|
1.INTRODUCTIONWildfires are natural ecosystem processes that considerably disturb the functioning of ecosystems. Monitoring post-fire forest regrowth is crucial for receiving knowledge to help forest ecosystem recovery after fires. Remote aerospace methods and data provide effective tool in agriculture1-3 as well as in forestry – forecasting, monitoring, mapping and restoration of burnt areas. Aerospace remote sensing methods are a high-tech tool for reliable and large-scale monitoring of recovery processes occurring in forest ecosystems after fire4-7. Spectral indices and reflectance values are mainly used for vegetation processes following disturbances after fire8-10. Spectral indices generally rely on greenness measurements of red – near-infrared vegetation indices on the basis of different algebraic combinations between original spectral bands11. Vegetation indices help the study of forest ecosystems disturbance, but they are not accurate enough to study the regrowth processes in forest ecosystems observed after a fire. The differences in fire damage, caused by differences in undergrowth, species diversity, and the different regenerative abilities of forest with different tree species are the main reason for this. Disturbance Index (DI)12 was examined to be a relatively efficient approach to detect the forest disturbance and monitor its change. The higher accuracy of the index in comparison to standard vegetation indices is based on the linear orthogonal transformation of multispectral satellite images – Tasseled Cap Transformation (TCT)13-15 which increases the degree of identification of the three main components changing during fire – soil, vegetation and moisture16. Different sensors use various transformation matrices fixed only to them. For monitoring post-fire regrowth dynamics in this study we used transformation matrices fixed for Landsat sensors17,18. DI highlights the unvegetated spectral signatures associated with stand-replacing disturbance and separates them from all other forest signatures. When viewed sequentially, the time series of DI images provide a direct way to highlight pixels that changed from an average to a disturbed forest condition12. The Landsat data is widely used in forest monitoring19-23, but their applicability for detailed vegetation studies is limited, especially by their spatial and spectral resolution24. However, Landsat program offers a chance to study the long-term dynamics of forest ecosystems via long-term data series. 2.STUDY AREAMonitoring post-fire forest regrowth was performed on the territory of a burnt area near Chernyovtsi village, Bulgaria. The selection of the study area was based on several criteria: the burnt area should be large enough (> 5 ha) to allow mapping with Landsat data, the fire event should be before 2016 so there is enough time for regrowth processes to start, the presence of aerial images for visual interpretation and validation. Thus taking into account the above mentioned criteria and limitations the selected test fire is located next to Chernyovtsi village. The Chernyovtsi test fire (450 - 510 m above sea level) is situated in the northeastern part of Rhodope Mountains, near Chernyovtsi village, 15 km from the city of Kardzhali, Bulgaria (Figure 1). A fire took place on October 1, 2012 and affected an area of 15 ha with mixed forests and coniferous forests. Mixed forests consist of Turkey oak (Quercus cerris L.), Hungarian oak (Quercus frainetto T e n) and Oriental hornbeam (Carpinus orientalis M i l l.) with Mediterranean elements in places with secondary origin. The native deciduous forests in the region refer to the Thracian province of the European deciduous forest area. However, because of the erosion processes in the 1950s and the expansion of bare lands, a massive afforestation with coniferous forests – Black pine (Pinus nigra Arn.), has been performed. The