Thanks to the combination of satellite images, the burned areas in the reserve of the Roman coast of Castel Fusano on 17 and 21 July were mapped in detail, both for the purpose of assessing the severity of the damage and for future monitoring of vegetation regrowth. To develop a large-scale information database, a team of researchers from INGV
On 17 July a great fire broke out in the Castel Fusano nature reserve (Roman coast) destroying a large area of pine forest. On 21 July a second fire hit the reserve again with significant damage to the vegetation (about 200 hectares of pine forest and Mediterranean scrub). From the combination of satellite images (Sentinel2A-MSI and the American twin Landsat 8-OLI) the researchers of the National Institute of Geophysics and Volcanology (INGV) have identified the burned areas, in their temporal succession, taking a picture of the damaged area and outlining the perimeter, the state of degradation and laying the foundations for future monitoring activities of vegetation regrowth.
Italy is one of the Mediterranean countries with the greatest fire risk. From mid-June until 27 July 2017, 74.965 hectares of wooded areas went up in smoke, equal to 156,41% of the total area burned in all of 2016 (data from Lega Ambiente).
"For the investigation", explains Stefania Amici, a researcher at INGV, Sentinel 2 multispectral satellite images (at 20m spatial resolution), made available within the European Copernicus programme, were used to characterize the damaged area and delineate it the perimeter".
INGV has been carrying out research for the study of fires for years, using sensors at various resolutions (spatial/spectral) and satellite data both to characterize the active flame phase and to evaluate the impact in terms of level of degradation and change of land use.
“The techniques used and applied to sensors such as Sentinel 2 or next generation sensors such as PRISMA (hyperspectral mission) of the Italian Space Agency (ASI) represent a large-scale and long-lasting information support to complete often fragmented and localized information, even temporally, and with great resource-saving potential. For example, when there are numerous fires, as in the case of this summer, the characterization of the perimeters, with satellite data, can be provided to administrations for various applications, including support for recovery plans. After the fire of 21 July, images from the Landsat 8 satellite with a spatial resolution of 30m were also used to outline the perimeter of the burnt area in Castel Fusano”, continues the INGV researcher.
To assess the extent of the damage, locate the areas at risk and support the damage mitigation strategies with targeted maps, the following parameters are taken into consideration: the burnt area spectral index (Normalized Burned Radio-NBR), the vegetation index and aridity index. The results obtained from the synergistic use of the two satellites (Sentinel2A-MSI and Landsat 8-OLI) are shown in Figure 3.
“The work, still in progress, in addition to delimiting the area affected by fires, aims to use satellite data to estimate the degree of severity of the damage. For this purpose, in situ measurements are used to validate the correspondence with the values derived from the map. Sensors such as Sentinel 2A and its twin Sentinel 2B, launched last March 7, in synergy with American sensors, will give us the possibility of aggregating information over time and characterizing multiple fires, thanks to a frequency of images every 5 days". concludes Stefania Amici.
Extended
Italy is one of the Mediterranean country with highest open fire risk. On 2017 in the period of time ranging between mid June and 27 July, 74.965 hectares of woodlands burned according to the Lega Ambiente report, representing 156,41% of the total burned area on 2016.
On 17 July 2017 a wildfire dangerously affected the natural reserve of Castel Fusano destroying a huge area covered by valuable maritime pine wood. At INGV we have used relatively new Sentinel 2A-MSI satellite data from the Copernicus program to characterize the burned area with an improved resolution of 20m/px. We have used the difference of normalized burn ratio (dNBR) technique to delineate the burned perimeter and to compare it with that one derived by Landsat 8-OLI satellite at 30 m resolution.
The promising results encourage further investigation aimed at extract burn severity and to map the evolution of the state of vegetation in the next months.
Figure 1 - NBR difference index derived from Sentinel2A-MSI images before (June 20, 2017) and after (July 20, 2017) the fire that started on July 17, 2017. Black-grey indicates no change; shades of very light gray towards white indicate change. The perimeter was surveyed by photo-interpretation (Contains modified Copernicus Sentinel data [2017]; Landsat 8-OLI courtesy of US Geological Survey; Software used SNAP-ESA, data processing Stefania Amici-INGV)
Figure 2 - NBR difference index calculated using two Landsat 8 -OLI images, one before (6 July 2017) and one (22 July 2017) after the fire of 17 and 21 July. Black-gray indicates no change; very light gray tones towards white indicate the change. The image highlights the area affected by the two fires. The perimeter was detected by photo-interpretation. (Contains modified Copernicus Sentinel data [2017]; Landsat 8-OLI courtesy of the US Geological Survey, Software used HARRIS-ENVI, data processing Stefania Amici-INGV)
Figure 3 - Sentinel 2A - MSI false color image in the SWIR, NIR, and Red edge bands with 20m/px resolution acquired by Sentinel 2A–MSI three days after the fire that started on July 17th. The white line indicates the perimeter derived from Sentinel 2A-dNBR while the red line indicates the perimeter derived using Landsat 8–dNBR at a resolution of 30m. The latter also highlights the area affected by the fire of July 21, 2017.
[credits Contains modified Copernicus Sentinel data [2017]; Landsat 8-OLI courtesy of the US Geological Survey, HARRIS-ENVI, data processing Stefania Amici-INGV]