Model Calibration Result and Discussion

62 examine the relationship between driving factors and land use change in study area.  Calibration result hydrology model gave value or R 2 achieve 0.524, while the values of Nash-Sutcliffe Efficient NSE is 0.67 and 42.9 of relative volume errors RV E . By using three tests it can be stated that the model is satisfactory accepted.  The values of water yield from 4 scenarios for Year 2030 scenario 1, scenario 2, scenario 3 and scenario 4 are not significantly different due to limited sized of research area and low dynamic of land use driving factor. However, based on the comparison about values of water yield between scenario-based simulation for Year 2030 and existing condition of Year 2010 data shows increasing of water yield from Year 2010 to Year 2030.  The increasing values of water yield influenced by forest rehabilitation activity by the government inside forest area and the development of community forest outside the forest area. Increasing values of water yield is quite high for scenario 2 and scenario 4, where government policy about restriction of land use inside forest area applied. That means government policy to prohibit land use conversion inside forest is appropriate to apply.

4.2. Further Research Directions

By considering some limitations that are discussed before and possible improvement of the model approach to support forecasting land use change and analysis on impact to water yield, further development of this research would be interesting to consider some factors:  Further research could be implemented in cooperation with planners and decision makers. By implementing the methodology in practice, some benefits can be obtained, including the results of this research could be effectively communicated with local managers, improvement of the applicability of the model and the methods, could involve more data in the analysis and more relevant policy could be put into practice. 63  This research was completely done; however further research could apply scenario-based approach in the larger scale by using higher spatial resolution data. CLUE-S model framework gives possibility to accommodate more detailed scale data and level of analysis. The advantage is that the research could explore more information from the specific locations and improve the understanding about a specific problem on upstream watershed zone.  Improvement of the model could be made in the driving factors selection before they are incorporated in the logistic regression. All of influencing variables could be examined by using expert knowledge approach to improve the selection process and reduce subjective preference. Moreover, the dynamic social-economic might need to include on the analysis for future research, to determine the best watershed management that can be implemented in Upstream Cisadane Watershed. 64