area is characterized by Continental-Mediterranean climate. The soils are Chromic Cambisols. The forest ecosystems in this part of Bulgaria have been under stress in the summers during the last years due to frequent and prolonged droughts related to climate change25,26. The topography influenced the development of the wildfire as well. However, the forest ecosystems were not entirely damaged, since the fire occurred during the winter. The cold and humid conditions during the winter, the higher moisture content in the forest ecosystems determined the lower intensity of the fire, which affected the forest ecosystems less. 3.METHODS AND DATA3.1.Data acquisitionMonitoring the degree of disturbance and post-fire regrowth processes was performed on the territory of the test area for the study period – 2012-2022. The imagery acquisition was carried out taking into account the vegetation period of the forest ecosystems and the absence of clouds and cloud shadows over the study area. Landsat (ETM+, OLI and OLI-2) satellite imageries were used once per year in August. The satellite images from Landsat are freely available through the US Geological Survey’s online platform – Earth Explorer (https://earthexplorer.usgs.gov/)27. The dates of the satellite images used for the purpose of the post-fire regrowth monitoring and the sensor of which they were obtained are shown in Table 1. Table 1:Image acquisition dates
Aerial images28 with very high resolution (VHR) from 2013 (one year after the fire) were used for visual interpretation and test area selection as well as validation. Their spatial resolution is ≤ 0.4 m. The proposed approach using Differenced DI classification for post-fire regrowth monitoring was validated in a previous study with the help of a method involving the delineation of dynamic boundaries for spatial accuracy assessment. That previous study used VHR satellite data, including World View (2/3) and GeoEye (1) sensors for validation29. 3.2.Multispectral Image ProcessingThe Differenced DI calculated for vegetation regrowth dynamics are considered as classified raster thematic maps. The data processing of multispectral satellite images included basic operations such as georeferencing, subsetting, stacking multiband images, tasseled cap transformation, generating spectral indices – DI, dDI (described in Table 2). Table 2.Spectral indices used for classified raster thematic maps.
Generation of DI was based on TCT, applied on stacked multi-band images. After applying TCT, the results were multi-band images containing three layers – Wetness (TCW), Brightness (TCB), and Greenness (TCG). Normalization steps followed in order to normalize the radiometric change. Afterwards TCB, TCG and TCW were combined linearly to acquire DI14. The classified output rasters have a spatial resolution of 30 m. 4.RESULTS AND DISCUSSIONSFor the purpose of post-fire regrowth monitoring dDI rasters were generated on a yearly basis, indicating the areas and intensity of forest disturbance and regrowth after fire, actual at that certain year. The thematic raster classified by the intensity of recovery are compared with the values one month before the fire (2012), in quantitative values (%). The concept of DI assumes that high TCB and low TCG and TCW values are typical for disturbed stands and DI values are high positive, while undisturbed or fully recovered stands exhibit low TCB and high TCG and TCW values resulting in low negative DI values15. Figure 2 shows the dDI classified thematic raster for the study period – 2012-2022 on the territory of the burnt forest area. Figure 2 A) shows the post-fire disturbance map one year after the fire (2013). Two Landsat images were used – 09/03/2012 (Landsat ETM+) and 13/08/2013 (Landsat OLI) for representing the disturbance one year after the fire. The dark green color depicts areas with 0% disturbance – actually the unburnt