ESTIMATION OF THE HIEC1 LOSS PARAMETERS FOR ROUTING THE PROBABLE MAXIMUM FLOOD1

ABSTRACT: Twenty storm events were used to select design values of the HEC1 loss parameters STRTL and CNSTL in order to route the probable maximum flood, PMF, through the Englewood watershed, Ohio. The parameter STRTL represents the initial volume of water lost due to interception and incomplete saturation of the soil prior to the storm. The parameter CNSTL represents a continuous loss rate and depends only on the watershed. When optimized from each storm event, STRTL varied between 0.0 and 3.4 inches with an average of 1.0 inch; CNSTL varied between 0.02 and 0.26 inch/hour, and it followed a normal probability distribution with a mean of about 0.1 inch/hour. The absence of correlation between optimum CNSTL values and each of total rainfall, total loss, and runoff duration supported the selection of the mean CNSTL as a design value. PMF routing through the Englewood watershed revealed that the PMF at the outlet is not sensitive to STRTL, but highly affected by CNSTL variations. The insensitivity to STRTL was due to the presence of a dam at the outlet of the watershed that caused the buildup of water in the watershed, thereby masking the storage effect of STRTL. The peak PMF increased by about 27 percent when the design CNSTL was decreased to 0.05 inch! hour, and decreased by about 18 percent when the design CNSTL was increased to 0.15 inch/hour.     

APPLICATION OF THE GREEN-AMPT INFILTRATION EQUATION TO WATERSHED MODELING1

ABSTRACT: A computer model was developed, based on the Green-Ampt infiltration equation, to computed rainfall excess for a single precipitation event. The model requires an estimate of parameters related to hydraulic conductivity, wetting front section, and fillable porosity of the soil layers. Values of parameters were estimated from soil textural averages or regression equations based on percent sand, percent clay, and porosity. Average values of effective porosity and wetting front suction were largely acceptable due to the relatively low variability and low model sensitivity to the parameters. Hydraulic conductivity was the most erratic constituent of the loss rate computation due to the high variability and the high sensitivity of the computed infiltration to the parameter. The performance of the Green-Ampt infiltration model was tested through a comparison with the SCS curve number procedure. Seven watersheds and 23 storms with precipitation of one inch or greater were used in the comparison. For storms with less than one inch of rainfall excess, the SCS curve number procedure generally gave the best results; however, for six of the seven storms with precipitation excess greater than one inch, the Green-Ampt procedure delivered better results. In this comparison, both procedures used the same initial abstractions. The separation of rainfall losses into infiltration, interception, and surface retention is, in theory, an accurate method of estimating precipitation excess. In the second phase of the study using nine watersheds and 39 storms, interception and surface retention losses were computed by the Horton equations. Green-Ampt and interception parameters were estimated from value sin the literature, while the surface retention parameter was calibrated so that the computed runoff volumes matched observed volumes. A relationship was found between the surface retention storage capacity and the 15-day antecedent precipitation index, month of year, and precipitation amount.     

APPLICATION OF PARAMETER OPTIMIZATION METHOD FOR CALIBRATING TANK MODEL1

ABSTRACT: Many automatic calibration processes have been proposed to efficiently calibrate the 16 parameters involved in the four‐layered tank model. The Multistart Powell and Stuffed Complex Evolution (SCE) methods are considered the best two procedures. Two rainfall events were designed to compare the performance and efficiency of these two methods. The first rainfall event is short term and the second designed for long term rainfall data collection. Both rainfall events include a lengthy no‐rainfall period. Two sets of upper and lower values for the search range were selected for the numerical tests. The results show that the Multistart Powell and SCE methods are able to obtain the true values for the 16 parameters with a sufficiently long no‐rainfall period after a rainfall event. In addition, by using two selected objective functions, one based on root mean square error and one based on root mean square relative error criteria, it is found that the no‐rainfall period lengths necessary to obtain the converged true values for the 16 parameters are roughly the same. The SCE method provides a more efficient search based on an appropriate preliminary search range. The Multistart Powell method, on the other hand, leads to more accurate search results when there is no suitable search range selected based on the parameter calibration experience.     

A PRECIPITATION ESTIMATION TECHNIQUE FOR DEVELOPING ISOHYETS IN AN ARID AREA USING VEGETATION AND TOPOGRAPHIC PARAMETERS1

A technique is presented for developing an isohyet map for the Hualapai Valley, a closed hydrologic basin of about 315 square miles in the northwestern Great Basin in Nevada. In this basin there is practically no climatic data, and in the northwest Great Basin there are too few stations for determination of rainfall on a detailed basis. Using a vegetational typing to represent a range in elevation and precipitation, an initial mean annual rainfall is determined for selected points on a grid pattern. This rainfall is then modified by using topographic parameters of slope, orientation, exposure, and rainfall shadow effect. The resulting point determinations of mean annual rainfall are then smoothed using a trend surface analysis, and an isohyetal map is drawn from the smoothed points. The technique provides an estimated accuracy of one inch of mean annual precipitation and one mile of resolution on isohyets.     

