The Fire Severity Analyst App is a Google Earth Engine (GEE) based tool that allows the generation of time series plots to assess fire severity and recovery. It allows the creation of these graphics using specific burned areas (selected from known sources) or user-specified areas of interest.
The app currently uses two spectral indices to assess fire severity and recovery:
- NBR – Normalized Burn Ratio and,
- NDVI – Normalized Difference Vegetation Index,
Five satellites/sensors are currently available:
- Terra/MODIS (MOD09A1, 500m, 8-day composites),
- Landsat 5/TM (SR/L2A, 30m, 16-day)
- Landsat 7/ETM+ (SR/L2A, 30m, 16-day)
- Landsat-8/OLI (SR/L2A, 30m, 16-day)
- Sentinel-2/MSI (SR/L2A, 20m, 5-day).
Two burned area datasets are currently available for mainland Portugal to check and evaluate fire scars. These are:
- EFFIS – European Forest Fire Information System (from the year 2000 to 2022), and,
- ICNF – burned areas (from the Instituto da Conservação da Natureza e Florestas, from the year 2000 to 2021)
A brief tutorial
Use the left panel and option to set up the data source of burned areas and the year. Press “+ Add layer” to insert the annual dataset. In the top panel/option “Layers” option, you can activate/inactivate each layer.
In the left panel, select the following options:
- Satellite/sensor – select one available EO platform for the analysis. These are MODIS, Landsat-5,6,7,8 or Sentinel-2;
- Spectral index – the spectral index that will be used for assessment. Currently, only NBR and NDVI are available.
- Start and end dates – depending on the year/date of the fire event, the user can customize the temporal window for the analysis. Note: very large periods may take longer to process, especially when large areas or complex polygons are selected. Eventually, this may cause GEE to stop or throw out-of-memory errors. In these cases, consider decreasing the time window and the area.
In the right-side panel, the user can select the geometry that will be used to perform spatiotemporal analyses on GEE. Available options for geometry inputs are:
- User-defined shape/drawing – this can be either a square, polygon or point;
- Specific longitude/latitude coordinates – these must be inserted in decimal degrees plus an optional radius for making a circular buffer (in meters), or,
- Selected burned area – after adding an annual burned area layer, the user can select one polygon for inspection and for plotting satellite data. Use the ‘crosshair’ cursor to select the area on the map. The polygon will appear with a yellow colour highlight, and info will be shown at the top about the selection (i.e., polygon ID, area and fire event date, if this info exists, which may not be the case for some ICNF polygons).
The “Selected input” checkbox is automatically turned on/off depending on the last input selected by the user. Note: in some cases, after selecting a burned area from the ICNF or EFFIS datasets, this may not activate automatically. In that instance, confirm if the option ‘Selected burned area‘ is active.
After selecting all the inputs, plot the time series data for the inputted geometry using the ” Plot time series data” button. In the right-side panel, a plot will appear after a few seconds. Keep in mind that large/complex geometries or long analysis windows may take longer.
After the plot is displayed on the panel, a secondary interface appears below it, enabling the user to generate a second graph with pre- and post-fire deltas (i.e., differences between pre and post-fire conditions).
This analysis is based on multi-temporal average composites of satellite images, which size (in days) is selected by the user.
The date of the fire must be specified in yyyy-mm-dd (year-month-day) format. This date can be retrieved from the ICNF or EFFIS databases or obtained from inspecting the plot for breaks in the time series. Abrupt breaks will typically be associated with fire.
Use the slider to select the number of days of the temporal window. A minimum of 30 days and a maximum of 365 days are allowed. After selecting the window value, the plot will be generated automatically.
This second plot depicts each delta as (example for NBR):
(p – each analysis period of size equal to n days, t – total number of periods from the fire till the present)
The x-axis of this second plot shows the days from the fire date, and the y-axis the deltas based on window averages.
Export each graph by clicking on the top-right button.