forests. The light green color shows slightly affected forests – 0-20 %, the yellow color – forests with 20-50 % disturbance, the orange color – forests with 50-100 % disturbance and the red color – forests with 100-126 % disturbance. Figure 2 B) – G) show the post-fire regrowth maps representing the disturbance and regrowth for the post-fire monitoring (2 – 10 years after the fire) compared with the values from one year before the fire, in percentages. The green, yellow and orange colors depict regrowth, while the red colors indicate for disturbance. The percentages are shown in the legends (Fig. 2). After visual interpretation of the results obtained for post-fire disturbance and regrowth maps, some main conclusions are summarized as follows:
ACKNOWLEDGEMENTSThe authors thank to U.S. Geological Survey for providing free access to Landsat satellite data and to the Ministry of Agriculture of Republic of Bulgaria for the VHR aerial imageries. REFERENCESChanev, M., Filchev, L. and Valcheva, D.,
“Methodology for remote sensing monitoring of organic wheat crops,”
Ecological Engineering and Environment Protection, 2 56
–59
(2022). https://doi.org/10.32006/eeep Google Scholar
Ganeva, D., Chanev, M., Filchev, L., Jelev, G. and Valcheva, D.,
“Evaluation of Phenocam phenology of barley,”
in Proc. SPIE,
(2022). https://doi.org/10.1117/12.2636335 Google Scholar
Chanev, M., Filchev, L. and Ivanova, D.,
“Possibilities for the application of remote methods in the biological cultivation of cereals – a review,”
Journal of the Bulgarian Geographical Society, 43 31
–36
(2021). https://doi.org/10.3897/jbgs.2020.43.5 Google Scholar
Stankova, N. and Nedkov, R.,
“Monitoring forest regrowth with different burn severity using aerial and Landsat data,”
Geoscience and Remote Sensing Symposium (IGARSS), IEEE International, 2166
–2169
(2015). https://doi.org/10.1109/IGARSS.2015.7326233 Google Scholar
Stankova, N., Nedkov, R., Ivanova, I. and Avetisyan, D.,
“Modeling of forest ecosystems recovery after fire based on orthogonalization of multispectral satellite data,”
in Proc. SPIE,
(2018). https://doi.org/10.1117/12.2325643 Google Scholar
Stankova, N., Nedkov, R., Ivanova, I. and Avetisyan, D.,
“Integration of multispectral and SAR data for monitoring forest ecosystems recovery after fire,”
in Proc. SPIE,
(2017). https://doi.org/10.1117/12.2277313 Google Scholar
Avetisyan, D., Velizarova, E. and Filchev, L.,
“Post-Fire Forest Vegetation State Monitoring through Satellite Remote Sensing and In Situ Data,”
Remote Sens., 14
(24), 6266
(2022). https://doi.org/10.3390/rs14246266 Google Scholar
Gikov, A. and Dimitrov, P.,
“Mapping of burned area and assessment of burn severity of the 2017 wild fires in Kresna gorge,”
Journal of the Bulgarian Geographical Society, 40 10
–16
(2019). https://doi.org/10.3897/jbgs.2019.40.2 Google Scholar
Nedkov, R., Velizarova, E., Avetisyan, D. and Georgiev, N.,
“Assessment of forest vegetation state through remote sensing in response to fire impact,”
in Proc. SPIE,
(2020). https://doi.org/10.1117/12.2570808 Google Scholar
Cha, S. and Lim, J.,
“Deep Learning-based Forest Fire Classification Evaluation for Application of CAS500-4,”
Korean Journal of Remote Sensing, 38
(6), 1273
–1283
(2023). https://doi.org/10.7780/kjrs.2022.38.6.1.22 Google Scholar
Pérez-Cabello, F., Montorio, R. and Borini Alves, D.,
“Remote sensing techniques to assess post-fire vegetation recovery,”
Current Opinion in Environmental Science & Health, 21 100251
(2021). https://doi.org/10.1016/j.coesh.2021.100251 Google Scholar
Healey, S., Cohen, W., Yang, Z. and Krankina, O.,
“Comparison of Tasseled Cap-based Landsat data structures for use in forest disturbance detection,”
Remote Sensing of Environment, 97 301
–310
(2005). https://doi.org/10.1016/j.rse.2005.05.009 Google Scholar
Kauth, R. and Thomas, G.,
“The Tasseled Cap – a graphic description of the spectral – temporal development of agricultural crops as seen by Landsat,”
