GROUNDWATER LEVEL MODELING FOR PREDICTION OF POTENTIAL GROUNDWATER AVAILABILITY IN THE FUTURE BASED ON ARTIFICIAL NEURAL NETWORKS, SLEMAN REGENCY

• Main Author: Nur Angga Rahayu, Naya
• Format: Theses
• Language: Indonesia
• Subjects: Teknik saniter dan perkotaan; teknik perlindungan lingkungan
• Online Access: https://digilib.itb.ac.id/gdl/view/78242
• Institution: Institut Teknologi Bandung

Human life cannot be separated from the need for water. Population growth increases the amount of water demand. The most widely used water source is groundwater. DIY province has experienced an increase in water demand, one of which is from the business sector. Sleman Regency has experienced rapid growth that affects population density and land use change. It can disrupt the sustainability of groundwater availability because Sleman Regency is the upstream part of the Yogyakarta-Sleman groundwater basin. Sleman Regency covers the catchment, transition, and groundwater discharge areas. Land use conversion in transition areas and population increases affects groundwater decline. Some areas in DIY have experienced a groundwater decline by 0-9 m. Modeling is fundamental as a first step to maintaining the sustainability of groundwater availability. The method used is ANN. The piezometric groundwater level model requires influential variables. Five variables affect the piezometric groundwater level: precipitation, temperature, discharge (non-domestic water demand), hydraulic conductivity, and transmissivity. The six variables act as input variables in modeling, while the piezometric groundwater level is the output variable. The whole correlation between variables input and variable output was weak. The variables that have the level of influence from the largest to the smallest are discharged with a correlation value of 0.47; transmissivity 0.16; rainfall -0.15; hydraulic conductivity 0.08; and temperature -0.03. However, variable discharge was more dominant in influencing piezometric groundwater level. Therefore, to analyze the trend of piezometric groundwater level based on discharge changes. The trend of groundwater discharge had decreased, followed by the piezometric groundwater level getting shallower within five years, namely from 2018 – 2022 in Sleman Regency. The ANN method has modeling capability in this case study. The model's accuracy with actual conditions is assessed from the R2 and RMSE values. At the training stage, it was found that the R2 value was 0.84, and RMSE was 0,15. Furthermore, validation is needed to give confidence in the model's correctness. The testing stage was used as validation. The R2 value was 0.92, and RMSE was 0.07 in the testing process. The results of the model have been good and entirely accurate, so the ideal weight values obtained can be used computationally to predict the piezometric groundwater level in the future. Predictions were made based on various scenarios to match the piezometric groundwater level under different environmental conditions. This study made three scenarios, the first scenario describes the dry season, the second scenario describes the rainy season, and the third scenario determines the piezometric groundwater level in Sleman Regency. The precipitation in the first scenario was 0 mm, 20 mm, and 50 mm, describing cloudy, light rain and moderate rain. The temperature was 30°C. The discharge increase assumption during the drought was 9%. The rainy season scenario was designed with heavy (100 mm), very heavy (150 mm), and extreme (300 mm) precipitation conditions with an air temperature of 26°C, then the assumption of an increase in discharge was 4%. The first and second scenarios were created to see the change in piezometric groundwater level in each sub-district. The third scenario for depicting Sleman Regency was made with average conditions: rainfall 200 mm, temperature 26°C, hydraulic conductivity 0.000340544 m/sec and transmissivity 270.263 m 2 /day. The prediction results for the first and second scenarios showed significant and insignificant changes in piezometric groundwater level. Ngaglik sub-district experiences significant changes during the dry and rainy seasons. The groundwater decline on average during drought will be 10.7 m and 5 m during rain. An insignificant change in piezometric groundwater level will occur in Tempel District, in the rainy season a decrease in piezometric groundwater level by 0.17 m; and during the dry season by 1 m. The insignificance of the piezometric groundwater level decrease was due to the relatively low discharge in the Tempel sub-district compared to other sub-districts. The decrease in piezometric groundwater level in Sleman Regency ranged from 18% to 58%. This percentage shows that changes in the piezometric groundwater level of confined aquifers are classified as safe to vulnerable.