Remotely sensed time series reveal varying levels of association between burned area and severity across regions in mainland Portugal

Abstract

Wildfires can pose severe threats to human lives and assets as well as for biodiversity at a global scale. Due to climate, land-use changes, and often inadequate forest management, wildfire regimes are also in transition, causing modifications in these disturbance events' spatial distribution, timing, size, frequency, and severity. In particular, increasing fire severity causes slower post-fire recovery times and a depletion in the resistance and resilience capacities which potentially trigger regime shifts in the state of ecosystems. Satellite remote sensing (SRS) Earth Observations (EO) allow us to characterize the ecological impacts of wildfires and assess spatiotemporal trends in fire severity. The SeverusPT project is currently pursuing the objective of harnessing SRS/EO time series to characterize wildfire severity and assess its impacts on ecosystems and human activities. Our primary objectives of this exploratory research paper are two-fold: (i) assess national and regional spatiotemporal trends in fire severity and burnt area, and (ii) evaluate if fire severity regionally scales up with the total burnt area. Satellite image time series (SITS) from Terra/MODIS were obtained to calculate fire severity through the Normalized Burn Ratio (NBR) and the difference between pre- and post-fire (ΔNBR). Trend analyses were employed to quantify fire severity and burned area spatiotemporal patterns. Linear regression assessed the association level between total burnt area by year/region (predictor) and fire severity (response). Preliminary results show that at a national level from 2001 until ca. 2008 - 2009 there was a general decrease in fire severity, followed by a reversal of this trend. This turning point has led to a general increase with new severity highs formed in 2017 and 2020. We also found wide variation in fire severity at the regional level (NUTS-III), and trend analysis display that most regions increased both in burned area and severity. Linear regression showed that burned area and fire severity are correlated despite this association being highly structured at the regional level, forming a continuous spectrum from highly area-severity coupled regions (e.g., AM Porto, Médio Tejo, Viseu, Coimbra, Alto-Minho) to less coupled ones (e.g., Cávado, Trás-os-Montes, Alentejo). These results may support that the increasing amount and size of burnt area will scale up into higher fire severity for specific regions. These preliminary results from the SeverusPT project show that MODIS image time series allow assessing the spatial variation and the temporal trends of fire severity in a standardized and synoptic fashion. Mapping fire severity, its spatiotemporal variation and addressing its environmental drivers are now more crucial than ever to understand its dynamics and support fire management and prevention.