Muhammad Raza Ul Mustafa

River Suspended Sediment Prediction Using Various Multilayer Perceptron Neural Network Training Algorithms—A Case Study in Malaysia

Estimation of suspended sediment discharge in rivers has a vital role in dealing with water resources problems and hydraulic structures. In this study, a Multilayer Perceptron (MLP) feed forward neural network with four different training algorithms was used to predict the suspended sediment discharge of a river (Pari River at Silibin) in Peninsular Malaysia. The training algorithms are Gradient Descent (GD), Gradient Descent with Momentum (GDM), Scaled Conjugate Gradient (SCG), and Levenberg Marquardt (LM). Different statistical measures, time of convergence and number of epochs to reach the required accuracy were used to evaluate the performance of training algorithms. The analysis showed that SCG and LM performed better than GD and GDM. While the performance of the superior algorithms (i.e., SCG and LM) is similar, LM required considerably shorter time of convergence. It was concluded that both training algorithms SCG and LM could be recommended for suspended sediment prediction using MLP networks. However, LM was the faster (1/7 of SCG convergence time) of the two algorithms.

Cr(VI) adsorption from aqueous solution by an agricultural waste based carbon

The study examined the adsorption of hexavalent chromium [Cr(VI)] from aqueous solution by acidically prepared rice husk carbon (APRHC). APRHC was characterized in terms of surface area, micropore area, micropore volume, average pore diameter and surface morphology. The effects of pH, contact time, initial Cr(VI) concentration and adsorbent dose on the adsorption of Cr(VI) from aqueous solution were investigated. Batch adsorption tests showed that Cr(VI) adsorption depends on initial concentration, contact time and pH. Equilibrium adsorption was achieved in 120 min, while maximum Cr(VI) adsorption occurred at pH 2. An artificial neural network (ANN) was used to model Cr(VI) adsorption. The Levenberg–Marquardt (LM) training algorithm was found to be the best among the 11 backpropagation (BP) algorithms tested, with a lowest mean square error (MSE) of 8.8876 and highest coefficient of determination (R2) of 0.987. Adsorption of Cr(VI) by APRHC followed pseudo-second order kinetics. Langmuir and Freundlich isotherm equations were fitted to the equilibrium adsorption data; the former isotherm yielded a better fit. The thermodynamic results indicate that the process of Cr(VI) adsorption by APRHC was endothermic in nature. Desorption of Cr(VI) was very low, i.e. in the range from 0.1 to 9%. Cr(VI) adsorption capacity by APRHC was compared with that of various adsorbents. APRHC showed a high capacity for adsorption of Cr(VI). APRHC can be employed as an effective adsorbent and substitute for commercially available activated carbon for the removal of Cr(VI) from aqueous solutions and wastewater systems.

Drought Analysis and Water Resources Management Inspection in Euphrates–Tigris Basin

Growing population, increasing basin development, and progressively declining water supplies are typical water resources issues in the Middle East. Drought is one of the most damaging climate‐related hazards that affect more people than any other. For identify‐ ing drought‐prone areas in the Euphrates–Tigris Basin, multifold aspects of drought and its features such as the frequency of drought occurrence and its spatial distribution were assessed. The long‐term precipitation data were collected from different meteorological stations of Turkey and Iran, and standard precipitation index (SPI) was calculated. Due to the lack of raw data, the literature works on drought were used in Syria and Iraq to obtain a drought perception in these countries. Moreover, the policy of water resources management and the hydraulic works in these regions were considered. The results show significant changes in the precipitation in these regions over the past decades. The projects undertaken in the basin are not in line with the principles of integrated water resources management and intensify the drought and caused marshland demise in the down‐ stream of the basin. The results of a comprehensive analysis of precipitation variation and water management in this research can alter the policy of water resources management in order to avoid drought in the basin.

