, among these models, Model 2 (R² = 1.00%; r = 1.00) was selected and described. Model 2 was generated using temperature, relative humidity, and rainfall as input data. This model can be exploited to predict and prevent forest fire hazards in the study area.

• The study aimed to develop forest fire risk models using gene expression programming (GEP) for Solan district of Himachal Pradesh.

• Model 2, Model 5 and Model 8 were the best-performing models.

• Model 2 (R² = 1.00%; r = 1.00) was selected for further description, and Model 2 was generated using temperature, relative humidity and rainfall as input data.