APPLICATION OF GREY MODEL AND ARTIFICIAL NEURAL NETWORKS TO FLOOD FORECASTING1

The main focus of this study was to compare the Grey model and several artificial neural network (ANN) models for real time flood forecasting, including a comparison of the models for various lead times (ranging from one to six hours). For hydrological applications, the Grey model has the advantage that it can easily be used in forecasting without assuming that forecast storm events exhibit the same stochastic characteristics as the storm events themselves. The major advantage of an ANN in rainfall‐runoff modeling is that there is no requirement for any prior assumptions regarding the processes involved. The Grey model and three ANN models were applied to a 2,509 km² watershed in the Republic of Korea to compare the results for real time flood forecasting with from one to six hours of lead time. The fifth‐order Grey model and the ANN models with the optimal network architectures, represented by ANN1004 (34 input nodes, 21 hidden nodes, and 1 output node), ANN1010 (40 input nodes, 25 hidden nodes, and 1 output node), and ANN1004T (14 input nodes, 21 hidden nodes, and 1 output node), were adopted to evaluate the effects of time lags and differences between area mean and point rainfall. The Grey model and the ANN models, which provided reliable forecasts with one to six hours of lead time, were calibrated and their datasets validated. The results showed that the Grey model and the ANN1010 model achieved the highest level of performance in forecasting runoff for one to six lead hours. The ANN model architectures (ANN1004 and ANN1010) that used point rainfall data performed better than the model that used mean rainfall data (ANN1004T) in the real time forecasting. The selected models thus appear to be a useful tool for flood forecasting in Korea.     

COMPARISON OF PROCESS‐BASED AND ARTIFICIAL NEURAL NETWORK APPROACHES FOR STREAMFLOW MODELING IN AN AGRICULTURAL WATERSHED1

ABSTRACT: The performance of the Soil and Water Assessment Tool (SWAT) and artificial neural network (ANN) models in simulating hydrologic response was assessed in an agricultural watershed in southeastern Pennsylvania. All of the performance evaluation measures including Nash‐Sutcliffe coefficient of efficiency (E) and coefficient of determination (R²) suggest that the ANN monthly predictions were closer to the observed flows than the monthly predictions from the SWAT model. More specifically, monthly streamflow E and R² were 0.54 and 0.57, respectively, for the SWAT model calibration period, and 0.71 and 0.75, respectively, for the ANN model training period. For the validation period, these values were ?0.17 and 0.34 for the SWAT and 0.43 and 0.45 for the ANN model. SWAT model performance was affected by snowmelt events during winter months and by the model's inability to adequately simulate base flows. Even though this and other studies using ANN models suggest that these models provide a viable alternative approach for hydrologic and water quality modeling, ANN models in their current form are not spatially distributed watershed modeling systems. However, considering the promising performance of the simple ANN model, this study suggests that the ANN approach warrants further development to explicitly address the spatial distribution of hydrologic/water quality processes within watersheds.     

EFFECTS OF INITIAL ABSTRACTION AND URBANIZATION ON ESTIMATED RUNOFF USING CN TECHNOLOGY1

ABSTRACT: Few studies have been conducted to explore the effects of initial abstraction on estimated direct runoff despite the widespread use of the curve number (CN) method in many hydrologic models to estimate direct runoff. In this study, use of a 5 percent ratio of initial abstraction (I_a) to storage (S) to estimate daily direct runoff with modified CN values for a 5 percent I_a/S value was investigated using the Long‐Term Hydrologic Impact Assessment (L‐THIA) geographic information system (GIS). In addition, the effects on estimated runoff of altering the hydrologic soil group due to urbanization were investigated. The L‐THIA model was applied to the Indiana Little Eagle Creek watershed with 5 percent and 20 percent I_a/S values, considering hydrologic soil group alteration due to urbanization. The results indicate that uses of a 5 percent l_a/S and modified CN values and Hydrologic Soil Group D for urbanized areas in model runs can improve long term direct runoff prediction.     

