|
Estimating Water Surface Area of A Reservoir Using Satellite Image Analysis Technique |
|---|---|
| รหัสดีโอไอ | |
| Creator | Thawatchai Paothui |
| Title | Estimating Water Surface Area of A Reservoir Using Satellite Image Analysis Technique |
| Contributor | Varawoot Vudhivanich, Chirakarn Sirivitmaitrie |
| Publisher | Thai Socities of Agricultural Engineering |
| Publication Year | 2565 |
| Journal Title | Thai Socities of Agricultural Engineering Journal |
| Journal Vol. | 28 |
| Journal No. | 2 |
| Page no. | 1-9 |
| Keyword | Water Surface Area, Deep Learning, Satellite Image, Land classifications, Reservoir |
| URL Website | https://li01.tci-thaijo.org/index.php/TSAEJ/index |
| Website title | Thai Socities of Agricultural Engineering Journal |
| ISSN | 2651-222X |
| Abstract | The objective of this research is to estimate the water surface area from satellite imagery by developing an algorithm based on deep learning techniques from Google Earth Pro satellite imagery in the reservoir area of Khun Dan Prakarnchon Dam, Nakhon Nayok Province, 3 days, consisting of 8th May 2014, 3rd November 2017, and 2nd December 2019. There are 2 methods for estimating the water surface area: area measurements in Google Earth Pro, and a deep learning algorithm recursively learns from the dispersion of colors in satellite imagery to provide statistical data for the algorithm's decision-making which consists of 3 important steps; 1) Learning the actual color segmentation of water surface obtained from satellite images, 2) Classifying between water surface and non-water surface area, and 3) Learning to estimate the water surface area from the satellite image. From the results of the comparison between these two methods of estimating water surface area and the reservoir capacity curve of the Khun Dan Prakan Chon Dam, Nakhon Nayok Province, found that the area measurement in Google Earth Pro has percentage aberrations of 6.50%, 2.14%, and 1.64%, respectively, while the deep learning algorithm recursively learns from the dispersion of colors in satellite imagery has 0.78%, 0.38% and 0.77%, respectively. Therefore, algorithm for dispersion of the color data to estimate the water surface area has more accuracy and precision. This can be further enhanced by using it to estimate the surface area of medium-sized and small reservoirs which there reservoir capacity curve has not yet been generated for further management. |