Artificial Neural Network Modeling for Spatial and Temporal Variations of Pore-Water Pressure Responses to Rainfall

Knowledge of spatial and temporal variations of soil pore-water pressure in a slope is vital in hydrogeological and hillslope related processes (i.e., slope failure, slope stability analysis, etc.). Measurements of soil pore-water pressure data are challenging, expensive, time consuming, and difficult task. This paper evaluates the applicability of artificial neural network (ANN) technique for modeling soil pore-water pressure variations at multiple soil depths from the knowledge of rainfall patterns. A multilayer perceptron neural network model was constructed using Levenberg-Marquardt training algorithm for prediction of soil pore-water pressure variations. Time series records of rainfall and pore-water pressures at soil depth of 0.5 m were used to develop the ANN model. To investigate applicability of the model for prediction of spatial and temporal variations of pore-water pressure, the model was tested for the time series data of pore-water pressure at multiple soil depths (i.e., 0.5 m, 1.1 m, 1.7 m, 2.3 m, and 2.9 m). The performance of the ANN model was evaluated by root mean square error, mean absolute error, coefficient of correlation, and coefficient of efficiency. The results revealed that the ANN performed satisfactorily implying that the model can be used to examine the spatial and temporal behavior of time series of pore-water pressures with respect to multiple soil depths from knowledge of rainfall patterns and pore-water pressure with some antecedent conditions.

EVALUATION OF RAINFALL-RUNOFF EROSIVITY FACTOR FOR CAMERON HIGHLAND, PAHANG, MALAYSIA

Soil is one of the vital components of the natural environment that is non-renewable on a human time-scale [communication from commission to the council, 2006]. Soil erosion by water or wind has been a global threat posing significant challenges in terms of land degradation and desertification [Valentin et al., 2005], aquatic imbalance and deterioration of water quality in rivers and reservoirs. Most soils are exposed to erosion through poor agricultural practices, indiscriminate deforestation, overgrazing, forest fires hazard, land slide, construction and mining activities among others. Soil erosion has both onand off-site impacts on land and water resources. On-site impact can be degradation of soil quality due to immediate loss of its upper layer [Bakker et al., 2004], while off-site impacts resulted in increased water turbidity and pollutants, flooding, reduced crop yield, poor water quality, loss of reservoir and river capacity, which may lead to significant economic issues and environmental degradation [Quinton et al., 2001; Haygarth, 2005; Delmas et al. 2012, Oh and Jung, 2005]. In the recent past, several techniques or models have been developed and utilized to assess soil erosion by water. These models are broadly clasEVALUATION OF RAINFALL-RUNOFF EROSIVITY FACTOR FOR CAMERON HIGHLANDS, PAHANG, MALAYSIA

Application Of Statistical Downscaling Model (SDSM) For Long Term Prediction Of Rainfall In Sarawak, Malaysia

Long-term prediction of rainfall over a catchment is a challenge for hydrologists. It is required for water resources management, hydropower energy forecasting and flood risks assessment in river basins. Several large scale climate phenomena affect the occurrence of rainfall around the world i.e El Nino Southern Oscillation (ENSO) and Indian Ocean Dipole (IOD) are most famous for their effect on India, North and South America and Australia. This study is motivated to evaluate the performance of Statistical Downscaling Model (SDSM) developed by annual and monthly sub models for rainfall downscaling from Global Climate Models (GCMs) over the two districts in Sarawak. It is noted that the monthly sub-models have better performance over the annual sub-models. However, both monthly and annual sub-models have poor correlation with the recorded rainfall for the calibration and validation period. Results indicate that both stations show increasing trend in the future annual rainfall under H3A2 and H3B2 scenarios of HadCM3. SDSM predict that the annual rainfall at Belaga and Limbang is expected to increase by 37.8% and 22.7% respectively by 2074. Overall the SDSM approximates the average rainfall very well during the calibration and validation period but the correlation between observed and forecasted rainfall was not so good. And there is a need to improve the statistical downscaling modelling to develop better correlation between predictand and predictors to have better model performance over the wet regions like Sarawak.

