scholarly journals Modelling streamflow response to hydro-climatic variables in the Upper Mkomazi River, South Africa

2014 ◽  
Author(s):  
◽  
Oluwaseun Kunle Oyebode
Keyword(s):  
South Africa ◽  
Water Resources ◽  
Genetic Programming ◽  
Predictive Models ◽  
Model Development ◽  
De Algorithm ◽  
Planning And Management ◽  
Ann Models ◽  
Artificial Neural

Streamflow modelling remains crucial to decision-making especially when it concerns planning and management of water resources systems in water-stressed regions. This study proposes a suitable method for streamflow modelling irrespective of the limited availability of historical datasets. Two data-driven modelling techniques were applied comparatively so as to achieve this aim. Genetic programming (GP), an evolutionary algorithm approach and a differential evolution (DE)-trained artificial neural network (ANN) were used for streamflow prediction in the upper Mkomazi River, South Africa. Historical records of streamflow and meteorological variables for a 19-year period (1994- 2012) were used for model development and also in the selection of predictor variables into the input vector space of the models. In both approaches, individual monthly predictive models were developed for each month of the year using a 1-year lead time. Two case studies were considered in development of the ANN models. Case study 1 involved the use of correlation analysis in selecting input variables as employed during GP model development, while the DE algorithm was used for training and optimizing the model parameters. However in case study 2, genetic programming was incorporated as a screening tool for determining the dimensionality of the ANN models, while the learning process was further fine-tuned by subjecting the DE algorithm to sensitivity analysis. Altogether, the performance of the three sets of predictive models were evaluated comparatively using three statistical measures namely, Mean Absolute Percent Error (MAPE), Root Mean-Squared Error (RMSE) and coefficient of determination (R2). Results showed better predictive performance by the GP models both during the training and validation phases when compared with the ANNs. Although the ANN models developed in case study 1 gave satisfactory results during the training phase, they were unable to extensively replicate those results during the validation phase. It was found that results from case study 1 were considerably influenced by the problems of overfitting and memorization, which are typical of ANNs when subjected to small amount of datasets. However, results from case study 2 showed great improvement across the three evaluation criteria, as the overfitting and memorization problems were significantly minimized, thus leading to improved accuracy in the predictions of the ANN models. It was concluded that the conjunctive use of the two evolutionary computation methods (GP and DE) can be used to improve the performance of artificial neural networks models, especially when availability of datasets is limited. In addition, the GP models can be deployed as predictive tools for the purpose of planning and management of water resources within the Mkomazi region and KwaZulu-Natal province as a whole.

scholarly journals Analysis of Droughts in the Central Region of South Africa and Their Association with SST Anomalies

2014 ◽  
Vol 2014 ◽  
pp. 1-8 ◽  
Author(s):  
Desalegn C. Edossa ◽  
Yali E. Woyessa ◽  
Worku A. Welderufael
Keyword(s):  
South Africa ◽  
Water Resources ◽  
Central Region ◽  
El Niño ◽  
Time Scale ◽  
El Nino ◽  
Lag Time ◽  
Planning And Management ◽  
El Niño Events ◽  
El Nino Events

The objective of this study was to characterise meteorological droughts in the Central Region of South Africa using Standardised Precipitation Evapotranspiration Index (SPEI) and to examine if there is a relationship between drought and El Niño events. The SPEI was used to quantify the precipitation deficit over time and space across the catchment for the time-scales that are important for planning and management of water resources. Based on 12-month time-scale, the total number of drought events identified in the area using SPEI ranged between 13 and 20 during the period of analysis (1952–1999). Considering the effects of event magnitude and duration as severity parameters, the most severe drought event was identified during 1973 followed by 1995 based on 12-month time-scale. Moreover, it was also found that the number of moderate, severe, and extreme drought events identified by SPEI follows increasing trend with decade during the period of analysis. Results of Spearman’s rank correlation test revealed that the trends exhibited by mild (SPEI-3 and SPEI-6), moderate (SPEI-12), severe (SPEI-12), and extreme (SPEI-3) drought categories are statistically significant at 5% significance level. The study also revealed that drought events in the central region of South Africa are preceded by El Niño events in the tropical Pacific (Nino 3.4) with an average lag time of 8 months between the onsets of the two events. It was found that hydrological drought events in the study area lag behind meteorological drought events with an average lag time of 7.4 months. Findings of this study can be used to forecast drought events in the area for the proper planning and management of water resources.

scholarly journals Prediction of CSO Chamber Level Using Evolutionary Artificial Neural Networks

10.29007/8pr7 ◽  
2018 ◽  
Author(s):  
Talia Rosin ◽  
Michele Romano ◽  
Kevin Woodward ◽  
Ed Keedwell ◽  
Zoran Kapelan
Keyword(s):  
Water Level ◽  
Evolutionary Strategy ◽  
Combined Sewer Overflows ◽  
Radar Rainfall ◽  
Untreated Wastewater ◽  
Input Structure ◽  
Ann Models ◽  
Artificial Neural ◽  
Trial And Error Method