in Proceedings second ann. symp. machine processing of remotely sensed data. West Lafayette’ Purdue University Lab. App. Remote Sensing,
(1976). Google Scholar
Crist, E. and Cicone, R.,
“A physicaly-based transformation of Thematic Mapper data – the TM Tasseled Cap,”
IEEE Transactions on Geoscience and Remote Sensing, 22 256
–263
(1984). https://doi.org/10.1109/TGRS.1984.350619 Google Scholar
Stoyanov, A.,
“Application of Tasseled Cap Transformation of Sentinel-2—MSI Data for Forest Monitoring and Change Detection on Territory of Natural Park “BLUE STONES,”
in Environ. Sci. Proc.,
(2022). https://doi.org/10.3390/IECF2022-13073 Google Scholar
Crist, E. and Kauth, R.,
“The Tasseled Cap de-mystified,”
Photogrammetric Engineering and Remote Sensing, 52 81
–86
(1986). Google Scholar
Huang, C., Wylie, B., Homer, C. and Zylstra, G.,
“Derivation of a tasseled cap transformation based on Landsat 7 at-satellite reflectance,”
International Journal of Remote Sensing, 23
(8), 1741
–1748
(2002). https://doi.org/10.1080/01431160110106113 Google Scholar
Baig, M., Zhang, L., Shuai, T. and Tong, Q.,
“Derivation of a tasseled cap transformation based on Landsat 8 at-satellite reflectance,”
Remote Sensing Letters, 5
(5), 423
–431
(2014). https://doi.org/10.1080/2150704X.2014.915434 Google Scholar
Dimitrova, M., Gochev, D. and Trenchev, P.,
“Detection of Forest Vegetation Types by Differences in Seasonal Vegetation Change,”
in Proceedings SES2017,
195
–200
(2017). Google Scholar
Dimitrova, M., Gochev, D. and Trenchev, P.,
“Seasonal Changes of the Reflective Characteristics, NDVI and NDWI Indexes for Grass Vegetation, Coniferous and Broadleaved Forest, Obtained from the TM and OLI Data,”
in Proceedings SES2016,
224
–230
(2017). Google Scholar
Pickell, P.D., Hermosilla, T., Frazier, R.J., Coops, N.J. and Wulder, M.A.,
“Forest recovery trends derived from Landsat time series for North American boreal forests,”
International journal of remote sensing, 37
(1), 138
–149
(2016). https://doi.org/10.1080/2150704X.2015.1126375 Google Scholar
Masek, J. G., Huang, C., Wolfe, R., Cohen, W., Hall, F., Kutler, J. and Nelson, P.,
“North American forest disturbance mapped from a decadal Landsat record,”
Remote Sensing of Environment, 112 2914
–2926
(2008). https://doi.org/10.1016/j.rse.2008.02.010 Google Scholar
Dimitrova, M., Gochev, D. and Trenchev, P.,
“Reflective Characteristics, NDVI and NDWI Indexes for a Coniferous Forest and Changing Grass Vegetation, Obtained from the TM, ETM+ and OLI Data,”
in Proceedings SES2016,
217
–223
(2017). Google Scholar
Mišurec, J., Kopacková, V., Lhotáková, Z., Campbell, P. and Albrechtová, J.,
“Detection of Spatio-Temporal Changes of Norway Spruce Forest Stands in Ore Mountains Using Landsat Time Series and Airborne Hyperspectral Imagery,”
Remote Sens., 8
(92),
(2016). https://doi.org/10.3390/rs8020092 Google Scholar
Avetisyan, D., Velizarova, E., Nedkov, R. and Borisova, D,
“Assessment and mapping of the current state of the landscapes in Haskovo region (Southeastern Bulgaria) in relation to ecosystem services using remote sensing and GIS,”
in Proc. SPIE,
(2018). https://doi.org/10.1117/12.2325894 Google Scholar
Avetisyan, D. and Nedkov, R.,
“Application of remote sensing and GIS for determination of predicted status of the ecosystem/landscape services in changing environmental conditions,”
in Proc. SPIE,
(2009). https://doi.org/10.1117/12.2532609 Google Scholar
. Available online:USGS. Earth Explorer,
(2022) https://earthexplorer.usgs.gov/ October ). 2022). Google Scholar
. Available online:Executive Environment Agency,
(2021) http://pdbase.government.bg/zpo/en/index.jsp May ). 2021). Google Scholar
Stankova, N. and Avetisyan D.,
“Monitoring the condition and recovery processes after three forest fires in Bulgaria using remote aerospace methods,”
Geograph,
(2022). Google Scholar
|