Mapping Forest Burned Areas in the Indio Maiz Biological Reserve Using SENTINEL-3 Slstr Imagery
Abstract
Wildfires are considered as one of the major hazards and environmental issues around the world. Considerable ecological damage is caused by the large amount of greenhouse gases emitted during the biomass burning process, in addition to the acceleration of the soil erosion rate as a result of the changes in soil properties. Earth Observation Satellites provide information useful for detecting and monitoring forest fire events, as well as for managing and evaluating environmental damages. In the last decades, optical sensors have been proven to be effective for wildfires detection and tracking, although the quality and usefulness of optical data are often hindered by the presence of clouds. One practical workaround is to combine different dataset from multiple sensors.
This research presents a methodology of day/night monitoring and tracking wildfires events by employing optical data from Sentinel-2, Sentinel-3 SLSTR, Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS). For this study, a single forest fire event that took place in the Indio Maiz Biological Reserve, Nicaragua, during April 3-13th of 2018, was selected as the study case. A total of six S3-SLSTR images were downloaded and preprocessed. The images were visualized with different combinations of bands, including those specific for active fire detection. Spectral indices such as Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI) and Normalized Burned Ratio (NBR) were derived from each scene, and all the imagery was subsequently segmented into image objects. These objects were merged into burned area image objects. Comparisons of burned area obtained from each sensors were made in order to corroborate S3-SLSTR imagery fire-detection capabilities. Furthermore, VIIRS Day Night Band (DNB) Sensor Data Record (SDR) were provided by the National Oceanic and Atmospheric Administration (NOAA), which were acquired between 0000-0200 AM local time (UTC-6). The scenes were digitally enhanced to increase the visibility of dim light sources, while preserving detail in brighter areas. The analysis of S3-SLSTR data reveals that 5870.66 Ha of forest were affected during the wildfire event that took place in the Indio Maiz Biological Reserve, in April 2018. This value is close to the 5945 Ha reported by local authorities, compared to the 5033.74 Ha and 5187.29 Ha obtained from MODIS and Sentinel-2 data, respectively. This study also produced a nighttime-sensed time series of a fire event in the Central American region using the VIIRS DNB band. This VIIRS product was capable of detecting the actively flaming forest fire in the IMBR and keep track of it by means of capturing radiance in the visible spectrum. In conclusion, S3-SLSTR imagery is capable of detecting active fire events in remote areas. Despite its coarse resolution of 500 m, S3-SLSTR is a valuable for disaster management teams, who will definitely benefit from the fire-monitoring optimized images.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFMNH23E0875U
- Keywords:
-
- 1616 Climate variability;
- GLOBAL CHANGEDE: 1920 Emerging informatics technologies;
- INFORMATICSDE: 4313 Extreme events;
- NATURAL HAZARDSDE: 4341 Early warning systems;
- NATURAL HAZARDS