HEAVY RAINFALL DISTRIBUTIONS BY SEASON IN LOUISIANA: SYNOPTIC INTERPRETATIONS AND QUANTILE ESTIMATES1

ABSTRACT: In most studies, quantile estimates of extreme 24-hour rainfall are given in annual probabilities. The probability of experiencing an excessive storm event, however, differs throughout the year. As a result, this paper explored the differences between heavy rainfall distributions by season in Louisiana. It was concluded by using the Kruskal-Wallis and Mann-Whitney tests that the distribution of heavy rainfall events differs significantly between particular seasons at the sites near the Gulf Coast. Furthermore, seasonal frequency curves varied dramatically at the four sites examined. Mixed distributions within these data were not found to be problematic, but the mechanisms that produced the events were found to change seasonally. Extreme heavy rainfall events in winter and spring were primarily generated by frontal weather systems, while summer and fall events had high proportions of events produced by tropical disturbances and airmass (free-convective) conditions.     

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.     

RUNOFF MODELING OF A MOUNTAINOUS CATCHMENT USING TOPMODEL: A CASE STUDY1

ABSTRACT: The rainfall‐runoff response of the Tygarts Creek Catchment in eastern Kentucky is studied using TOPMODEL, a hydrologic model that simulates runoff at the catchment outlet based on the concepts of saturation excess overland flow and subsurface flow. Unlike the traditional application of this model to continuous rainfall‐runoff data, the use of TOPMOEL in single event runoff modeling, specifically floods, is explored here. TOPMODEL utilizes a topographic index as an indicator of the likely spatial distribution of rainfall excess generation in the catchment. The topographic index values within the catchment are determined using the digital terrain analysis procedures in conjunction with digital elevation model (DEM) data. Select parameters in TOPMODEL are calibrated using an iterative procedure to obtain the best‐fit runoff hydrograph. The calibrated parameters are the surface transmissivity, T_O, the transmissivity decay parameter, m, and the initial moisture deficit in the root zone, S_r0. These parameters are calibrated using three storm events and verified using three additional storm events. Overall, the calibration results obtained in this study are in general agreement with the results documented from previous studies using TOPMODEL. However, the parameter values did not perform well during the verification phase of this study.     

PREDICTING MEASUREMENT ERROR OF AREAL MEAN PRECIPITATION DURING EXTREME EVENTS1

ABSTRACT: The areal mean precipitation (AMP) over a catchment is normally calculated using point measurements at rainfall gages. Error in AMP estimates occurs when an insufficient number of gages are used to sample precipitation which is highly variable in space. AMP error is investigated using historic, severe rainfalls with a set of hypothetical catchments and raingage networks. The potential magnitude of error is estimated for typical gage network densities and arrangements. Possible sources of error are evaluated, and a method is proposed for predicting the magnitude of error using data that are commonly available for severe, historic rainfall.     

FLOOD EVENT MODELING - A STUDY OF TWO METHODS1

: Despite the advances in catchment modeling in recent years, engineers still face major problems in estimating flood flows. Application of unit hydrograph and runoff routing models to five United Kingdom catchments shows that either can be tuned to predict, on a test event, the routing effects of a catchment with equal accuracy. The larger remaining problem is the prediction of losses from rainfall and this study shows how alternative ways of describing the within event distribution of these losses can be an important factor controlling the success of the overall model. Other problems include the risks of extrapolation to larger events, baseflow separation methods, rainfall patterns, and inevitable errors in the data.     

TEMPORAL AND SPATIAL CHARACTERIZATION OF RAINFALL OVER CENTRAL AND SOUTH FLORIDA1

requency evaluation and spatial characterization of rainfall in Central and South Florida are presented. Point frequency analysis performed at all available sites has shown that the 2‐parameter Gamma probability density function is the best model for monthly rainfall frequency over Central and South Florida. The model's parameters estimated at 145 stations were used to provide monthly rainfall estimates for 10‐ and 100‐year dry and wet return periods. Experimental and theoretical variograms computed for these estimates, as well as the Kriging estimation variance maps, show that the existing rain gage network is less capable of resolving monthly rainfall variation in the wet season than the dry season. May is the dry‐to‐wet transition month, while October is the wet‐to‐dry transition month with average rainfall of 4.5 inches. Monthly average rainfall is above 7 inches during the wet season and below 3 inches during the dry season. Two‐thirds of the annual rainfall is accumulated in the wet season. Annual average rainfall is maximum (above 60 inches) in many areas along the east coast, and is minimum (below 45 inches) in many areas over Lake Okee‐chobee and Central Florida. Rainfall maps show a changing pattern between the wet and the dry seasons. Frontal rainfall occurs in the dry season, while convective rainfall, tropical depression, and hurricanes occur in the wet season. Average rainfall is higher along the east coast area in the dry season and it is higher along the west coast area in the wet season.     

