THE EFFECT OF FOREST INUNDATION ON WATER QUALITY AT THE INTAKE MANGGAR RESERVOIR BALIKPAPAN CITY USING ARTIFICIAL NEURAL NETWORK METHOD

• Main Author: Oksa Rusadi, Tersa
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/52744
• Institution: Institut Teknologi Bandung

In Balikpapan city was built a reservoir which is located in upstream of Manggar river called Manggar Reservoir which is useful of public needs in Balikpapan city especially water needs in Manggar’s protected forest.This reservoir is located in protected forest where there is community life especially for transmigration program since 1960. This situation causing reservoir problem because of activity of the people. But, the problem is not only because of activity of the people, the main problem is increased capacity of the reservoir that be enlarged from 3,27 Million m3 become 10,3 m3 to use the increasing need for clean water for Balikpapan city residents so that take effect the area of inundation from 198 Ha become 443 Ha, this causes 70 Ha acacia tree that planted in protected forest area being submerged and died, but because of this reservoir included in protected forest area then logging of dead acacia due to reservoir inundation is not permitted. Other than that, the presence of aquatic plants Salvina Molesta can suppress vegetation growth and reduce water quality and can speed up the sedimentation process due to uncontrolled plants that are buried and rot in the lake. Therefore in controlling the quality of reservoir water modeling is needed which can provide an overview of the nutrients in the water bodies of the Manggar reservoir on the nutrients that affect the problems that occur. This modeling uses the artificial neural network backpropagation scheme method by modeling all related elements that have a close correlation with the decay of plants in the water bodies of the Manggar Reservoir. By paying attention to the close correlation, there were 5 scenarios with nitrogen and BOD targets that affected the inundation area which resulted in plant rot at 1 and 2 weeks after. Each target parameter and the specified time are modeled with 5 scenarios. The input parameters were used the inundated forest area (Aht), total area (AT), the area that was not inundated (Att), volume, incoming discharge, discharge out, NH3, NO2, NO3 and BOD. From the 5 scenarios have different input combinations. So that the modeling results from each target are the best results based on the correlation coefficient and root mean square error (RMSE) value and validation. From the best model result then it trained and tested again by eliminating the element of inundated forest area to see the effect of the input of the area of forest inundation on the modeling results. The results show that the modeled influence of inundated forest area is different. This influence is seen from the model performance.