Using self-organizing maps to infill missing data in hydro-meteorological time series from the Logone catchment, Lake Chad basin

• Published in: Environmental monitoring and assessment Vol. 188; no. 7; p. 400
• Main Authors: Nkiaka, E., Nawaz, N. R., Lovett, J. C.
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 01-07-2016; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Atmospheric Protection/Air Quality Control/Air Pollution; Chad; Data mining; Datasets; Developing countries; Earth and Environmental Science; Ecology; Ecotoxicology; Environment; Environmental Management; Environmental monitoring; Environmental Monitoring - methods; Lakes; LDCs; Meteorology - methods; Missing data; Monitoring/Environmental Analysis; Neural networks; Neural Networks (Computer); River discharge; River flow; Rivers; Time series; Water resources management; Logone catchment; Infilling missing data; Hydro-meteorological data; Self-organizing maps; Artificial neural networks
• ISSN: 0167-6369; 1573-2959
• DOI: 10.1007/s10661-016-5385-1

Hydro-meteorological data is an important asset that can enhance management of water resources. But existing data often contains gaps, leading to uncertainties and so compromising their use. Although many methods exist for infilling data gaps in hydro-meteorological time series, many of these methods require inputs from neighbouring stations, which are often not available, while other methods are computationally demanding. Computing techniques such as artificial intelligence can be used to address this challenge. Self-organizing maps (SOMs), which are a type of artificial neural network, were used for infilling gaps in a hydro-meteorological time series in a Sudano-Sahel catchment. The coefficients of determination obtained were all above 0.75 and 0.65 while the average topographic error was 0.008 and 0.02 for rainfall and river discharge time series, respectively. These results further indicate that SOMs are a robust and efficient method for infilling missing gaps in hydro-meteorological time series.