EVALUATING THE ACCURACY OF RAINFALL CATCH BY THREE DIFFERENT GAGES1

ABSTRACT: A U.S. standard gage, a weighing-type recording gage, a standard gage fitted with an Alter windshield, and a pit gage were installed to evaluate the accuracy and wind effects on rainfall catch. The study was conducted at the Stephen F. Austin Experimental Forest, about 20 km SW of Nacogdoches, Texas. A recording anemometer was also installed at a height corresponding to the standard gage orifice. Based on data from 67 storms collected over a one-year period (July 1995-August 1996), all three conventional gages consistently caught less rainfall than the reference pit gage with an average percent deficiency greater than 10 percent. However, the recording gage caught 2.7 percent less and the shielded gage caught 1 percent more than the standard gage—differences less than those reported elsewhere. The deficiencies were highly correlated with storm intensity, duration, or total rainfall. When the correction for wind effect on angle of raindrop inclination is included, the percent catch deficiency of the standard gage was reduced from 11 percent to 6 percent. The remaining errors may be attributed to wind effects (streamline vs. turbulent flow), nonrandom errors, or other unknown sources.     

RELIABILITY OF THE DESIGN STORM CONCEPT IN EVALUATING RUNOFF PEAK FLOW1

A comparative study was undertaken to evaluate peak runoff flow rates using (1) a continuous series of actual rainfall events and (2) design storms. The ILLUDAS computer model was used to simulate runoff over a catchment within the city of Montreal, Canada. A ten-year period, five-minute increment rainfall data base was used to derive peak flow frequency curves. Two types of design storms were analyzed: one derived from intensity duration frequency curves (Chicago type), the other from averaging actual rainfall patterns (Huff type). Antecedent soil moisture conditions were considered in the analyses. It was found that the probability distribution of runoff peak flow was sensitive to the choice of design storm pattern and to the antecedent soil moisture condition. A symmetrical, Chicago-type design storm with antecedent dry soil moisture produced a flow frequency curve similar to the one obtained from a series of historical rainfall events.     

PREDICTING STORMFLOW AND PEAKFLOW FROM SMALL BASINS IN HUMID AREAS BY THE R-INDEX METHOD1

ABSTRACT: A relatively simple nonlinear equation was fitted to 468 stormflows larger than 0.05 area inches on 11 forested basins from New Hampshire to South Carolina, providing a predictive method for use on forest and wildlands in humid regions. Stormflow in area inches (Q?) was: where R is the mean value of Q/P for all P larger than one inch, P is storm rainfall in inches, and I is the initial flow rate in ft³/sec/mi². S.E. was 0.3 inch of stormflow. Peakflow was similarly estimated, S.E. 26 ft³/sec/mi². The R-index method is proposed as a practical tool in forest and wildland management. Similar to the SCS runoff curve number method, the R-index method requires no prior assumptions about infiltration capacities of forest lands, but calls for the mapping of all first-order streams for the average storage capacity index R, i.e., the mean hydrologic response of the source areas. Tested against the runoff curve method on four independent basins, predictions by the R-index method were considerably more accurate when field information normally available to planners and managers was used in both methods.     

Curve Number Derivation for Watersheds Draining Two Headwater Streams in Lower Coastal Plain South Carolina,USA

The objective of this study was to assess curve number (CN) values derived for two forested headwater catchments in the Lower Coastal Plain (LCP) of South Carolina using a three‐year period of storm event rainfall and runoff data in comparison with results obtained from CN method calculations. Derived CNs from rainfall/runoff pairs ranged from 46 to 90 for the Upper Debidue Creek (UDC) watershed and from 42 to 89 for the Watershed 80 (WS80). However, runoff generation from storm events was strongly related to water table elevation, where seasonally variable evapotranspirative wet and dry moisture conditions persist. Seasonal water table fluctuation is independent of, but can be compounded by, wet conditions that occur as a result of prior storm events, further complicating flow prediction. Runoff predictions for LCP first‐order watersheds do not compare closely to measured flow under the average moisture condition normally associated with the CN method. In this study, however, results show improvement in flow predictions using CNs adjusted for antecedent runoff conditions and based on water table position. These results indicate that adaptations of CN model parameters are required for reliable flow predictions for these LCP catchments with shallow water tables. Low gradient topography and shallow water table characteristics of LCP watersheds allow for unique hydrologic conditions that must be assessed and managed differently than higher gradient watersheds.     

