Exploring the utility of Sentinel-2 MSI derived spectral indices in mapping burned areas in different land-cover types
The difference in structure and composition of landcover types requires accurate mapping of burned areas for post-fire ecological assessments. Spectral indices for burned area mapping are mostly hard-coded to particular environments. However, the best post-fire spectral index to use for burned area...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2020-11-01
|
| Series: | Scientific African |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2468227620303033 |
| _version_ | 1818622054343114752 |
|---|---|
| author | Kudzai Shaun Mpakairi Henry Ndaimani Blessing Kavhu |
| author_facet | Kudzai Shaun Mpakairi Henry Ndaimani Blessing Kavhu |
| author_sort | Kudzai Shaun Mpakairi |
| collection | DOAJ |
| description | The difference in structure and composition of landcover types requires accurate mapping of burned areas for post-fire ecological assessments. Spectral indices for burned area mapping are mostly hard-coded to particular environments. However, the best post-fire spectral index to use for burned area mapping in most unstudied landcover types is not known. In this study, out of nine burned mapping indices optimised using the red-edge band, we tested which index outperformed the others in different land cover types. We used the Random Forest (RF) classifier to detect burned areas from Sentinel 2A imagery in four study sites and assessed the classification accuracy. We found out that, the Burned Area Index (BAI) and Global Environmental Monitoring Index (GEMI) spectral indices outperformed other indices in open shrublands, evergreen forest and in needle-leaved and semi-deciduous forests. The lowest performing spectral indices in the four study sites were Optimised Soil Adjusted Vegetation Index (OSAVI), Normalise Burn Ratio (NBR), and Normalise Difference Vegetation Index (NDVI). We recommend that for future studies, researchers and ecologists should use BAI and GEMI in mapping fires in open shrublands, evergreen, needle-leaved and semi-deciduous forests. Our results provide necessary insight for burned mapping algorithms and the accurate estimation of post-fire carbon emission with uni-temporal spectral indices in open shrublands, evergreen, needle-leaved and semi-deciduous forests. |
| first_indexed | 2024-12-16T18:19:04Z |
| format | Article |
| id | doaj.art-fa3ec76b211f488da83b4717c22f566a |
| institution | Directory Open Access Journal |
| issn | 2468-2276 |
| language | English |
| last_indexed | 2024-12-16T18:19:04Z |
| publishDate | 2020-11-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Scientific African |
| spelling | doaj.art-fa3ec76b211f488da83b4717c22f566a2022-12-21T22:21:36ZengElsevierScientific African2468-22762020-11-0110e00565Exploring the utility of Sentinel-2 MSI derived spectral indices in mapping burned areas in different land-cover typesKudzai Shaun Mpakairi0Henry Ndaimani1Blessing Kavhu2University of Zimbabwe, Department of Geography and Environmental Science, Box MP 167, Mount Pleasant, Harare, Zimbabwe; Corresponding author.University of Zimbabwe, Department of Geography and Environmental Science, Box MP 167, Mount Pleasant, Harare, ZimbabweZimbabwe Parks and Wildlife Management Authority Headquarters, P O Box CY140 Causeway, Borrowdale, Harare, Zimbabwe; Department of Geography and Environmental Studies, Stellenbosch University, Private Bag X1, Matieland 7602, South Africa; Centre for Complex Systems in Transition, Stellenbosch University, Stellenbosch 7600, South AfricaThe difference in structure and composition of landcover types requires accurate mapping of burned areas for post-fire ecological assessments. Spectral indices for burned area mapping are mostly hard-coded to particular environments. However, the best post-fire spectral index to use for burned area mapping in most unstudied landcover types is not known. In this study, out of nine burned mapping indices optimised using the red-edge band, we tested which index outperformed the others in different land cover types. We used the Random Forest (RF) classifier to detect burned areas from Sentinel 2A imagery in four study sites and assessed the classification accuracy. We found out that, the Burned Area Index (BAI) and Global Environmental Monitoring Index (GEMI) spectral indices outperformed other indices in open shrublands, evergreen forest and in needle-leaved and semi-deciduous forests. The lowest performing spectral indices in the four study sites were Optimised Soil Adjusted Vegetation Index (OSAVI), Normalise Burn Ratio (NBR), and Normalise Difference Vegetation Index (NDVI). We recommend that for future studies, researchers and ecologists should use BAI and GEMI in mapping fires in open shrublands, evergreen, needle-leaved and semi-deciduous forests. Our results provide necessary insight for burned mapping algorithms and the accurate estimation of post-fire carbon emission with uni-temporal spectral indices in open shrublands, evergreen, needle-leaved and semi-deciduous forests.http://www.sciencedirect.com/science/article/pii/S2468227620303033Random forestBAIGEMIClassification |
| spellingShingle | Kudzai Shaun Mpakairi Henry Ndaimani Blessing Kavhu Exploring the utility of Sentinel-2 MSI derived spectral indices in mapping burned areas in different land-cover types Scientific African Random forest BAI GEMI Classification |
| title | Exploring the utility of Sentinel-2 MSI derived spectral indices in mapping burned areas in different land-cover types |
| title_full | Exploring the utility of Sentinel-2 MSI derived spectral indices in mapping burned areas in different land-cover types |
| title_fullStr | Exploring the utility of Sentinel-2 MSI derived spectral indices in mapping burned areas in different land-cover types |
| title_full_unstemmed | Exploring the utility of Sentinel-2 MSI derived spectral indices in mapping burned areas in different land-cover types |
| title_short | Exploring the utility of Sentinel-2 MSI derived spectral indices in mapping burned areas in different land-cover types |
| title_sort | exploring the utility of sentinel 2 msi derived spectral indices in mapping burned areas in different land cover types |
| topic | Random forest BAI GEMI Classification |
| url | http://www.sciencedirect.com/science/article/pii/S2468227620303033 |
| work_keys_str_mv | AT kudzaishaunmpakairi exploringtheutilityofsentinel2msiderivedspectralindicesinmappingburnedareasindifferentlandcovertypes AT henryndaimani exploringtheutilityofsentinel2msiderivedspectralindicesinmappingburnedareasindifferentlandcovertypes AT blessingkavhu exploringtheutilityofsentinel2msiderivedspectralindicesinmappingburnedareasindifferentlandcovertypes |