Calibration of Hydrologic Design Inputs for a Small Urban Watershed

Transcription

1 Research Project Proposal for Johnson County Stormwater Management Program and City of Overland Park, Kansas Calibration of Hydrologic Design Inputs for a Small Urban Watershed Bruce M. McEnroe & C. Bryan Young Department of Civil, Environmental and Architectural Engineering University of Kansas Introduction The design of stormwater drainage, detention and BMP systems requires realistic values of hydrologic inputs such as Rational C factors, runoff curve numbers and volumetric runoff coefficients. These inputs are strongly dependent on local climate as well as land cover and soil conditions; therefore, their values vary geographically. Tables of runoff coefficients and curve numbers published by national organizations such as ASCE and NRCS do not account for this geographic variability. The recommended values in the KC-APWA Section 5600 Specifications and the MARC BMP Manual were taken from these standard references. Their applicability to the Kansas City metro area is uncertain. However, a recent study found that the Rational C factors for rural Kansas watersheds are strongly dependent on geographic location and recurrence interval (Young and McEnroe, 2009). The data records from one particular ALERT site provide an opportunity to determine actual values of hydrologic inputs for a small urban watershed in the Kansas City area. The City of Overland Park and Johnson County had the foresight to install ALERT precipitation and water-level gages on the 178-acre Wilshire Woods watershed in The gages are located at a box culvert on a small tributary of Tomahawk Creek at 127 th Terrace west of Grant Street. The Wilshire Woods watershed is the smallest gaged watershed in the ALERT system. It is also much smaller than the USGS-gaged watershed in the Kansas City area. The land use is mostly single-family residential with two schools and a large church. The drainage system consists of curb-and-gutter streets and concrete pipes and box conduits, with only 300 feet of open channel upstream of the gages. The data archives now contain twelve years of continuous precipitation and water-level data. These data are much more detailed than the data recorded at USGS gages. The ALERT gages record the exact times of each millimeter of precipitation and each small change in stage. However, the usefulness of the data is limited by the lack of a reliable rating curve to convert the stage record to a discharge record. Objectives and Benefits 1. We will develop a reliable stage-discharge relationship (rating curve) for the gage by means of a physical hydraulic model study. We will use this relationship to convert the archived stage record to a discharge record. A frequency analysis of this record will yield estimates of discharges with return periods up to 10 years. With a reliable rating curve, the ALERT data-processing system could be programmed to report and archive discharge as well as stage. 2

2 2. We will analyze the precipitation, stage, and discharge records to determine the values of the following hydrologic design inputs: Rational C values for return periods of 2, 5 and 10 years NRCS runoff curve numbers for antecedent moisture conditions AMC I, II and III Average volumetric runoff coefficient for the 1.37-inch water-quality storm Hydrologic lag times for events of different magnitudes. We will compare the hydrologic design inputs derived from the data record with the values recommended in KC-APWA Section 5600 and the MARC BMP Manual, and examine the practical implications of any substantial differences. These findings will be of interest to the KC-APWA and MARC committees responsible for the Section 5600 specifications and the BMP manual. This project might lead to better stormwater design practices for the Kansas City area or provide support for current practices. Required Tasks and Methods 1. Develop a rating curve for the stage gage The Wilshire Woods stage gage is located on the wing-wall of a box culvert at 127 th Terrace. The culvert is a two-cell 6 x 6 RCB culvert with a square-edged headwall, wing-walls at 45º angles, and straight entrance conditions. The channel downstream of the culvert is wide and fairly clean with a sizeable drop in the channel bed a short distance downstream. It is clear that culvert s headwater-discharge relationship would be unaffected by the tailwater level; i.e., the culvert would operate under inlet control. FHWA has performed model tests on culverts of this design and developed equations for the rating curve, with the culvert s span and rise as the only inputs. However, these equations relate the discharge to the total head upstream of the culvert, i.e., the stage at a point upstream where the velocity is negligible. The Wilshire Woods stage gage consists of a plastic tube with an open bottom attached to the culvert wing-wall approximately one foot from the culvert entrance. The bottom of the tube is set near the level of the culvert floor, except during winter months when it is raised to avoid ice damage. A pressure transducer inside the tube transmits a voltage, which the ALERT system converts to a water level. The reported water level is actually the piezometric head at the bottom of the tube. The piezometric head at this location is lower than the total head upstream of the culvert for two reasons: the velocity is relatively high and the vertical distribution of pressure is non-hydrostatic due to the downward convective acceleration of the flow. At moderate and high flows, the difference is relatively large and cannot be neglected. We propose to establish the relationship between the piezometric head at the measurement point and the total head upstream of the culvert by means of a physical hydraulic model study. A 1:10 scale model of the culvert and the channel directly upstream will be constructed and tested in the 2.5-foot-wide re-circulating flume in the University of Kansas hydraulics laboratory. A piezometer tube will be mounted on the wing-wall of the model culvert in the position of the actual stage gage. The piezometric 3

