Optimization Of Optical And Radar Satellite Data In Google Earth Engine For Monitoring Oil Palm Changes In Tropical River Basins
Accurate mapping of oil palm plantations is crucial for planning agricultural best management practices. Google Earth Engine (GEE), a cloud-based computing platform, allowing users to process multi-source satellite images more quickly and effectively. In fact, it is difficult to distinguish oil palm...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/60889/ http://eprints.usm.my/60889/1/ZENG%20JU%20-%20TESIS24.pdf |
| _version_ | 1848884563825131520 |
|---|---|
| author | Zeng, Ju |
| author_facet | Zeng, Ju |
| author_sort | Zeng, Ju |
| building | USM Institutional Repository |
| collection | Online Access |
| description | Accurate mapping of oil palm plantations is crucial for planning agricultural best management practices. Google Earth Engine (GEE), a cloud-based computing platform, allowing users to process multi-source satellite images more quickly and effectively. In fact, it is difficult to distinguish oil palm from other crops using only optical satellites due to the issue of cloud cover in tropical regions. Unfortunately, there is only little scientific understanding about how various satellite images within GEE can be helpful for mapping oil palm plantations. Hence, this study aims to determine the optimal combination of open-source optical and radar satellite data for mapping oil palm plantations in tropical river basins using the Muda River Basin (MRB) and the Johor River Basin (JRB) as test sites. First, the two machine learning classifiers available in GEE, random forest (RF) and support vector machine (SVM), were compared to identify which is the most effective classifier for mapping oil palm plantations. Then, eight different data combinations have been constructed from the satellite images and indices such as C-band Sentinel-1, L-band PALSAR2, Landsat8, Sentinel-2, topographic, Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), etc. Lastly, the optimal data combination was employed to project future oil palm distribution using the CA-Markov approach. The findings demonstrate that RF outperformed SVM in mapping oil palm plantations in both river basins. |
| first_indexed | 2025-11-15T19:08:42Z |
| format | Thesis |
| id | usm-60889 |
| institution | Universiti Sains Malaysia |
| institution_category | Local University |
| language | English |
| last_indexed | 2025-11-15T19:08:42Z |
| publishDate | 2023 |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | usm-608892024-07-31T08:11:03Z http://eprints.usm.my/60889/ Optimization Of Optical And Radar Satellite Data In Google Earth Engine For Monitoring Oil Palm Changes In Tropical River Basins Zeng, Ju P1-1091 Philology. Linguistics(General) Accurate mapping of oil palm plantations is crucial for planning agricultural best management practices. Google Earth Engine (GEE), a cloud-based computing platform, allowing users to process multi-source satellite images more quickly and effectively. In fact, it is difficult to distinguish oil palm from other crops using only optical satellites due to the issue of cloud cover in tropical regions. Unfortunately, there is only little scientific understanding about how various satellite images within GEE can be helpful for mapping oil palm plantations. Hence, this study aims to determine the optimal combination of open-source optical and radar satellite data for mapping oil palm plantations in tropical river basins using the Muda River Basin (MRB) and the Johor River Basin (JRB) as test sites. First, the two machine learning classifiers available in GEE, random forest (RF) and support vector machine (SVM), were compared to identify which is the most effective classifier for mapping oil palm plantations. Then, eight different data combinations have been constructed from the satellite images and indices such as C-band Sentinel-1, L-band PALSAR2, Landsat8, Sentinel-2, topographic, Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), etc. Lastly, the optimal data combination was employed to project future oil palm distribution using the CA-Markov approach. The findings demonstrate that RF outperformed SVM in mapping oil palm plantations in both river basins. 2023-09 Thesis NonPeerReviewed application/pdf en http://eprints.usm.my/60889/1/ZENG%20JU%20-%20TESIS24.pdf Zeng, Ju (2023) Optimization Of Optical And Radar Satellite Data In Google Earth Engine For Monitoring Oil Palm Changes In Tropical River Basins. PhD thesis, Universiti Sains Malaysia. |
| spellingShingle | P1-1091 Philology. Linguistics(General) Zeng, Ju Optimization Of Optical And Radar Satellite Data In Google Earth Engine For Monitoring Oil Palm Changes In Tropical River Basins |
| title | Optimization Of Optical And Radar Satellite Data In Google Earth Engine For Monitoring Oil Palm Changes In Tropical River Basins |
| title_full | Optimization Of Optical And Radar Satellite Data In Google Earth Engine For Monitoring Oil Palm Changes In Tropical River Basins |
| title_fullStr | Optimization Of Optical And Radar Satellite Data In Google Earth Engine For Monitoring Oil Palm Changes In Tropical River Basins |
| title_full_unstemmed | Optimization Of Optical And Radar Satellite Data In Google Earth Engine For Monitoring Oil Palm Changes In Tropical River Basins |
| title_short | Optimization Of Optical And Radar Satellite Data In Google Earth Engine For Monitoring Oil Palm Changes In Tropical River Basins |
| title_sort | optimization of optical and radar satellite data in google earth engine for monitoring oil palm changes in tropical river basins |
| topic | P1-1091 Philology. Linguistics(General) |
| url | http://eprints.usm.my/60889/ http://eprints.usm.my/60889/1/ZENG%20JU%20-%20TESIS24.pdf |