- Library Home /
- Search Collections /
- Open Collections /
- Browse Collections /
- UBC Research Data /
- Remote Sensing Seagrasses in Cambodia: Comparing Sentinel-2...
Open Collections
UBC Research Data
Remote Sensing Seagrasses in Cambodia: Comparing Sentinel-2 and Planetscope for detecting and monitoring seagrasses in Cambodia’s Koh Rong and Koh Sdach Archipelagos Tan, Nigel
Description
Tropical seagrass is an integral part of the world’s fisheries as it acts as a nursery for benthic reproduction, leading to it being the center of many conservation efforts. Monitoring seagrass by high resolution satellite is commonplace, but such data quality is often unavailable for many conservation groups, leading to a question of whether open-source satellite data is sufficient to meet the rigorous needs of seagrass patch detection. 10m images from the European Space Agency’s (ESA) Sentinel-2 and 3m images from Planet Labs’ Planetscope covering the Koh Sdach and Koh Rong Archipelagos in Cambodia were collected, processed, and classified, with outputs being compared for seagrass detection ability and suitability. Image scenes were merged and processed to produce depth invariant indexes using the Lyzenga algorithm. Random forest classification with ground truth data yielded accuracies of 79.5% for Sentinel and 79.7% for Planetscope using unprocessed rasters, while depth invariant rasters resulted in accuracies of 80.8% and 70.0% respectively. Sentinel-2 and Planetscope are qualitatively compared based on technical specifications and factors such as band wavelength, radiometric quality, and revisit time. Sentinel-2 shows that it is able to match the seagrass detection ability of the higher resolution Planetscope due to its additional blue band and deeper radiometric depth. Results of this analysis are placed in the context of wider research on detecting seagrass patches through satellite image classification.
Item Metadata
Title |
Remote Sensing Seagrasses in Cambodia: Comparing Sentinel-2 and Planetscope for detecting and monitoring seagrasses in Cambodia’s Koh Rong and Koh Sdach Archipelagos
|
Creator | |
Contributor | |
Date Issued |
2021-04-12
|
Description |
Tropical seagrass is an integral part of the world’s fisheries as it acts as a nursery for benthic reproduction, leading to it being the center of many conservation efforts. Monitoring seagrass by high resolution satellite is commonplace, but such data quality is often unavailable for many conservation groups, leading to a question of whether open-source satellite data is sufficient to meet the rigorous needs of seagrass patch detection. 10m images from the European Space Agency’s (ESA) Sentinel-2 and 3m images from Planet Labs’ Planetscope covering the Koh Sdach and Koh Rong Archipelagos in Cambodia were collected, processed, and classified, with outputs being compared for seagrass detection ability and suitability. Image scenes were merged and processed to produce depth invariant indexes using the Lyzenga algorithm. Random forest classification with ground truth data yielded accuracies of 79.5% for Sentinel and 79.7% for Planetscope using unprocessed rasters, while depth invariant rasters resulted in accuracies of 80.8% and 70.0% respectively. Sentinel-2 and Planetscope are qualitatively compared based on technical specifications and factors such as band wavelength, radiometric quality, and revisit time. Sentinel-2 shows that it is able to match the seagrass detection ability of the higher resolution Planetscope due to its additional blue band and deeper radiometric depth. Results of this analysis are placed in the context of wider research on detecting seagrass patches through satellite image classification.
|
Subject | |
Type | |
Language |
English
|
Notes |
Project on comparing Sentinel-2 and Planetscope for detecting and monitoring seagrasses in Cambodia’s Koh Rong and Koh Sdach Archipelagos. Files include final report along with 6 polygon shapefiles and a descriptive Readme.
Shapefiles include training and validation data, along with seagrass bed outputs from both Sentinel-2 and Planetscope data. Each satellite type has 2 output polygon datasets, one made using raw imagery and one made using imagery corrected with the Lyzenga algorithm Depth Invariant Index (DII).
|
Date Available |
2021-04-08
|
Provider |
University of British Columbia Library
|
License |
CC-BY 4.0
|
DOI |
10.14288/1.0396662
|
URI | |
Publisher DOI | |
Rights URI | |
Country |
Cambodia
|
Aggregated Source Repository |
Dataverse
|
Item Media
Item Citations and Data
Licence
CC-BY 4.0