3 head difference between the wing-wall piezometer and a stilling well upstream of the culvert will be measured with a differential pressure transducer connected to a data acquisition device. We will determine the relationship between the heads at the two locations over the range of possible flows conditions. This relationship will be combined with the FHWA equations to obtain the desired rating curve for the ALERT stage gage. 2. Quality control of rainfall and stage data The Wilshire Woods rainfall and stage records are nearly complete and appear to be of good quality generally. However, the record does include some brief periods during which the data are obviously incorrect due to various malfunctions. We will carefully review the rainfall and stage records in tandem to identify anomalous data and make appropriate corrections wherever possible. 3. Streamflow record and frequency analysis A discharge record will be constructed from the corrected stage record and the previously developed rating curve. A frequency analysis will be performed on the discharge record. Appropriate adjustments will be made for any gaps in the data record. We will estimate discharges with return periods up to 10 years and place statistical confidence limits on these estimates. 4. Watershed characteristics Watershed characteristics relevant to hydrologic behavior will be determined from Johnson County s digital planimetric data and other resources. These characteristics include the area of impervious surfaces, land-cover classifications, soil hydrologic classifications, and lengths and slopes of overland flow, gutter flow, pipe flow and channel flow. 5. Rainfall-runoff analysis All rainfall events exceeding a threshold depth of 0.50 inches will be identified. The total depth of rainfall and the corresponding depth of runoff for each event will be computed from the rainfall and discharge records. We will plot a graph of runoff depth versus rainfall depth for these significant events. This graph will exhibit considerable scatter due to differences in storm characteristics (intensities, durations and temporal patterns), antecedent moisture conditions, and seasonal factors; however, a definite trend will be evident. 6. Volumetric runoff coefficient for water-quality event Stormwater BMPs are designed to store or treat the runoff from a specified water-quality storm event. The runoff depth is determined by multiplying the water-quality storm depth by a volumetric runoff coefficient. The water-quality storm depth for the Kansas City area is 1.37 inches. We will determine the average volumetric runoff coefficient for a 4

4 1.37-inch rainfall through a statistical analysis of the rainfall and runoff depths for individual events from Task 5. This runoff coefficient will be compared with the value that the BMP Manual recommends for this type of watershed. 7. Runoff curve numbers for AMC I, II and III The runoff curve numbers for antecedent moisture conditions I, II and III (CN I, CN II and CN III ) will be determined by plotting NRCS rainfall-runoff relationships for different curve numbers on the graph of runoff depth versus rainfall depth from Task 5. According to the NRCS, CN II is the curve number for which equal numbers of data points lie above and below the plotted curve. CN I is the curve number for which 90% of the data points lie above the curve. CN III is the curve number for which 10% of the data points lie above the curve. We will compare these curve numbers with the values recommended for this type of watershed in Section 5600 and in NRCS guidance. 8 Rational C values for return periods of 2, 5 and 10 years. Rational C values for return periods 2, 5 and 10 years will be computed from the estimated discharges for these return periods from Task 3, and rainfall intensities for these return periods from KC-APWA s rainfall intensity-duration-frequency table for the Kansas City area (Young and McEnroe, 2002). The rainfall duration will be set equal to the watershed s time of concentration computed by the method in Section We will compare these Rational C values with the value that Section 5600 recommends for this type of watershed. 9. Lag times Lag time is an essential input to flood hydrograph simulation by the NRCS unithydrograph method. It is defined as the time interval between the occurrence of a short, intense burst of runoff-producing rainfall and the arrival of the peak discharge at the watershed outlet (gage). We will determine lag times for significant rainfall events by calibration of a flood-hydrograph simulation model using the ALERT gage data. Rainfall data at one-minute intervals will be input to the model, and the lag time will be adjusted so that the computed peak discharge occurs at the same time as the recorded peak stage. We will compare the lag times for different events and examine whether lag time varies with event magnitude. The lag times derived from the gage data will be compared with the lag time predicted by the regression equation for urban watersheds in the KDOT Design Manual (KDOT, 2008; McEnroe and Zhao, 2001). We will also investigate validity of the lag-time method in Section 5600, in which lag time is set equal to threefifths of the time of concentration computed by the Section 5600 method. 10. Analysis of findings We will examine the practical implications of any substantial differences between the hydrologic design inputs derived from the gage records and the values recommended in Section 5600 and the BMP Manual. 5

