A patch filling method for thematic map refinement: a case study on forest cover mapping in the Greater Mekong Subregion and Malaysia
Accurate forest cover mapping is essential for monitoring the status of forest extent in Southeast Asia. However, tropical areas frequently experience cloud cover, resulting in invalid or missing data in thematic maps. The initial 2005 and 2010 forest cover maps produced by the collaboration of the...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Taylor & Francis Group
2023-12-01
|
| Series: | GIScience & Remote Sensing |
| Subjects: | |
| Online Access: | http://dx.doi.org/10.1080/15481603.2023.2252225 |
| _version_ | 1827811623246495744 |
|---|---|
| author | Shili Meng Yong Pang Kebiao Huang Zengyuan Li |
| author_facet | Shili Meng Yong Pang Kebiao Huang Zengyuan Li |
| author_sort | Shili Meng |
| collection | DOAJ |
| description | Accurate forest cover mapping is essential for monitoring the status of forest extent in Southeast Asia. However, tropical areas frequently experience cloud cover, resulting in invalid or missing data in thematic maps. The initial 2005 and 2010 forest cover maps produced by the collaboration of the Greater Mekong Subregion and Malaysia (GMS+) economies contain unclassified pixels in the areas affected by cloud or cloud shadow. To enhance the usability and effectiveness of the 2005 and 2010 GMS+ forest cover maps for further analysis and applications, we present a novel method for accurately mapping forest cover in the presence of cloud cover. We employed a pixel-based algorithm to create clear view composites and automatically generated land cover training labels from the existing forest cover maps. We then reclassified the invalid areas and produced updated maps. The land cover types for all previously missing pixels have been successfully reclassified. The accuracy of this method was assessed at both the pixel and region level, with an overall accuracy of 94.2% at the forest/non-forest level and 86.6% at the finer classification level by pixel level assessment across all reclassified patches, and 93.2% at the forest/non-forest level and 89.9% at the finer level by region level for the selected site. There are 2.6% of forest and 0.7% of non-forest areas in the 2005 map, as well as 2.7% of forest and 0.6% of non-forest in the 2010 map have been reclassified from invalid pixels. This approach provides a framework for filling invalid areas in the existing thematic map toward improving its spatial continuity. The updated outputs provide more accurate and reliable information than the initial maps on the status of forest extent in the GMS+, which is critical for effective forest management and sustainable use in the region. |
| first_indexed | 2024-03-11T23:08:15Z |
| format | Article |
| id | doaj.art-3d49703b593c43ac955bf440a77bf349 |
| institution | Directory Open Access Journal |
| issn | 1548-1603 1943-7226 |
| language | English |
| last_indexed | 2024-03-11T23:08:15Z |
| publishDate | 2023-12-01 |
| publisher | Taylor & Francis Group |
| record_format | Article |
| series | GIScience & Remote Sensing |
| spelling | doaj.art-3d49703b593c43ac955bf440a77bf3492023-09-21T12:43:10ZengTaylor & Francis GroupGIScience & Remote Sensing1548-16031943-72262023-12-0160110.1080/15481603.2023.22522252252225A patch filling method for thematic map refinement: a case study on forest cover mapping in the Greater Mekong Subregion and MalaysiaShili Meng0Yong Pang1Kebiao Huang2Zengyuan Li3Chinese Academy of ForestryChinese Academy of ForestryChinese Academy of ForestryChinese Academy of ForestryAccurate forest cover mapping is essential for monitoring the status of forest extent in Southeast Asia. However, tropical areas frequently experience cloud cover, resulting in invalid or missing data in thematic maps. The initial 2005 and 2010 forest cover maps produced by the collaboration of the Greater Mekong Subregion and Malaysia (GMS+) economies contain unclassified pixels in the areas affected by cloud or cloud shadow. To enhance the usability and effectiveness of the 2005 and 2010 GMS+ forest cover maps for further analysis and applications, we present a novel method for accurately mapping forest cover in the presence of cloud cover. We employed a pixel-based algorithm to create clear view composites and automatically generated land cover training labels from the existing forest cover maps. We then reclassified the invalid areas and produced updated maps. The land cover types for all previously missing pixels have been successfully reclassified. The accuracy of this method was assessed at both the pixel and region level, with an overall accuracy of 94.2% at the forest/non-forest level and 86.6% at the finer classification level by pixel level assessment across all reclassified patches, and 93.2% at the forest/non-forest level and 89.9% at the finer level by region level for the selected site. There are 2.6% of forest and 0.7% of non-forest areas in the 2005 map, as well as 2.7% of forest and 0.6% of non-forest in the 2010 map have been reclassified from invalid pixels. This approach provides a framework for filling invalid areas in the existing thematic map toward improving its spatial continuity. The updated outputs provide more accurate and reliable information than the initial maps on the status of forest extent in the GMS+, which is critical for effective forest management and sustainable use in the region.http://dx.doi.org/10.1080/15481603.2023.2252225forest cover mapgms+clear view compositesinvalid areasrandom forest |
| spellingShingle | Shili Meng Yong Pang Kebiao Huang Zengyuan Li A patch filling method for thematic map refinement: a case study on forest cover mapping in the Greater Mekong Subregion and Malaysia GIScience & Remote Sensing forest cover map gms+ clear view composites invalid areas random forest |
| title | A patch filling method for thematic map refinement: a case study on forest cover mapping in the Greater Mekong Subregion and Malaysia |
| title_full | A patch filling method for thematic map refinement: a case study on forest cover mapping in the Greater Mekong Subregion and Malaysia |
| title_fullStr | A patch filling method for thematic map refinement: a case study on forest cover mapping in the Greater Mekong Subregion and Malaysia |
| title_full_unstemmed | A patch filling method for thematic map refinement: a case study on forest cover mapping in the Greater Mekong Subregion and Malaysia |
| title_short | A patch filling method for thematic map refinement: a case study on forest cover mapping in the Greater Mekong Subregion and Malaysia |
| title_sort | patch filling method for thematic map refinement a case study on forest cover mapping in the greater mekong subregion and malaysia |
| topic | forest cover map gms+ clear view composites invalid areas random forest |
| url | http://dx.doi.org/10.1080/15481603.2023.2252225 |
| work_keys_str_mv | AT shilimeng apatchfillingmethodforthematicmaprefinementacasestudyonforestcovermappinginthegreatermekongsubregionandmalaysia AT yongpang apatchfillingmethodforthematicmaprefinementacasestudyonforestcovermappinginthegreatermekongsubregionandmalaysia AT kebiaohuang apatchfillingmethodforthematicmaprefinementacasestudyonforestcovermappinginthegreatermekongsubregionandmalaysia AT zengyuanli apatchfillingmethodforthematicmaprefinementacasestudyonforestcovermappinginthegreatermekongsubregionandmalaysia AT shilimeng patchfillingmethodforthematicmaprefinementacasestudyonforestcovermappinginthegreatermekongsubregionandmalaysia AT yongpang patchfillingmethodforthematicmaprefinementacasestudyonforestcovermappinginthegreatermekongsubregionandmalaysia AT kebiaohuang patchfillingmethodforthematicmaprefinementacasestudyonforestcovermappinginthegreatermekongsubregionandmalaysia AT zengyuanli patchfillingmethodforthematicmaprefinementacasestudyonforestcovermappinginthegreatermekongsubregionandmalaysia |