Integrating Meteorological and Remote Sensing Geospatial Data for Forest Fire Risk Modelling in East Kalimantan Akhmad Haris Karsena (a,*), Tang-Huang Lin (a)
a) Center for Space and Remote Sensing Research, National Central University
No. 300, Zhongda Rd., Zhongli District, Taoyuan City 320, Taiwan
*hariskarsena[at]gmail.com
Abstract
Forest and land fires pose a recurrent dry season threat in East Kalimantan, Indonesia, undermining peatland ecosystems, public health, and economic security. This study evaluates whether an integrated remote sensing and meteorological geospatial framework that merges MODIS FIRMS hotspot density with meteorological, biophysical, and anthropogenic variables can deliver a more accurate and timely fire risk map for 2000 to 2020 than hotspot counts alone. A Multi Criteria Evaluation combined with a Weighted Linear Combination assimilates CHIRPS precipitation, ERA5 air temperature, relative humidity, 10 m wind speed, MODIS land surface temperature and NDVI, ESA WorldCover land cover classes, 30 m digital elevation metrics, and distances to roads and settlements, after each layer is fuzzy normalised and weighted via the Analytic Hierarchy Process. Model performance is evaluated against independent hotspot subsets using receiver operating characteristic analysis, area under the curve (AUC), and success rate metrics. Expected results point to clusters of high and very high risk in coastal peatlands and degraded agricultural fronts where August-October rainfall deficits, elevated temperatures, and low humidity coincide with intensive human activity, the composite model is anticipated to achieve an AUC above 0.80, indicating robust predictive skill. The aim of this research is to explore the spatial link between environmental conditions and forest fire activity, with the expectation that the results can support improved fire risk planning.
Keywords: Remote sensing, Meteorological data, Forest fire risk modelling, Fuzzy logic normalisation, Multi Criteria Evaluation (MCE), East Kalimantan