A SEMI-DISTRIBUTED ADAPTIVE MODEL FOR REAL-TIME FLOOD FORECASTING1

ABSTRACT: A semi-distributed deterministic model for real-time flood forecasting in large basins is proposed. Variability of rainfall and losses in space is preserved and the effective rainfall-direct runoff model segment based on the Clark procedure is incorporated. The distribution of losses in space is assumed proportional to rainfall intensity and their evolution in time is represented by the φ-index; furthermore, an initial period without production of effective rainfall is considered. The first estimation of losses and the associated forecasts of flow are performed at the time corresponding to the first rise observed in the hydrograph. Then the forecasts of flow are corrected at each subsequent time step through the updating of the φ-index. The model was tested by using rainfall-runoff events observed on two Italian basins and the predictions of flow for lead times up to six hours agree reasonably well with the observations in each event. For example, for the coefficient of persistence, which compares the model forecasts with those generated by the no-model assumption, appreciable positive values were computed. In particular, for the larger basin with an area of 4,147 km², the mean values were 0.4, 0.4 and 0.5 for forecast lead times of two hours, four hours and six hours, respectively. Good performance of the model is also shown by a comparison of its flow predictions with those derived from a unit hydrograph based model     

CHARACTERISTICS AND CONTRIBUTING CAUSES OF AN ABNORMAL FREQUENCY OF FLOOD-PRODUCING RAINSTORMS AT CHICAGO1

ABSTRACT: Major hydrometeorological factors pertinent to defining and understanding the hydrologic characteristics of urban and other small-basin storms were investigated using data from a continuous 44-year operation of a recording raingage network in Chicago. Factors included: the frequency distribution of basin mean rainfall and its relation to storm maximum precipitation; the spatial distribution characteristics of each storm, including storm rainfall reduction factors which are widely used in hydrologic design problems; and weather-related causes related to the frequency and intensity of severe rainstorms in the Chicago area in recent years. Results have indicated that urban mean rainfall frequencies were overestimated in earlier studies in which they were derived from point/areal mean rainfall ratios obtained from much shorter records on rural networks. Reduction factors were found to vary between urban and rural storm systems due to urban-related effects. Two factors were found to be potential contributors to the characteristics of severe rainstorm occurrences at Chicago. These include urban-induced rain enhancement and an upward climatic trend in the occurrence of heavy rainfall events during the sampling period. Study results should be generally applicable to other large urban areas in the Midwest and other regions of similar precipitation climate.     

THE CHANCE A FLOOD WILL BE EXCEEDED IN A PERIOD OF YEARS1

ABSTRACT. Much has been written about the chance that a hydrologic event, such as a flood peak of a given size or greater, will occur during a given period of years. Four variables are involved, and any one of the four can be the dependent variable: (1) the probability of encountering such an event in a single year, (2) the probability of encountering one or several of these events in a period of years, (3) the least number of times of encountering the event in that period of years, and (4) the number of years in the period involved. Most of these problems are not difficult to solve, but they are tedious to calculate, not well understood, and consequently seldom used in water resources planning and development. The most popular approach is based on the binomial distribution. Graphical procedures similar to those developed by Riggs [1961] were further elaborated and are presented here with illustrative examples to facilitate their use in solving the many related problems. The confidence that one can place in these probability estimates is also explained and illustrated by tables and further examples. To help assure proper use of these methods, commonly used terms such as “recurrence interval” and “partial duration series” are discussed. No new theory is developed: at issue is a deeper understanding of the significance of design levels and their ease of computation.     

RAINGAGE NETWORK DESIGN USING NEXRAD PRECIPITATION ESTIMATES1

ABSTRACT: A general framework is proposed for using precipitation estimates from NEXRAD weather radars in raingage network design. NEXRAD precipitation products are used to represent space time rainfall fields, which can be sampled by hypothetical raingage networks. A stochastic model is used to simulate gage observations based on the areal average precipitation for radar grid cells. The stochastic model accounts for subgrid variability of precipitation within the cell and gage measurement errors. The approach is ideally suited to raingage network design in regions with strong climatic variations in rainfall where conventional methods are sometimes lacking. A case study example involving the estimation of areal average precipitation for catchments in the Catskill Mountains illustrates the approach. The case study shows how the simulation approach can be used to quantify the effects of gage density, basin size, spatial variation of precipitation, and gage measurement error, on network estimates of areal average precipitation. Although the quality of NEXRAD precipitation products imposes limitations on their use in network design, weather radars can provide valuable information for empirical assessment of rain‐gage network estimation errors. Still, the biggest challenge in quantifying estimation errors is understanding subgrid spatial variability. The results from the case study show that the spatial correlation of precipitation at subgrid scales (4 km and less) is difficult to quantify, especially for short sampling durations. Network estimation errors for hourly precipitation are extremely sensitive to the uncertainty in subgrid spatial variability, although for storm total accumulation, they are much less sensitive.