Combined Sewer Overflows (CSOs) are a major source of pollution, spilling untreated wastewater directly into water bodies and/or the environment. If spills can be predicted in advance then interventions are available for mitigation. This paper presents Evolutionary Artificial Neural Network (EANN) models designed to predict water level in a CSO chamber up to 6 hours ahead using inputs of past CSO level, radar rainfall and rainfall forecast data. An evolutionary strategy algorithm is used to automatically select the optimal ANN input structure and parameters, allowing the ANN models to be constructed specifically for different CSO locations and forecast horizons. The methodology has been tested on a real world case study CSO and the EANN models were found to be superior to ANN models constructed using the trial and error method. This methodology can be easily applied to any CSO in a sewer network without substantial human input. It is envisioned that the EANN models could be beneficially used by water utilities for near real-time modelling of the water level in multiple CSOs and the generation of alerts for upcoming spills events.

scholarly journals The evaluation on artificial neural networks (ANN) and multiple linear regressions (MLR) models over particulate matter (PM10) variability during haze and non-haze episodes: A decade case study

2019 ◽  
Vol 15 (2) ◽  
pp. 164-172 ◽  
Author(s):  
Ku Mohd Kalkausar Ku Yusof ◽  
Azman Azid ◽  
Muhamad Shirwan Abdullah Sani ◽  
Mohd Saiful Samsudin ◽  
Siti Noor Syuhada Muhammad Amin ◽  
...  
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Particulate Matter ◽  
Meteorological Factors ◽  
Model Development ◽  
Eastern Region ◽  
Multiple Linear Regressions ◽  
Artificial Neural ◽  
Linear Regressions

The comprehensives of particulate matter studies are needed in predicting future haze occurrences in Malaysia. This paper presents the application of Artificial Neural Networks (ANN) and Multiple Linear Regressions (MLR) coupled with sensitivity analysis (SA) in order to recognize the pollutant relationship status over particulate matter (PM10) in eastern region. Eight monitoring studies were used, involving 14 input parameters as independent variables including meteorological factors. In order to investigate the efficiency of ANN and MLR performance, two different weather circumstances were selected; haze and non-haze. The performance evaluation was characterized into two steps. Firstly, two models were developed based on ANN and MLR which denoted as full model, with all parameters (14 variables) were used as the input. SA was used as additional feature to rank the most contributed parameter to PM10 variations in both situations. Next, the model development was evaluated based on selected model, where only significant variables were selected as input. Three mathematical indices were introduced (R2, RMSE and SSE) to compare on both techniques. From the findings, ANN performed better in full and selected model, with both models were completely showed a significant result during hazy and non-hazy. On top of that, UVb and carbon monoxide were both variables that mutually predicted by ANN and MLR during hazy and non-hazy days, respectively. The precise predictions were required in helping any related agency to emphasize on pollutant that essentially contributed to PM10 variations, especially during haze period.

Estimation of Groundwater Level Using Artificial Neural Networks: a Case Study of Hatay-Turkey

2017 ◽  
Author(s):  
Fatih Üneş ◽  
Mustafa Demirci ◽  
Eyup Ispir ◽  
Yunus Ziya Kaya ◽  
Mustafa Mamak ◽  
...  
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Groundwater Level ◽  
Underground Water ◽  
Population Increase ◽  
Ann Model ◽  
Total Water Consumption ◽  
Ann Models ◽  
Artificial Neural

Groundwater, which is a strategic resource in Turkey, is used for drinking-use, agricultural irrigation and industrial purposes. Population increase and total water consumption are constantly increasing. In order to meet the need for water, over-shoots from underground water have caused significant falls in groundwater level. Estimation of water level is important for planning an efficient and sustainable groundwater management. In this study, groundwater level, monthly mean precipitation and temperature observations of Turkish General Directorate of State Hydraulic Works (DSI) in Hatay, Amik Plain, Kumlu district were used between 2000 and 2015 years. The performance evaluation was done by creating Multi Linear Regression (MLR) and Artificial Neural Networks (ANN) models. The ANN model gave better results than the MLR model.

scholarly journals Prediction of Compressive Strength of Concrete Using Artificial Neural Network and Genetic Programming

2016 ◽  
Vol 2016 ◽  
pp. 1-10 ◽  
Author(s):  
Palika Chopra ◽  
Rajendra Kumar Sharma ◽  
Maneek Kumar
Keyword(s):  
Compressive Strength ◽  
Genetic Programming ◽  
Prediction Models ◽  
Model Development ◽  
Strength Prediction ◽  
Ann Model ◽  
Concrete Compressive Strength ◽  
Compressive Strength Of Concrete ◽  
Artificial Neural

An effort has been made to develop concrete compressive strength prediction models with the help of two emerging data mining techniques, namely, Artificial Neural Networks (ANNs) and Genetic Programming (GP). The data for analysis and model development was collected at 28-, 56-, and 91-day curing periods through experiments conducted in the laboratory under standard controlled conditions. The developed models have also been tested on in situ concrete data taken from literature. A comparison of the prediction results obtained using both the models is presented and it can be inferred that the ANN model with the training function Levenberg-Marquardt (LM) for the prediction of concrete compressive strength is the best prediction tool.

Sign in / Sign up

Export Citation Format

Share Document