AUTHOR=Kale Manish P. , Mishra Asima , Pardeshi Satish , Ghosh Suddhasheel , Pai D. S. , Roy Parth Sarathi 

TITLE=Forecasting wildfires in major forest types of India

JOURNAL=Frontiers in Forests and Global Change

VOLUME=Volume 5 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/forests-and-global-change/articles/10.3389/ffgc.2022.882685

DOI=10.3389/ffgc.2022.882685

ISSN=2624-893X

ABSTRACT=Severity of wildfires witnessed in different parts of the world in the recent times has posed a significant challenge to fire control authorities. Even when the different fire alert systems have been developed and deployed to provide the quickest possible alerts about the wildfire location, severity and danger, often it is difficult to deploy the resources quickly to contain the wildfire at short notice. Response time is further delayed when the terrain is complex. 
Though, the sophisticated numerical two-way coupled models such as WRF-SFIRE have been developed and deployed in different parts of the world to quickly (hourly basis) predict the wildfire spread once the ignition location is detected, the statistical models based on long time-series datasets have their own importance as they are extremely useful in providing monthly fire forecasts. This gives fire control authorities sufficient time to prepare to control the possible region-specific fire events and deploy the resources at vulnerable locations. This becomes extremely important when terrain is complex. 
Considering the problem in hand the present research has been carried out with the aim of forecasting the number of wildfires in different forest types of India on monthly basis using ‘Autoregressive Integrated Moving Average’ time series models (both univariate and with regressors) at 25km × 25km spatial resolution (grid). 
The performance of different models was validated based on the autocorrelation function, partial autocorrelation function, cumulative periodogram and Portmanteau (L-Jung Box) test. The univariate model was preferred to have better parsimony. The R software package was used to run and test the model. 
The forecasted result was tested against the original fire counts for three years monthly forecast from 2015 to 2017 and variation in coefficient of determination from 0.94 for year 1 forecast (2015) to 0.64 when all the three-year forecasts were considered together was observed for tropical dry deciduous forests. These values varied from 0.98 to 0.89 for tropical moist deciduous forest and from 0.97 to 0.88 for tropical evergreen forest. 
The forecasted fire counts were used to estimate future forest fire frequency ratio which has been used as an indicator of the intensity of wildfire.