Assessing swidden land use in Myanmar by decision tree-based detection method using landsat imagery
Abstract Swidden agriculture is a common land use found in the mountainous regions, especially in Southeast Asia. In Myanmar, the swidden agriculture has been practicing as an important livelihood strategy of millions of people, mainly by the ethnic groups. However, the extent of swidden agriculture...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
CABI
2022-10-01
|
| Series: | CABI Agriculture and Bioscience |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s43170-022-00132-4 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832572957846667264 |
|---|---|
| author | Nyein Chan Khin Nilar Swe Khin Thu Wint Kyaw La Minn Ko Ko Kyaw Win Nway Nway Aung Thet Oo Zwe Maung Maung Zar Zar Win Thein |
| author_facet | Nyein Chan Khin Nilar Swe Khin Thu Wint Kyaw La Minn Ko Ko Kyaw Win Nway Nway Aung Thet Oo Zwe Maung Maung Zar Zar Win Thein |
| author_sort | Nyein Chan |
| collection | DOAJ |
| description | Abstract Swidden agriculture is a common land use found in the mountainous regions, especially in Southeast Asia. In Myanmar, the swidden agriculture has been practicing as an important livelihood strategy of millions of people, mainly by the ethnic groups. However, the extent of swidden agriculture in Myanmar is still in question. Therefore, we attempted to detect swidden patches and estimate the swidden extent in Myanmar using free available Landsat images on Google Earth Engine in combination with a decision tree-based plot detection method. We applied the commonly used indices such as dNBR, RdNBR, and dNDVI, statistically tested their threshold values to select the most appropriate combination of the indices and thresholds for the detection of swidden, and assessed the accuracy of each set of index and thresholds using ground truth data and visual interpretation of sample points outside the test site. The results showed that dNBR together with RdNBR, slope and elevation demonstrated higher accuracy (84.25%) compared to an all-index combination (dNBR, RdNBR, dNDVI, slope, and elevation). Using the best-fit pair, we estimated the extent of swidden at national level. The resulting map showed that the total extent of swidden in Myanmar was about 0.1 million ha in 2016, which is much smaller than other previously reported figures. Also, swidden patches were mostly observed in Shan State, followed by Chin State. In this way, this study primarily estimated the total extent of swidden area in Myanmar at national level and proved that the use of a decision tree-based detection method with appropriate vegetation indices and thresholds is highly applicable to the estimation of swidden extent on a regional basis. Also, as Myanmar is the largest country in mainland Southeast Asia in area with a great majority of the population living in rural areas, and many in the mountains, its land resources are of great relevance to the people’s livelihoods and thereby the nation’s progress. Therefore, this study will contribute to sustainable land management planning on both regional and national scale. |
| format | Article |
| id | doaj-art-9298885beffc4bd9af776349c3ae1a0d |
| institution | Kabale University |
| issn | 2662-4044 |
| language | English |
| publishDate | 2022-10-01 |
| publisher | CABI |
| record_format | Article |
| series | CABI Agriculture and Bioscience |
| spelling | doaj-art-9298885beffc4bd9af776349c3ae1a0d2025-02-02T05:59:52ZengCABICABI Agriculture and Bioscience2662-40442022-10-013111810.1186/s43170-022-00132-4Assessing swidden land use in Myanmar by decision tree-based detection method using landsat imageryNyein Chan0Khin Nilar Swe1Khin Thu Wint Kyaw2La Minn Ko Ko3Kyaw WinNway Nway AungThet OoZwe Maung MaungZar Zar Win Thein4National Institute for Environmental StudiesLaboratory of Vehicle Robotics, Division of Fundamental AgriScience, Research Faculty of Agriculture, Hokkaido UniversityGraduate School of Bioresource and Bioenvironmental Sciences, Kyushu UniversityGraduate School of Science and Engineering, Saitama UniversityGraduate School of Asian and African Area Studies, Kyoto UniversityAbstract Swidden agriculture is a common land use found in the mountainous regions, especially