5 Project Deliverables 1. Fully documented report with digital appendices 2. Oral presentation on project and findings, upon completion Project Duration 12 months Project Cost $65,000 PROPOSED BUDGET Year 1: 10/01/10 to 09/30/11 SALARIES AND WAGES Senior Personnel % time Months Rate Bruce McEnroe, Co-I summer ,571 10,720 C. Bryan Young, Co-I summer ,680 8,968 Total senior personnel 19,688 Other Personnel Undergraduate Student(s) Persons Hours Rate calendar ,403 Total other personnel 15,403 Total salaries and wages 35,091 FRINGE BENEFITS 34% faculty and staff 6,694 6% students (employed 75% or less) 924 Total fringe benefits 7,618 Total salaries, wages & fringe benefits 42,709 TRAVEL (b) regional travel # Persons Trips Days Amount Transportation ( Total (b) 500 Total travel 500 OTHER DIRECT COSTS Research materials & supplies 1,009 Total Other Direct Costs 1,009 TOTAL DIRECT COSTS 44,218 BASE 44,218 INDIRECT COSTS (47% of total direct costs excluding equipment and tuition allowance) 20,782 TOTAL PROPOSED COSTS - YEAR 1 $65,000 6

6 Budget Justification Senior Personnel Funds are requested for three summer months at % effort for Professors Bruce M. McEnroe and C. Bryan Young, the co-principal investigators. The co-principal investigators will be responsible for performance of all tasks listed the research proposal, timely delivery of the final report and presentation, and supervision of research assistants. Other Personnel Two or more civil engineering students will be employed as research assistants on an hourly basis for a total of 1101 hours over the 12-month project period. The research assistants will be responsible for assisting with construction and testing of a hydraulic model and for hydrologic data analysis. Fringe Benefits Fringe benefits are calculated at 34% for faculty and staff, 14% for graduate students working 76% or more, and 6% for graduate students working 75% or less. Travel Project personnel will make a total of 12 day-trips to the research site in Overland Park and to meetings at the offices of the Overland Park and Johnson County public works departments. Expenses cover mileage for travel in personal vehicles. Other Direct Costs $1009 is requested to purchase materials and supplies for construction and testing of a hydraulic model of a box culvert, and for other miscellaneous supplies needed to conduct the research. Facilities & Administrative Costs Facilities & Administrative (F&A) costs are calculated at 47% of modified total direct costs (MTDC), where MTDC equals total direct costs, less equipment, tuition, and subcontract amounts over $25,000. References Kansas Department of Transportation (2008). KDOT Design Manual, Vol. I, Road Section, Part C, Elements of Drainage and Culvert Design McEnroe, B. M and H. Zhao (2001). Lag Times of Urban and Developing Watersheds in Johnson County, Kansas, Report No. K-TRAN: KU-99-5, Kansas Department of Transportation. Young, C. B., B. M. McEnroe and A. C. Rome (2009). Empirical Determination of Rational Method Runoff Coefficients, Journal of Hydrologic Engineering, Vol. 14, No. 12, 2009, pp Young, C. B. and B. M. McEnroe (2002). Precipitation Frequency Estimates for the Kansas City Metropolitan Area, Kansas City Metropolitan Chapter of the American Public Works Association. 7