The Feitsui reservoir is a major water supply source for more than five million people in Northern Taiwan. The reservoir water quality has been good, but is threatened by eutrophication due to excessive nutrient input and siltation due to sediment loads. Recently, the water authorities in Taiwan have made considerable efforts to devise strategies using watershed conservation practices for the protection of Feitsui reservoir water quality. The control of non-point source pollution (NPS) represents one of the major strategies and the use of best management practices (BMPs) is under careful consideration. The objective of this study was to assess the effectiveness and cost of a full Feitsui watershed implementation of riparian buffer strips and other appropriate conservation practices. Based on the use of watershed simulation models and a statistical relationship between pollution reduction rate and the width and slope of a buffer strip, a methodology for the planning and design of riparian buffer strips was addressed. Data from field experiments were used to calibrate the coefficients of the regression equations. Several planning scenarios were evaluated by means of cost-benefit analysis coupled with net present value method. Data on local construction and maintenance costs for the selected design and location of buffer strips were used in the analysis. Based on several cost-benefit analyses, the scenario for installing buffer strips with 30 m width and 5% slope along both sides

It is vitally important to define the critical condition for a receiving water body in the total maximum daily load (TMDL) development process. One of the major disadvantages of using a continuous simulation approach is that there is no guarantee that the most critical condition will be covered within the subjectively selected representative hydrologic period, which is usually several years depending on the availability of data. Another limitation of the continuous simulation approach, compared to a design storm approach, is the lack of an estimate of the risk involved. Because of the above limitations, a storm event-based critical flow-storm (CFS) approach was previously developed to explicitly address the critical condition as a combination of a prescribed stream flow and a storm event of certain magnitude, both having a certain frequency of occurrence and when combined, would create a critical condition. The CFS approach was tested successfully in a TMDL study for Muddy Creek in Virginia. The present paper reports results of a comparative study on the applicability of the CFS approach in Taiwan. The Dy-yu creek watershed in northern Taiwan differs significantly from Muddy Creek in terms of climate, hydrology, terrain, and other characteristics. Results show that the critical condition for different watersheds might be also different, and that the CFS approach could clearly define that critical condition and should be considered as an alternative method for TMDL development to a continuous simulation approach.

Computer software is an effective tool for simulating urban rainfall–runoff. In hydrological analyses, the storm water management model (SWMM) is widely used throughout the world. However, this model is ineffective for parameter calibration and verification owing to the complexity associated with monitoring data onsite. In the present study, the general regression neural network (GRNN) is used to predict the parameters of the catchment directly, which cannot be achieved using SWMM. Then, the runoff curve is simulated using SWMM, employing predicted parameters based on actual rainfall events. Finally, the simulated and observed runoff curves are compared. The results demonstrate that using GRNN to predict parameters is helpful for achieving simulation results with high accuracy. Thus, combining GRNN and SWMM creates an effective tool for rainfall–runoff simulation.