The retrieval of the coastal water depths from data of multi- and hyperspectral remote sensing imagery

Grigorieva, O. V.; Zhukov, D. V.; Markov, Andrey; Mochalov, Viktor

doi:10.1134/s1024856017010067

Cited by 5 publications

(1 citation statement)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The images were downloaded and exported into the R software (R Core Team, 2021) for further analysis. The bathymetry was retrieved by using the support vector machine (SVM) algorithm, which yielded the best accuracy after comparing it with other commonly used methods such as Stumpf Method (Stumpf et al., 2003), Random Forest (Breiman et al., 2001), Gradient Boosting Machine (Friedman, 2001), Extreme Gradient Boosting (Kaixiang et al., 2020), and Artificial Neural Network (Grigorieva et al., 2017). We used log‐transformed bands (band2/band3 and band4/band2), principal component analysis (PCA) outputs (PC1, PC2, and PC3), and on‐site collected depth points to create raster layers of continuous water depth based on the SVM (see Mabula et al., 2023 for details).…”

Section: Methodsmentioning

confidence: 99%

Use of GIS‐based multicriteria evaluation for improved selection of suitable sites for cage fish farming in Mwanza Gulf, Lake Victoria

Mabula,

Kisanga,

Pamba

et al. 2023

Aquaculture Fish & Fisheries

View full text Add to dashboard Cite

Stagnating capture fisheries and increasing fish protein demand necessitate aquaculture production to bridge the gap. Lake Victoria is a potential water body for increasing fish production through cage farming. The sustainability of Lake Victoria for cage farming depends on timely and holistic site selection. However, current site selection methods involve complex, resource‐intensive field surveys that lack a holistic approach to integrate multiple factors. Accordingly, information regarding suitable sites for sustainable fish cage farming in Lake Victoria remains scarce. In this study, a transparent geographic information systems (GIS) and multicriteria evaluation (MCE) here after reffered to as GIS‐MCE was used to reveal potential sites to be permitted or avoided during cage aquaculture development in the Mwanza Gulf of Lake Victoria (Tanzania). Our analysis involved weighting and integration of sub‐models representing ecologically sensitive areas, physical environment, and socio‐economic and water quality variables into a single spatial model portraying different site suitability levels in the Mwanza Gulf. The results indicated that the sub‐models identified relatively larger suitable and most suitable sites compared to the overall model. No site maintained its status across all sub‐models. The overall model designated a small area (5.10 km2 or 1.52%) as the most suitable site, with 24.20 km2 (7.44%) as suitable, 64.47 km2 (19.82%) as less suitable, and 42.63 km2 (13.12%) as unsuitable for cage fish farming. The remaining area (188.84 km2 or 58.06%) was a constrained site to be avoided during cage aquaculture development. Taken together, the individual sub‐models are ineffective in designating potential sites for fish cage culture and thus should not be used solely. The GIS‐MCE general model provides a fast and timely method for identifying potential sites for cage farming in Lake Victoria. Fish farmers and managers should use the GIS‐MCE overall model in inland waters to facilitate site selection for complying with licensing requirements and decrease field extensive surveys.

show abstract