Evaluating the Effectiveness of Spectral Indices in Assessing Fire Severity Using Google Earth Engine: A Case Study of the Wildfires in Northern Rural Latakia, Syria (July 2025)

Abstract

This study aimed to assess forest fire severity in northern rural Latakia during July 2025 using five spectral indices: NBR, NBRT, NDVI, NDVIT, and VI6T, before and after the fires, and to relate them to climatic data (temperature and wind speed), utilizing Google Earth Engine for satellite image analysis. Results showed a decrease in mean index values post-fire, e.g., NBR dropped from 0.25 to 0.22 and NBRT from 0.05 to 0.03, reflecting fire impact. Thermal indices NDVIT, NBRT, and VI6T were more sensitive to fire than vegetation indices, especially in affected areas. Fires were mainly located in Qastal Ma‘af, Jabal Turkman, and Nab‘ al-Mar villages. NDVI and NDVIT estimated higher fire-affected areas in some regions due to drought influence. Evaluation results indicated NDVIT and NBRT achieved the highest overall accuracy (0.91) and kappa coefficient (0.82), highlighting their effectiveness. NDVI showed lower accuracy despite higher degraded area estimates. Degraded area estimates by NDVI and NDVIT reached ~21,959 ha (8%), exceeding the 20,000 ha reference by 9.8%, whereas NBR estimated 18,455 ha (7%), and NBRT and VI6T focused on severely burned areas (~13,867 ha, 5%). Correlation analysis indicated temperature was the main driver of fire spread (0.843 with NBR), while wind had less impact, confirming temperature as a key factor in fire dynamics in northern rural Latakia forests.