Short communicationEstimating spectral separability of satellite derived parameters for burned areas mapping in the Calabria region by using SPOT-Vegetation data
Introduction
Forest fires are one of the most critical issues in global change. Yearly, they affect vast areas throughout the world and cause devastating damages, such as, loss of biodiversity, decrease in forests, alteration of landscape, soil degradation, increase in greenhouse, etc. In the Mediterranean regions, fires are considered a major cause of land degradation. Every year, around 45,000 forest fires break out in the Mediterranean basin causing the destruction of about 2.6 million hectares (FAO, 2001). Several studies (Malanson, 1987, Trabaud and Lepart, 1981, Vila et al., 2001) dealing with the effects of fires on the vegetation within the Mediterranean basin found that fires induce significant alterations in short as well as long-term vegetation dynamics. Fires lead permanent changes in the composition of vegetation community, cause decrease in forests, loss of biodiversity, soil degradation, alteration of landscape patterns and ecosystem functioning thus speeding desertification processes up. Moreover, recent studies found that fires facilitate alien plant invasion, patch homogenization and create positive feedbacks in future fire susceptibility, fuel loading, fire spreading and intensity (Cochrane et al., 1999, Rahman and Dedieu, 1994, Malamud et al., 1998).
In Italy, as in other countries of the Mediterranean Basin, a small number of fires generally destroy a large percentage of the total burned areas every year. In these cases, the use of coarse resolution satellite sensors appears to be very useful for the discrimination of burned areas.
In this study, SPOT-Vegetation (SPOT-VGT) data at full spatial resolution were analysed in order to investigate the spectral characteristics of burned areas observed in the Mediterranean ecosystems of the Calabria Region (Southern Italy) during the 1998 fire season. VGT data have been successfully used for burnt area detection in boreal forests (Fraser and Li, 2002), Australia (Stroppiana et al., 2002a, Stroppiana et al., 2002b). This study represents one of the first examples of application of VGT data for burned land mapping in the Mediterranean ecosystems. These investigations can be very useful in helping the evaluation of the feasibility of SPOT VGT data for burned area discrimination and also for improving the capability of algorithms developed for burned area mapping. Satellite-based burned area mapping can be suitably used for comparing performance and reliability of fire spreading simulator and fire behaviour models which are useful tools to assist fire management (Miller and Yool, 2002). Currently, the inclusion of disturbance and other extreme events in coarse scale dynamic model is still in its infancy (Keane et al., 2004). An accurate mapping of fire-affected areas can support landscape simulation model for investigating the behavior of ecological systems and for evaluating modelling assumption (see, for example Wimberly, 2004; Boychuk et al., 1997). Fire disturbance regime may be crucial in determining the demographic characteristics of landscape-level dynamics. Satellite-based information can provide an invaluable support for evaluating fire disturbance patterns on a forest landscape, for testing the reliability of spatial simulation modes for fire and vegetation dynamics (see, for example Keane et al., 2004). Satellite-based mapping of burned area can provide a foundation for comparing models and to help identify the appropriate fire and vegetation processes and their simulation to include in coarse scale dynamic global vegetation models.
Section snippets
Study area
The study was carried out in the Calabria Region (South of Italy, see Fig. 1) selected because it is classified as high-risk areas on a global scale (FAO, 2001) for the high number of fires, which generally occur during the dry season from July to September. As an example, Fig. 2 shows hot spot and smoke plumes rising from some large fires that were active when the satellite advanced very Hhigh resolution radiometer (AVHRR) images were acquired on the 3rd of July, 1998.
The study area has a
Data analysis
SPOT-Vegetation (SPOT-VGT) data at full spatial resolution were analysed in order to investigate the spectral characteristics of burned areas observed in the Italian Peninsula. The investigations were performed on wildland fires that occurred during the 1998 fire season. Among the total fire events occurred in the considered period forest fires larger than 1000 ha were selected for this study. SPOT-VGT imagery acquired before and after fire events were analysed. Single channels or spectral
Results and discussion
Table 3 shows the SI indices for the considered fires. Significant SI values (higher than 1) are in bold. Fig. 2a and b, show the differences observed in each index in the VGT images acquired previous (III decade of June, 1998) and after fire occurrence (II decade of July, 1998) for two significant fire events. These differences were normalized to the values observed for each index in the VGT image acquired previous to fire events. A visual inspection of Fig. 2 clearly shows that the highest
Conclusion
Characterization of spectral signatures observed for fire scars in the Mediterranean ecosystems of the Calabria Region (Southern Italy) was performed by using 10-day composition of SPOT-VGT data. The analysis of burned area spectral signatures showed that among the spectral indices considered in this work, the NDII, SAVI, GEMI, BAI and NIR are the most appropriate variables for designing a burnt area algorithm due to their sensitivity to changes induced by the fire on the vegetation cover.
