Monitoring spatial-temporal variations of surface areas of small reservoirs in Ghana’s Upper East Region using Sentinel-2 satellite imagery and machine learning
Small reservoirs are one of the most important sources of water for irrigation, domestic and livestock uses in the Upper East Region (UER) of Ghana. Despite various studies on small reservoirs in the region, information on their spatial-temporal variations is minimal. Therefore, this study performed a binary Random Forest classification on Sentinel-2 images for five consecutive dry seasons between 2015 and 2020. The small reservoirs were then categorized according to landscape positions (upstream, midstream, and downstream) using a flow accumulation process. The classification produced an average overall accuracy of 98% and a root mean square error of 0.087 ha. It also indicated that there are currently 384 small reservoirs in the UER (of surface area between 0.09 and 37 ha), with 20% of them newly constructed between the 2016-17 and 2019-20 seasons. The study revealed that upstream reservoirs have smaller sizes and are likely to dry out during the dry season while downstream reservoirs have larger sizes and retain substantial amounts of water even at the end of the dry season. The results further indicated that about 78% of small reservoirs will maintain an average of 54% of their water surface area by the end of the dry season. This indicates significant water availability which can be effectively utilized to expand dry season irrigation. Overall, we demonstrate that landscape positions have significant impact on the spatial-temporal variations of small reservoirs in the UER. The study also showed the effectiveness of remote sensing and machine learning algorithms as tools for monitoring small reservoirs.
| Main Authors: | , , , , |
|---|---|
| Format: | Journal Article biblioteca |
| Language: | English |
| Published: |
Elsevier
2022-02
|
| Subjects: | reservoirs, remote sensing, climate variability, satellite imagery, machine learning, |
| Online Access: | https://hdl.handle.net/10568/117314 https://www.sciencedirect.com/science/article/pii/S147470652100125X/pdfft?md5=59bd7a98182c33a44b62aaf447495217&pid=1-s2.0-S147470652100125X-main.pdf https://doi.org/10.1016/j.pce.2021.103082 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Small reservoirs are one of the most important sources of water for irrigation, domestic and livestock uses in the Upper East Region (UER) of Ghana. Despite various studies on small reservoirs in the region, information on their spatial-temporal variations is minimal. Therefore, this study performed a binary Random Forest classification on Sentinel-2 images for five consecutive dry seasons between 2015 and 2020. The small reservoirs were then categorized according to landscape positions (upstream, midstream, and downstream) using a flow accumulation process. The classification produced an average overall accuracy of 98% and a root mean square error of 0.087 ha. It also indicated that there are currently 384 small reservoirs in the UER (of surface area between 0.09 and 37 ha), with 20% of them newly constructed between the 2016-17 and 2019-20 seasons. The study revealed that upstream reservoirs have smaller sizes and are likely to dry out during the dry season while downstream reservoirs have larger sizes and retain substantial amounts of water even at the end of the dry season. The results further indicated that about 78% of small reservoirs will maintain an average of 54% of their water surface area by the end of the dry season. This indicates significant water availability which can be effectively utilized to expand dry season irrigation. Overall, we demonstrate that landscape positions have significant impact on the spatial-temporal variations of small reservoirs in the UER. The study also showed the effectiveness of remote sensing and machine learning algorithms as tools for monitoring small reservoirs. |
|---|