Assignment # 7

Article Reference:
Cristina Perez-Pedini, James F. Limbrunner, Richard M. Vogel, Optimization Location of Infiltration Based Best Management Practices for Storm Water Management, Nov-Dec 2008, Journal of Water Resources Planning and Management

Summary:
The present paper discusses about the optimization of infiltration based best management practices of Aberjona river watershed which is carried out by combination of hydrologic model and genetic algorithm. Main objective of the study was to optimize the number and location of infiltration based BMP thereby reducing the peak flow at the outlet of the watershed.

The hydrologic model was built using the SCS approach (CN method). This model was optimized using GA to find the areas within the watershed where the infiltration based BMP would be more effective to reduce the impacts of the increased peak flow. By considering different optimization results, the trade off curve between flood flow reduction and program budget. The paper considers the option of optimizing the BMP on the upslope so that it is more effective at the downstream flood protection. The whole catchment is modeled as the Hydrologic Response Units (HRU). It is considered that each HRU unit drains to nearest upslope HRU’s and a maximum HRU’s which can drain into are limited to 7. It is seen that only at most 2 HRU’s drain into adjacent upslope HRU. The stream network collects the runoff from the upstream HRU and routes to the watershed outlet. The hydrologic parameters were developed using ArcGIS and the cell to cell connectivity is done using D8 algorithm. BMP is modeled using a binary variable; introduction of BMP decreases the CN of the HRU. In the scope of this paper, they have assumed only the planning of the whole watershed system instead of designing of each BMP. We could further take the inputs from this methodology to study the watershed which have larger impact on the flood peak flow and apply the particular BMP to each HRU. The study considers only single type of BMP for the whole study as project cost would increase if different BMP types were considered. This study did not consider the cost implication of setting a BMP in HRU with different land use and surface conditions.

The hydrologic model is modified so that it calculated the cumulative water at the HRU by considering the sum of the precipitation, excess groundwater at the HRU and excess runoff from the adjacent down slope HRU and subtracting the cumulative initial abstraction. As defined by CN method, initial abstraction is defined as a fraction of the soil storage capacity. Combining above mentioned procedures, cumulative runoff, and cumulative infiltration are estimated. Ground water storage is calculated as sum of storage of previous time step, cumulative infiltration and sum of the previous time step base flow from the adjacent upslope. Stream channel is routed to carry the runoff and the base flow to the watershed outlet with a time lag. One of the model disadvantages is that time step is considered to be the major control of the system, so the time step needs to be calibrated such that it meets both the peak of flow and the basin lag time.

The hydrologic model was calibrated using the observed and modeled hydrograph. The model could not be used to model long term recession which impacts the groundwater storage and characteristics of the outflow. The objective of the study was to choose the HRU where if BMP applied would reduce the peak flow at the watershed outlet. As only one type of BMP is considered therefore project cost is directly proportional to the number of BMP selected. The study used GA approach to analyze different groups of BMP location so that the peak flow of a October storm is reduced. Initially every HRU was considered to be having BMP which was found out to be infeasible, so they restricted the location of BMP based on the CN which targets the HRU which is more impervious and which will eventually cause more runoff into the stream network. GA was used to maximize the peak flow reduction for a given budget constraint that total number of BMPs selected should not exceed than that of the preselected.

By continuous optimization with varied number of BMP’s i.e varied costs, a tradeoff (Pareto frontier) curve was obtained between the cost and the peak flow reduction. All the solutions below this curve are feasible and all the solutions above the curve as infeasible. The paper recommends for a distributed physical representation of the basin for carrying out BMP planning analysis. It was concluded that for the present area of study by applying BMP at not less than 200 HRU we can achieve 20% reduction in peak flow. It is illustrated by the trade off curve that with the increase in the number of BMP’s the returns are decreased. Author suggests using the incremental approach in designing which is BMP is introduced at places which are most affected by the storm and then moving on to the places which aren’t that critical, such an approach would be helpful in targeting resource management.

Discussion:
The paper was very helpful in giving an insight of how the GA optimization could be carried out. As the authors specify that physical representation of the basin is required for the planning, taking up a research to study that and comparing the results with this model which did not take into the consideration without a physical model would be worth knowing. The other aspect would be considering how the cost optimization results will be affected by considering different BMP’s based on the land use pattern of each HRU.