Using Sentinel-2 and MODIS for Imagery Intelligence
Detect fire hotspots using thermal activity products and broadcasted news related to fire.
Introduction of Sentinel-2 use cases
Sentinel-2 SAR is a high-resolution, wide-swath imaging mission that supports Copernicus Land Monitoring investigations such as vegetation, soil, and water cover monitoring, as well as observation of inland waterways and coastal areas.
The Sentinel-2 Multispectral Instrument (MSI) samples 13 spectral bands with a spatial resolution of four bands of ten meters, six bands of twenty meters, and three bands at sixty meters. It can generate the geospatial information at local, regional, national, and worldwide scales thanks to the collected data, mission coverage, and high return frequency.
In 2012, the Copernicus land monitoring service became live, providing spatial data on land cover and factors relevant to subjects like plant status and the water cycle. It also helps with spatial planning, forest management, water management, agriculture, and food security.
The Copernicus Marine Environment Monitoring Service (CMEMS) offers frequent and systematic reference data on the physical state, variability, and dynamics of oceans and marine ecosystems throughout the world, as well as in European seas. Marine safety, marine resources, coastal and marine environment, weather, seasonal forecasts, and climate are among the application domains provided by CMEMS.
All organizations and entities involved in the management of natural disasters, man-made emergency situations, and humanitarian crises can benefit from the Copernicus emergency management service, which delivers accurate and timely geospatial information generated from satellite remote sensing. Open data sources are used to supplement this data wherever accessible. The Copernicus Emergency Management Service’s mapping component has global coverage and generates maps based on satellite images.
Introducing the fire detection use case
The Sentinel-2 SAR mission collects imagery data for forest monitoring, land cover change detection and natural disaster management. These data collections offer imagery analysis of the short-wave infrared spectrum. The electromagnetic spectrum allows methods for a better visualization of fire burns and scars.
The intensity of fire burns and scars directly shows the most important information about the fire hazard. Fire intensity is the energy generated from organic matter during combustion. It also refers to the fire’s intensity when it is burning. Burn severity describes how the intensity of the fire affects the ecosystem’s functioning in the burned area. The observed effects varied frequently between various ecosystems and within the same location. The burn severity defines the degree to which the fire has affected or disrupted a geographic region.
The Normalized Burn Ratio (NBR) is a measurement identifying burned areas in larger geographic regions. The formula is like the Normalized Difference Vegetation Index (NDVI), except it considers near infrared (NIR) and shortwave infrared (SWIR) wavelengths.
NBR uses the ratio between NIR and SWIR bands to benefit from the size of spectral difference. A high NBR score suggests healthy vegetation, whereas a low value shows barren land or recently burned regions. Non-burned geographic regions are usually have assigned values near zero.
The difference between the NBR gathered from the images before and after the fire is used to derive the severity of the fire. A higher difference value shows more severe damage, while areas with negative difference values may show recent regrowth after a fire. You can classify the difference values based on historical fire sources.
Inspecting the Sentinel-2 data near the Antonov Airport collected on 26 February show several fires caused by the Battle of Hostomel which began on 24 February during the Russian invasion of Ukraine.
Expectations from a geospatial intelligence analyst
As a geospatial intelligence analyst, I need feature extraction of fires areas and some confidence score how well the algorithm detected the fire areas by using continuous Sentinel-2 data. So that I can validate the automatic fire detection and focus on only a few tough spots.
A bunch of firebrands being geographically near to each other often lead to spotting fire bursts. There must be a kind of fire buffering and clustering for recognizing possible fire hotspots. So that I can easily use these fire cluster informations grouping spatial related fires.
From an engineer's perspective, I need to query fires as features by using a simple API. Our well-trained engineers are using Python-based data science environments. The API must offer some handy conversion from JSON into data frames.
Verified access to geospatial data
The world’s most comprehensive collection of verified geographic data of various themes is ArcGIS Living Atlas of the World. This living atlas has maps, geocentric web applications, and geospatial data layers to help any data scientist with common geospatial tasks.
ArcGIS Living Atlas of the World hosts curated preprocessed Sentinel-2 views. You can use these raster layers in a variety of business sectors, academic fields, and other geospatial focused areas. For our use case, these views allow highlighting of burned geographical areas and determining fire severity around the planet. The views contain historical raster Sentinel-2 data for at least 14 months and receive daily updates. So that your data science pipelines can use various spatial and temporal filtering of the underlying raster data.
Moderate Resolution Imaging Spectroradiometer (MODIS) is an important instrument viewing the entire surface of the Earth every one to two days. With its ability to image the entire surface, we can create observations describing the features of the land, ocean, and atmosphere. It is one of the many instruments solving the question of what is happening on our planet. However, most of those instruments have limitations, while the technology it uses is relatively mature.
Thermal activity products allow fire monitoring and smoke detection, and are used in a variety of applications, such as fire safety and fire protection. You can derive these compounds from the combustion of hydrocarbons. However, the cost of the fire products is high and they include several false positives.
The detection of anomalies using specific bands combining measured values of spectral intensity, surface temperature and sunlight reflection intensity is a complex task. Experts suggest alternative approaches for detecting surface-temperature anomalies, which is based on the possibility of simultaneous spectral analysis of three signals. The surface energy balance seems important for a correct interpretation of surface and interior temperatures, because you can use it both for temperature detection and for the evaluation of the radiation balance.
The fire products contain the fire location, the fire occurrence, and various specific fire attributes. These fire features represent detectable thermal activities for the last 48 hours and/or 7 days. Any data scientist can access this spatio-temporal dataset from ArcGIS Living Atlas of the World using the ArcGIS API for Python.
Summary
The amount of land burnt by fires across the European Union is over three times greater than the level of wildfires experts expected. Fire has threatened 5 billion hectares, and this seems to be an ongoing trend. A fire broke out in Berlin’s Grunewald forest early on Thursday, 28th August following an explosion at a munitions storage site, with the German capital still facing serious damages.
We need better geospatial analytics capabilities, train our geospatial skill set, use the existing cloud infrastructure and Open Data datasets more wisely. From an engineer’s perspective, we need more location services providing ready-to-use geospatial and intelligence know-how without being a geospatial intelligence expert.
References
[1] Living Atlas — Sentinel-2 Views
Sentinel-2, 10m Multi-spectral, Multi-temporal, 13-band images with visual renderings and indices
[2] Copernicus Marine Environment Monitoring Service (CMEMS)
Providing open marine data and services
[3] Copernicus in Germany (d-copernicus.de)
Copernicus landing page introducing the Sentinel-family
[4] Active Fire Data — NASA | LANCE | FIRMS
Fire information for Resource Management System
[5] Living Atlas — Satellite (MODIS) Thermal Hotspots and Fire Activity
Thermal activity detected by the MODIS sensors on the NASA Aqua and Terra satellites during the last 48 hours and 7 days.