ASSESSMENT OF POSSIBLE IMPACTS OF CLIMATE CHANGE IN AN URBAN CATCHMENT1

ABSTRACT: Stationarity of rainfall statistical parameters is a fundamental assumption in hydraulic infrastructure design that may not be valid in an era of changing climate. This study develops a framework for examining the potential impacts of future increases in short duration rainfall intensity on urban infrastructure and natural ecosystems of small watersheds and demonstrates this approach for the Mission/Wagg Creek watershed in British Columbia, Canada. Nonstationarities in rainfall records are first analyzed with linear regression analysis, and the detected trends are extrapolated to build potential future rainfall scenarios. The Storm Water Management Model (SWMM) is used to analyze the effects of increased rainfall intensity on design peak flows and to assess future drainage infrastructure capacity according to the derived scenarios. While the framework provided herein may be modified for cases in which more complex distributions for rainfall intensity are needed and more sophisticated stormwater management models are available, linear regressions and SWMM are commonly used in practice and are applicable for the Mission/Wagg Creek watershed. Potential future impacts on stream health are assessed using methods based on equivalent total impervious area. In terms of impacts on the drainage infrastructure, the results of this study indicate that increases in short duration rainfall intensity may be expected in the future but that they would not create severe impacts in the Mission/Wagg Creek system. The equivalent levels of imperviousness, however, suggest that the impacts on stream health could be far more damaging.     

AN INTEGRATED ONE‐DIMENSIONAL AND TWO‐DIMENSIONAL URBAN STORMWATER FLOOD SIMULATION MODEL1

ABSTRACT: Flash flooding is the rapid flooding of low lying areas caused by the stormwater of intense rainfall associated with thunderstorms. Flash flooding occurs in many urban areas with relatively flat terrain and can result in severe property damage as well as the loss of lives. In this paper, an integrated one‐dimensional (1‐D) and two‐dimensional (2‐D) hydraulic simulation model has been established to simulate stormwater flooding processes in urban areas. With rainfall input, the model simulates 2‐D overland flow and 1‐D flow in underground stormwater pipes and drainage channels. Drainage channels are treated as special flow paths and arranged along one or more sides of a 2‐D computational grid. By using irregular computation grids, the model simulates unsteady flooding and drying processes over urban areas with complex drainage systems. The model results can provide spatial flood risk information (e.g., water depth, inundation time and flow velocity during flooding). The model was applied to the City of Beaumont, Texas, and validated with the recorded rainfall and runoff data from Tropical Storm Allison with good agreement.     

HYDROLOGIC BEHAVIOR OF VEGETATED ROOFS1

ABSTRACT: Control of stormwater runoff from impervious surfaces is an important national goal because of disruptions to downstream ecosystems, water users, and property owners caused by increased flows and degraded quality. One method for reducing stormwater is the use of vegetated (green) roofs, which efficiently detain and retain stormwater when compared to conventional (black) roofs. A paired green roof‐black roof test plot was constructed at the University of Georgia and monitored between November 2003 and November 2004 for the green roof's effectiveness in reducing stormwater flows. Stormwater mitigation performance was monitored for 31 precipitation events, which ranged in depth from 0.28 to 8.43 cm. Green roof precipitation retention decreased with precipitation depth; ranging from just under 90 percent for small storms (< 2.54 cm) to slightly less than 50 percent for larger storms (> 7.62 cm). Runoff from the green roof was delayed; average runoff lag times increased from 17.0 minutes for the black roof to 34.9 minutes for the green roof, an average increase of 17.9 minutes. Precipitation and runoff data were used to estimate the green roof curve number, CN = 86. This information can be used in hydrologic models for developing stormwater mitigation programs.     

