| Assessment of Machine Learning Technique for Depth Estimation in Nile River |
| Paper ID : 1014-ICRSSSA-FULL |
| Authors |
|
Noha Kamal *1, Nagwa Elashmawy2 1Head of Information System Unit, Nile Research Institute, National water research center 2National water reserach center |
| Abstract |
| The Nile river is one of the largest rivers in the world, and is considered as a navigation channel. As a result of sedimentation and erosion processes in the natural water stream, bathymetry maps must be updated and surveyed regularly. Periodical bathymetry mapping is a necessity to conduct an updated bathymetry for better river management strategies and development plans, which are time consuming and costly. The Machine Learning (ML) Techniques based on the analysis of the multispectral satellite images has proven its effectiveness in producing bathymetry maps for the coastal zones of the seas and oceans.. This paper, aims at assessing the performance of Machine Learning (ML) techniques for deriving bathymetric data from multispectral satellite imageries (such as Landsat 8 and Sentinel 2). A study area in the fourth Reach of the Nile river in Egypt, between Assuit and Delta barrages is selected for this research, where large number of depth point data are available via field measurements more than 184162 points. The available data were collected in a hydrographic survey mission by a Valiport (MIDAS Surveyor) echo-sounder. The estimated shallow water depth out of the multispectral imageries data retrieved bathymetry are calibrated and validated by the field surveyed data. |
| Keywords |
| Machine Learning, Landsat 8, Sentinel 2, Bathymetry Mapping, Nile River. |
| Status: Accepted (Oral Presentation) |
