Evaluating the Best Spectral Indices for the Detection of Burn Scars at Several Post-Fire Dates in a Mountainous Region of Northwest Yunnan, China

Abstract

Abstract:Remote mountainous regions are among the Earth's last remaining wild spots, hosting rare ecosystems and rich biodiversity. Because of access difficulties and low population density, baseline information about natural and human-induced disturbances in these regions is often limited or nonexistent. Landsat time series offer invaluable opportunities to reconstruct past land cover changes. However, the applicability of this approach strongly depends on the availability of good quality, cloud-free images, acquired at a regular time interval, which in mountainous regions are often difficult to find. The present study analyzed burn scar detection capabilities of 11 widely used spectral indices (SI) at 1 to 5 years after fire events in four dominant vegetation groups in a mountainous region of northwest Yunnan, China. To evaluate their performances, we used M-statistic as a burned-unburned class separability index, and we adapted an existing metric to quantify the SI residual burn signal at post-fire dates compared to the maximum severity recorded soon after the fire. Our results show that Normalized Burn Ratio (NBR) and Normalized Difference Moisture Index (NDMI) are always among the three best performers for the detection of burn scars starting 1 year after fire but not for the immediate post-fire assessment, where the Mid Infrared Burn Index, Burn Area Index, and Tasseled Cap Greenness were superior. Brightness and Wetness peculiar patterns revealed long-term effects of fire in vegetated land, suggesting their potential integration to assist other SI in burned area detection several years after the fire event. However, in general, class separability of most of the SI was poor after one growing season, due to the seasonal rains and the relatively fast regrowth rate of shrubs and grasses, confirming the difficulty of assessment in mountainous ecosystems. Our findings are meaningful for the selection of a suitable SI to integrate in burned area detection workflows, according to vegetation type and time lag between image acquisitions.

Integration of Multiple Spectral Indices and a Neural Network for Burned Area Mapping Based on MODIS Data

“…The primary reason for the results is the differences of the fuels involved in the three wildfire cases, including the vegetation differences between the long grass of Hot Pot Fire [64], the tall grass of Range 12 Fire [65], and the oak canopy of Rey Fire. Also, the varying space and time of wildfires can cause inconsistent performance of the spectral band [5]. In addition, the results also highlight the regional limitation of the single spectral band that is not normalized to the pre-fire spectral condition, as found by Reference [66].…”

“…In this study, Landsat 8 OLI Collection 1 L1TP products were processed to generate the reference burned area map as ground truth. The OLI products have already been radiometrically calibrated and orthorectified using ground control points and digital elevation model (DEM) data to correct for relief displacement [5]. Here are the processing steps of Landsat-8 OLI data with ENVI software: Firstly, the L1TP products were atmospherically corrected to obtain surface reflectance free from the influence of atmosphere, solar illumination and similar interferences [27,47].…”

“…Fire also plays an important role in ecosystem succession [1], global carbon budget [2], climate cycles [3], and land cover change [4]. Since wildfires ruin the terrestrial vegetation layer, accurate burned area mapping is essential for disaster assessment, burned landscape management and vegetation recovery [5][6][7]. In recent years, the topic has become a research hotspot because of its important practical significance.…”

“…In comparison to manual field measurement of a burned area after wildfires, satellite remote sensing is more convenient, safe and applicable with fast response capability. So far, the satellite sensors applied to burned area research mainly consist of the VEGETATION (VEG) [8][9][10], Advanced Very High Resolution Radiometer (AVHRR) [2,[11][12][13][14][15], Moderate Resolution Imaging Spectroradiometer (MODIS) [1,[16][17][18][19], Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) [20][21][22], Visible Infrared Imaging Radiometer Suite (VIIRS) [7], Thematic Mapper (TM) [5,23,24], Enhanced Thematic Mapper plus (ETM+) [24][25][26][27], and Operational Land Imager (OLI) [28]. Besides, the availability of Synthetic Aperture Radar (SAR) for burned area mapping has also been investigated [25,29,30].…”

“…Furthermore, the spectral indices have been widely used to represent the spectral characteristics of the burned area, vegetation and other types. Stroppiana et al [10,20,24,25,27], Veraverbeke et al [22,34], and Xiao et al [5] evaluated the capacity of spectral indices in discrimination of the burned area and unburned types based on the satellite data. These algorithms based on multi-temporal satellite data are typical and can utilize the information on changes before and after the wildfire; whereas the traditional threshold-based methods with a single satellite image need to manually extract the samples to do analysis and cannot make the best of the spectral differences Remote Sens.…”