HOTSPOT service specifications

This service performs a hotspot detection algorithm. It takes as input a single calibrated optical dataset containing nir and swir22 spectral bands to generate a hotspot bitmask, representing areas of active fire.

The tutorial of the HOTSPOT service is available in this section.

Service Description

The Hotspot Detection (HOTSPOT) service employs an hotspot algorithm that requires in input a Calibrated Dataset having at least nir and swir22 assets (e.g. for Sentinel-2 MSI those are respectively the 8A and 12 bands). It calculates the ratio and the difference between those assets in TOA reflectance:

Pixels over water are masked and then excluded from the hotspot detection in case of TOA reflectance for swir22 asset is minor than 0.04. Hotspot and candidate hotspots are then identified using the following expressions:

The algorithm performs a background characterization around candidate hotspot pixels in case of a large fire event. The aggregated hotspot size is increased if the number of candidate fire pixels is less than the half of the total pixels in the window (approximately 2x2 km). Statistics (mean and standard deviation of the ratio) are also computed for pixels within the windows considered in the background characterization. In a later stage, the algorithm undertakes contextual tests which are needed to decide for all candidate hotspot pixels if they can be classified as hotspot or not.

In this process, two conditions needs to be satisfied to classify a candidate pixel as hotspot:

1. ratio > ratio_mean + max((3 * ratio_std), 0.5),

2. swir22 > swir22_mean + max((3 * swir22_std), 0.05).

Finally, the algorithm returns the hotspot classification of the entire scene as a binary mask (No-Fire=0, Hotspot=1). Output binary classification of active fires is given as single band raster in COG format.

Input

This processor uses as input a calibrated product from the Optical Products having swir22 and nir STAC assets. HOTSPOT supports Calibrated Datasets from the following EO missions:

• Sentinel-2,

• Landsat-8, Landsat-9.

Parameters

The HOTSPOT service requires a specified number of mandatory and optional parameters. All service parameters are listed in the table below.

Table 1 - Service parameters of the HOTSPOT processor.

Optical calibrated dataset

The first parameter represents the reference to the input optical calibrated dataset that is used to detect hotspots.

AOI (optional)

This second parameter (optional) may define the area of interest expressed as a Well-Known Text value.

Output

The HOTSPOT processor generates the following products:

1. Hotspot binary mask (1=true=fire, 0=false=not-fire),

2. Hotspot overview (true values in red with transparency set for false values).

HOTSPOT Product Specifications can be found in the below tables:

Vectorize HOTSPOT single band asset

HOTSPOT's binary mask can be converted to polygons by using the FilterVectorize service in Vectorize mode and selecting only true values DN=1 (hotspot).