Vegetation and Soil Fire Damage Analysis Based on Species Distribution Modeling Trained with Multispectral Satellite Data

Abstract

Forest managers demand reliable tools to evaluate post-fire vegetation and soil damage. In this study, we quantify wildfire damage to vegetation and soil based on the analysis of burn severity, using multitemporal and multispectral satellite data and species distribution models, particularly maximum entropy (MaxEnt). We studied a mega-wildfire (9000 ha burned) in North-Western Spain, which occurred from 21 to 27 August 2017. Burn severity was measured in the field using the composite burn index (CBI). Burn severity of vegetation and soil layers (CBIveg and CBIsoil) was also differentiated. MaxEnt provided the relative contribution of each pre-fire and post-fire input variable on low, moderate and high burn severity levels, as well as on all severity levels combined (burned area). In addition, it built continuous suitability surfaces from which the burned surface area and burn severity maps were built. The burned area map achieved a high accuracy level (κ = 0.85), but slightly lower accuracy when differentiating the three burn severity classes (κ = 0.81). When the burn severity map was validated using field CBIveg and CBIsoil values we reached lower κ statistic values (0.76 and 0.63, respectively). This study revealed the effectiveness of the proposed multi-temporal MaxEnt based method to map fire damage accurately in Mediterranean ecosystems, providing key information to forest managers.

Can Landsat-Derived Variables Related to Energy Balance Improve Understanding of Burn Severity From Current Operational Techniques?

“…From a multi-temporal perspective, both dET_i-and dLST_i-based estimates showed a lower accuracy than dNDVI_i-and dNBR_i-based ones (multi-temporal approach) ( Table 6). The obtained accuracy values were in accordance with accuracy values found in previous studies (e.g., [18,60,75,[113][114][115][116][117]).…”

“…As in previous studies [75,76], LST was calculated based on the single-channel method proposed by Yu et al [77]. This method is based on the radiative transfer equation (Equation 1).…”

“…Although thresholding the differenced form of NBR (dNBR) [6] has become a standard method to estimate burn severity due to its easy computation [60], some limitations of this method have been pointed out (see [66]) and new alternatives have been proposed. Among them, the ones based on char fraction from unmixing reflective bands stand out [66,75,[108][109][110].…”

A Synergetic Approach to Burned Area Mapping Using Maximum Entropy Modeling Trained with Hyperspectral Data and VIIRS Hotspots

“…That continuous suitability surface (or probability of occurrence map) is the most valuable information for forest managers, as they might select an appropriate threshold according to the project priorities [61,62]. Thus, the availability of continuous suitability surfaces as MaxEnt output constitutes one advantage over well-established classifiers [32,62].…”

“…MaxEnt has been commonly used as a model for species distributions [25][26][27], but it is being progressively used in remote sensing based applications as well [12,28,29]. Additionally, it has been successfully used to model fire occurrence [30,31] and burn severity [12,32]. [30] used MaxEnt to estimate both burn severity and burned area from Sentinel-2 Multispectral Instrument (MSI) Multiple Endmember Spectral Mixture Analysis (MESMA) fraction images and post-fire Landsat 8 OLI LST images.…”