HYDROLOGIC MODELING OF A BIOINFILTRATION BEST MANAGEMENT PRACTICE1

ABSTRACT: The goal of this research was to develop a methodology for modeling a bioinfiltration best management practice (BMP) built in a dormitory area on the campus of Villanova University in Pennsylvania. The objectives were to quantify the behavior of the BMP through the different seasons and rainfall events; better understand the physical processes governing the system's behavior; and develop design criteria. The BMP was constructed in 2001 by excavating within an existing traffic island, backfilling with a sand/soil mixture, and planting with salt tolerant grasses and shrubs native to the Atlantic shore. It receives runoff from the asphalt (0.26 hectare) and turf (0.27 hectare) surfaces of the watershed. Monitoring supported by the hydrologic model shows that the facility infiltrates a significant fraction of the annual precipitation, substantially reducing the delivery of nonpoint source pollution and erosive surges downstream. A hydrologic model was developed using HEC‐HMS to represent the site and the BMP using Green‐Ampt and kinematic wave methods. Instruments allow comparison of the modeled and measured water budget parameters. The model, incorporating seasonally variable parameters, predicts the volumes infiltrated and bypassed by the BMP, confirming the applicability of the selected methods for the analysis of bioinfiltration BMPs.     

HEAVY RAINSTORMS IN CHICAGO: INCREASING FREQUENCY,ALTERED IMPACTS,AND FUTURE IMPLICATIONS1

ABSTRACT: Operations of a dense raingage network in the Chicago area since 1989 provided data to assess the temporal and spatial distributions of heavy rainstorms. The 12‐year average was 4.4 storms per year, 40 percent more than in the 1948 to 1980 period, reflecting an ongoing Midwestern increase in heavy rains. The total rainfall from the 53 heavy rainstorms maximized over the city, reflecting previous observations that the influence of the city and Lake Michigan on the atmosphere causes an increase in heavy rains. Impacts from the record high number of eight storms in 2001 revealed that efforts to control flooding including the Deep Tunnel system, had reduced street and basement flooding in the moderate intensity storms, but the two most intense storms, each with 100‐year rainfall values, led to excessive flooding and a need to release flood waters into Lake Michigan. Results suggest continuing increases in the number of heavy rainstorms in future years, which has major implications for water managers in Chicago and elsewhere.     

DISCHARGE CHARACTERISTICS OF PERFORATED PIPE FOR USE IN INFILTRATION TRENCHES1

ABSTRACT: Infiltration trenches are an effective stormwater management alternative for the control of urban runoff from small areas. Perforated pipes buried within the gravel of an infiltration trench are used to distribute the inflowing runoff along the length of the trench. Laboratory tests are described that characterize the hydraulics of the orifices in perforated pipes. The results show that the steady-state exfiltration of water from the pipe into a surrounding gravel trench can be described by the orifice equation.     

EFFECTS OF SLUDGE AND CHEMICAL FERTILIZER APPLICATION ON RUNOFF WATER QUALITY1

ABSTRACT: Simulated rainfall was used on experimental field plots to compare the effect of chemical fertilizer and sludge application on sediment, nitrogen, and phosphorus in runoff from no-till and conventional tillage systems. Chemical fertilizer application under the no-till system resulted in the least amount of total N and P in surface runoff. However, sludge application under the no-till system resulted in the least amount of NO₃-N and sediment in surface runoff. The worst water quality scenarios were observed when either sludge or chemical fertilizer were surface-applied under a conventional tillage system. Nitrogen losses from the conventional tillage system were minimized when sludge was incorporated into the soil. However, phosphorus and sediment yield from such a system were significantly higher when compared to phosphorus and sediment yield from the no-till system. The results from this study indicate that the use of sludge on agricultural land under a no-till system can be a viable alternative to chemical fertilizer for nitrogen and phosphorus control in runoff. A more cautious approach is recommended when the sludge is incorporated into the soil in a conventional tillage system because of potential for high sediment and phosphorus yield in surface runoff.