Estimation of soil pore-water pressure variations using a thin plate spline basis function

Information of soil pore-water pressure changes due to climatic effect is an integral part for studies associated with hill slope analysis. Soil pore-water pressure variations in a soil slope due to rainfall were predicted using Artificial Neural Network (ANN) technique with Thin Plate Spline (TPS) radial basis function. A radial basis function (RBF) neural network with network architecture of 8-36-1 (input-hidden-output) was selected to develop RBF model. Number of hidden neurons was selected using trial and error procedure whereas spread of the basis function was established using normalization method. Time series data of rainfall and pore-water pressure was used for training and testing the RBF model. The performance of the model was evaluated using root mean square error, coefficient of correlation and coefficient of efficiency. The results of the model prediction revealed that the model produced promising results indicating that TPS basis function is able to predict time series of pore-water pressure responses to rainfall. Comparison with other studies showed that the RBF model using TPS basis function can be used as alternate of Gaussian basis function for prediction of soil pore-water pressure variations.

Evaluation of CMIP5 models for projection of future precipitation change in Bornean tropical rainforests

We present the climate change impact on the annual and seasonal precipitation over Rajang River Basin (RRB) in Sarawak by employing a set of models from Coupled Model Intercomparison Project Phase 5 (CMIP5). Based on the capability to simulate the historical precipitation, we selected the three most suitable GCMs (i.e. ACCESS1.0, ACCESS1.3, and GFDL-ESM2M) and their mean ensemble (B3MMM) was used to project the future precipitation over the RRB. Historical (1976–2005) and future (2011–2100) precipitation ensembles of B3MMM were used to perturb the stochastically generated future precipitation over 25 rainfall stations in the river basin. The B3MMM exhibited a significant increase in precipitation during 2080s, up to 12 and 8% increase in annual precipitation over upper and lower RRB, respectively, under RCP8.5, and up to 7% increase in annual precipitation under RCP4.5. On the seasonal scale, Mann-Kendal trend test estimated statistically significant positive trend in the future precipitation during all seasons; except September to November when we only noted significant positive trend for the lower RRB under RCP4.5. Overall, at the end of the twenty-first century, an increase in annual precipitation is noteworthy in the whole RRB, with 7 and 10% increase in annual precipitation under the RCP4.5 and the RCP8.5, respectively.

Prediction models for flow resistance in flexible vegetated channels

ABSTRACT The analysis of flow resistance due to vegetation remains an issue in the hydraulic industry, although it has been systematically studied for several decades, accurate prediction of the resistance is still a challenge. This is because most of the previous studies used synthetic vegetation to model flow–vegetation interactions. This paper presents the applications of the artificial neural network (ANN) and gene expression programming (GEP) as advanced tools, to predict the flow resistance (n) of natural vegetation using a grassed swale and laboratory channel, irrespective of the grass height with relative to flow depth. To achieve this, hourly discharges and water depths were measured in the grassed swale for different rainfall events using the electromagnetic current metre. Experiments were performed in the laboratory channel using the same grass, in order to get additional data. From the results obtained regression equation was developed for predicting the flow resistance through the use of dimensional analysis. The regression equation obtained was compared with the established models of ANN and GEP. The results show that ANN and GEP models gave a better prediction of n-values, based on performance indices. However, the GEP model would be preferred as it produced a physical equation that can be used in engineering practice.

Analysis of Manning’s and Drag Coefficients for Flexible Submerged Vegetation

Accurate determination of flow resistance is of great significance in modelling of open channels that will convey water efficiently. Although, resistance or drag induced by vegetation have been systematically studied for several decades, estimating of the resistance remain as a challenge. This is because most of previous studies use artificial vegetation to investigate flow – vegetation interactions. To overcome this, the present study evaluates the vegetation resistance in terms of Mannings roughness coefficient and drag coefficient using a natural flexible vegetation (cow grass) under submerged condition. From the experimental result obtained, it was observed that the Mannings and drag coefficients decreased with the increasing in average velocity. Also, graphical relationship between Mannings coefficient, n and drag coefficient, CD has been developed with R2 = 0.9465, which indicate that there exist a strong correlation between n and CD, and one can use the proposed graphical model to predict the n - values corresponding to the CD – values.