Generalized Additive Mixed Modelling of River Discharge in the Black Volta River

Abstract

River discharge data offer a rich source of information for reservoir management and flood control, if modelling can separate out the effects of rainfall, land use, soil type, relief, and weather conditions. In this paper, we model river discharge data from the Black Volta River, using Generalised Additive Mixed Models (GAMMs) with a space-time interaction represented via a tensor product of continuous time and discrete space. River discharge data from January 2000 to December 2009 for the four gauge stations along the Black Volta River namely, Lawra, Chache, Bui and Bamboi were obtained from the hydrological services department of Ghana and used for model fitting. Four GAMMs were explored, two with space-time interactions and two without space-time interactions. The comparison of the performance of the models with space-time interactions and those without space-time interactions based on Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) suggests that in this application, the former is better overall and in particular for modelling local variations. Further, a model with space and time main effects performed better compared with one without space and time main effects. After model selection, checking and validation, there is evidence for increasing river discharge from the most upstream gauge station to the most downstream gauge station for the study period.

Regional hydrological frequency analysis at ungauged sites with random forest regression

“…Generalized Additive Models (GAM) due to their considerable flexibility, are used in regional flood frequency analysis, water quality estimation, river discharge modeling, etc. (Chebana et al, 2014;Iddrisu et al, 2017;Morton and Henderson, 2008;Ouarda et al, 2018;Rahman et al, 2017). Other non-linear approaches used RFA include Projection Pursuit Regression (Durocher et al (2015), Non-Linear CCA Ouali et al (2015), and Adaptive Neuro-Fuzzy Inference Systems (ANFIS) (Shu and Ouarda, 2008).…”

Stream Temperature Modeling Using Functional Regression Models

“…This model was introduced by Hastie and Tibshirani () as an extension of the generalized linear model, without the assumption of linearity in the relationship between predictors and the predictand, as well as relaxing the normality assumption. Even if only few applications of the GAM in stream temperature have been reported (Wehrly et al ; Laanaya et al ), it has been widely used in hydrology (Chebana et al ; Zhang et al ; Falah et al ; Iddrisu et al ; Rahman et al ). GAM is defined as follows:where g is the link function, E ( y ) is the expected value of the predictand (in our case, the daily mean stream temperature), x j is the j th predictor, f j is the associated smooth nonlinear function (often combination of cubic splines), and ε is the error term.…”

“…This model was introduced by Hastie and Tibshirani (1990) as an extension of the generalized linear model, without the assumption of linearity in the relationship between predictors and the predictand, as well as relaxing the normality assumption. Even if only few applications of the GAM in stream temperature have been reported (Wehrly et al 2009;Laanaya et al 2017), it has been widely used in hydrology (Chebana et al 2014;Zhang et al 2015;Falah et al 2017;Iddrisu et al 2017;Rahman et al 2018). GAM is defined as follows:…”

Controls on Sediment Bed Erodibility in a Muddy, Partially-Mixed Tidal Estuary

“…Histograms of eroded mass at 0.2 Pa appeared log normal, and eroded mass at 0.2 Pa was logtransformed for subsequent analysis. River discharge was also log-transformed because it is common practice in the hydrological literature (e.g., Iddrisu et al, 2017). All variables were then linearly standardized by subtracting their means and dividing by their standard deviations.…”