in Southeast Asia. In Myanmar, the swidden agriculture has been practicing as an important livelihood strategy of millions of people, mainly by the ethnic groups. However, the extent of swidden agriculture in Myanmar is still in question. Therefore, we attempted to detect swidden patches and estimate the swidden extent in Myanmar using free available Landsat images on Google Earth Engine in combination with a decision tree-based plot detection method. We applied the commonly used indices such as dNBR, RdNBR, and dNDVI, statistically tested their threshold values to select the most appropriate combination of the indices and thresholds for the detection of swidden, and assessed the accuracy of each set of index and thresholds using ground truth data and visual interpretation of sample points outside the test site. The results showed that dNBR together with RdNBR, slope and elevation demonstrated higher accuracy (84.25%) compared to an all-index combination (dNBR, RdNBR, dNDVI, slope, and elevation). Using the best-fit pair, we estimated the extent of swidden at national level. The resulting map showed that the total extent of swidden in Myanmar was about 0.1 million ha in 2016, which is much smaller than other previously reported figures. Also, swidden patches were mostly observed in Shan State, followed by Chin State. In this way, this study primarily estimated the total extent of swidden area in Myanmar at national level and proved that the use of a decision tree-based detection method with appropriate vegetation indices and thresholds is highly applicable to the estimation of swidden extent on a regional basis. Also, as Myanmar is the largest country in mainland Southeast Asia in area with a great majority of the population living in rural areas, and many in the mountains, its land resources are of great relevance to the people’s livelihoods and thereby the nation’s progress. Therefore, this study will contribute to sustainable land management planning on both regional and national scale.https://doi.org/10.1186/s43170-022-00132-4Ethnic land useGoogle earth engineMyanmarSwiddenThreshold values |
| spellingShingle | Nyein Chan Khin Nilar Swe Khin Thu Wint Kyaw La Minn Ko Ko Kyaw Win Nway Nway Aung Thet Oo Zwe Maung Maung Zar Zar Win Thein Assessing swidden land use in Myanmar by decision tree-based detection method using landsat imagery CABI Agriculture and Bioscience Ethnic land use Google earth engine Myanmar Swidden Threshold values |
| title | Assessing swidden land use in Myanmar by decision tree-based detection method using landsat imagery |
| title_full | Assessing swidden land use in Myanmar by decision tree-based detection method using landsat imagery |
| title_fullStr | Assessing swidden land use in Myanmar by decision tree-based detection method using landsat imagery |
| title_full_unstemmed | Assessing swidden land use in Myanmar by decision tree-based detection method using landsat imagery |
| title_short | Assessing swidden land use in Myanmar by decision tree-based detection method using landsat imagery |
| title_sort | assessing swidden land use in myanmar by decision tree based detection method using landsat imagery |
| topic | Ethnic land use Google earth engine Myanmar Swidden Threshold values |
| url | https://doi.org/10.1186/s43170-022-00132-4 |
| work_keys_str_mv | AT nyeinchan assessingswiddenlanduseinmyanmarbydecisiontreebaseddetectionmethodusinglandsatimagery AT khinnilarswe assessingswiddenlanduseinmyanmarbydecisiontreebaseddetectionmethodusinglandsatimagery AT khinthuwintkyaw assessingswiddenlanduseinmyanmarbydecisiontreebaseddetectionmethodusinglandsatimagery AT laminnkoko assessingswiddenlanduseinmyanmarbydecisiontreebaseddetectionmethodusinglandsatimagery AT kyawwin assessingswiddenlanduseinmyanmarbydecisiontreebaseddetectionmethodusinglandsatimagery AT nwaynwayaung assessingswiddenlanduseinmyanmarbydecisiontreebaseddetectionmethodusinglandsatimagery AT thetoo assessingswiddenlanduseinmyanmarbydecisiontreebaseddetectionmethodusinglandsatimagery AT zwemaungmaung assessingswiddenlanduseinmyanmarbydecisiontreebaseddetectionmethodusinglandsatimagery AT zarzarwinthein assessingswiddenlanduseinmyanmarbydecisiontreebaseddetectionmethodusinglandsatimagery |