Acknowledgments
The author wishes to thank D. Pontani and G. Saragosa of the Italian National Forestry Service for providing fire data archives, which made this study possible. This work has been funded by the European Union and Basilicata Region within the framework of the project “Sviluppo dei distretti industriali di osservazioni della Terra”, Finanziamento MIUR, 2002–2005.
References (20)
- et al.
Modelling the effect of spatial scale and correlated fire disturbances on forest age distribution
Ecol. Modell.
(1997) - et al.
Estimating fire-related parameters in boreal forest using SPOT VEGETATION
Rem. Sens. Environ.
(2002) A soil-adjusted vegetation index (SAVI)
Rem. Sens. Environ.
(1988)- et al.
Detection of changes in leaf water content using near- and middle infrared reflectances
Rem. Sens. Environ.
(1989) - et al.
A classification of landscape fire succession models: spatial simulations of fire and vegetation dynamics
Ecol. Modell.
(2004) - et al.
Modeling fire in semi-desert grassland/oak woodland: the spatial implications
Ecol. Modell.
(2002) - et al.
Radiometric analysis of SPOT-VEGETATION images for burnt area detection in Northern Australia
Rem. Sens. Environ.
(2002) - et al.
Positive fire-grass feedback in Mediterranean basin shrub land
Forest Ecol. Manage.
(2001) Fire and forest landscapes in the Georgia Piedmont:an assessment of spatial modelling assumptions
Ecol. Modell.
(2004)- et al.
Assessment of different spectral indices in the red-near-infrared spectral domain for burned land discrimination
Int. J. Rem. Sens.
(2002)
Cited by (62)
Response of active layer thickening to wildfire in the pan-Arctic region: Permafrost type and vegetation type influences
2023, Science of the Total EnvironmentMapping fire-impacted refugee camps using the integration of field data and remote sensing approaches
2022, International Journal of Applied Earth Observation and GeoinformationCitation Excerpt :The MSI consists of 13 spectral channels ranging from 0.433 µm to 2.19 µm, with four 10 m visible and near-infrared bands (B2, B3, B4, and B8), six 20 m bands of red edge, near-infrared (NIR), and short wave infrared (SWIR) wavelengths (i.e. B5, B6, B7, B8a, B11, and B12), and three 60 m bands (B1, B9, and B10). Visible wavelengths (0.490 µm to 0.665 µm) are generally not useful for burned area detection as they are more sensitive to atmospheric contamination than longer wavelengths (Huang et al., 2016; Lasaponara, 2006). In addition, they are susceptible to fire smoke aerosols that are difficult to atmospherically correct.
Informational analysis of MODIS NDVI and EVI time series of sites affected and unaffected by wildfires
2022, Physica A: Statistical Mechanics and its ApplicationsShifting cultivation induced burn area dynamics using ensemble approach in Northeast India
2022, Trees, Forests and PeopleCitation Excerpt :Escuin et al. (2008) used a threshold-based method, where they employed multiple spectral indices, such as NDVI, NBR, and dNBR, for deforestation monitoring. Lasaponara (2006) employed NIR spectral band with multiple indices as NDVI, normalized difference of infrared index (NDII), burned area index (BAI), global environmental monitoring index (GEMI), and soil adjusted vegetation index (SAVI), and evaluated the performance of these indices in discriminating the land cover classes and fire-affected areas. Boschetti et al. (2010) also employed multiple spectral indices, such as NBR, BAI, char soil index (CSI), SAVI, and mid-infrared burn index (MIRBI), and applied the fuzzy classification technique for burned area mapping.
Spectral signature analysis of false positive burned area detection from agricultural harvests using Sentinel-2 data
2021, International Journal of Applied Earth Observation and Geoinformation