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.     

AI-based runoff simulation based on remote sensing observations: A case study of two river basins in the United States and Canada

Data-driven techniques are used extensively for hydrologic time-series prediction. We created various data-driven models (DDMs) based on machine learning: long short-term memory (LSTM), support vector regression (SVR), extreme learning machines, and an artificial neural network with backpropagation, to define the optimal approach to predicting streamflow time series in the Carson River (California, USA) and Montmorency (Canada) catchments. The moderate resolution imaging spectroradiometer (MODIS) snow-coverage dataset was applied to improve the streamflow estimate. In addition to the DDMs, the conceptual snowmelt runoff model was applied to simulate and forecast daily streamflow. The four main predictor variables, namely snow-coverage (S-C), precipitation (P), maximum temperature (T_max), and minimum temperature (T_min), and their corresponding values for each river basin, were obtained from National Climatic Data Center and National Snow and Ice Data Center to develop the model. The most relevant predictor variable was chosen using the support vector machine-recursive feature elimination feature selection approach. The results show that incorporating the MODIS snow-coverage dataset improves the models' prediction accuracies in the snowmelt-dominated basin. SVR and LSTM exhibited the best performances (root mean square error = 8.63 and 9.80) using monthly and daily snowmelt time series, respectively. In summary, machine learning is a reliable method to forecast runoff as it can be employed in global climate forecasts that require high-volume data processing.     

MODEL ACCURACY IN SNOWMELT-RUNOFF FORECASTS EXTENDING FROM 1 TO 20 DAYS1

ABSTRACT: This paper examines the performance of snowmelt-runoff models in conditions approximating real-time forecast situations. These tests are one part of an intercomparison of models recently conducted by the World Meteorological Organization (WMO). Daily runoff from the Canadian snowmelt basin Illecille. waet (1155 km², 509–3150 m a.s.l.) was forecast for 1 to 20 days ahead. The performance of models was better than in a previous WMO project, which dealt with runoff simulations from historical data, for the following reasons: (1) conditions for models were more favorable than a real-time forecast situation because measured input data and not meteorological forecast inputs were distributed to the modelers; (2) the selected test basin was relatively easy to handle and familiar from the previous WMO project; and (3) all kinds of updating were allowed so that some models even improved their accuracy towards longer forecast times. Based on this experience, a more realistic follow-up project can be imagined which would include temperature forecasts and quantitative precipitation forecasts instead of measured data.     

EFFECTS OF IMPROVED PRECIPITATION ESTIMATES ON AUTOMATED RUNOFF MAPPING: EASTERN UNITED STATES1

ABSTRACT: We evaluated maps of runoff created by means of two automated procedures. We implemented each procedure using precipitation estimates of both 5-km and 10-km resolution from PRISM (Parameter-elevation Regressions on Independent Slopes Model). Our goal was to determine if using the 5-km PRISM estimates would improve map accuracy. Visual inspection showed good general agreement among our runoff maps, as well as between our maps and one produced using a manual method. A quantitative uncertainty analysis comparing runoff interpolated from our maps with gage data that had been withheld showed slightly smaller actual and percentage interpolation errors for the 5-km PRISM-based maps. Our analyses suggest a modest region-wide improvement in runoff map accuracy with the use of PRISM-based precipitation estimates of 5-km (compared to 10-km) resolution.     

EVALUATION OF PRACTICES IN WATER SUPPLY FORECASTING1

ABSTRACT: Snowmelt runoff is a primary source of water supply in much of the Western United States. Multipurpose planning requires long-range forecasts and the accuracy of the forecasts has a significant effect on economic benefits. In an effort to increase the accuracy of snowrnelt runoff forecasts, selected practices in water supply forecasting were evaluated. These practices include 1) using multiple regression in developing forecasting models;2) using a model that was calibrated to make forecasts an April 1 for making forecasts at other times;3) using maximum snow water equivalent measurements in forecast equations; and 4) using weighted snow water equivalent measurements for making forecasts. The results of a case study indicate that forecasting accuracy is significantly affected by these practices. Goodness-of-fit statistics may not be indicative of the accuracy of forecasts when the prediction equations are used to make forecasts for dates other than that used in calibration. The use of maximum snow water equivalentmeasurements and weighted averages did not improve forecast accuracy.     

GAGE DENSITY AND LOCATION FOR ESTIMATING MEAN ANNUAL PRECIPITATION IN MOUNTAINOUS AREAS1

ABSTRACT: Data from a network of 45 shielded precipitation gages on the Reynolds Creek Experimental Watershed in Southwestern Idaho were analyzed to determine the optimum gage density for estimating mean annual precipitation. Four subsets of the 45 gage network were used to derive a curve of mean annual precipitation versus number of gages with a confidence band at the 95 percent level. When less than 20 gages were used in the estimate, the confidence interval widens rapidly. Estimates were improved by stratifying gages on the basis of plant cover class or by elevation bands. Sixty-four percent of the variation in mean annual precipitation was accounted for by elevation and cover class. The aspect and hydrologic soil classification were not statistically significant.     

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.     

基于SPOT VGT/NDVI的陕西省关中地区冬小麦遥感估产

冬小麦是关中地区的主要粮食作物之一。对冬小麦产量估算的研究可使国家有关部门及时掌握粮食生产状况,对国家制定粮食政策和经济计划,及时进行宏观调控有着重要的意义。以陕西省关中地区为研究区域,利用1999—2007年的SPOT VGT/NDVI遥感数据对冬小麦产量进行预测。首先,选取冬小麦关键生长期内0.2—0.8范围内的旬NDVI数据,利用最大值合成法合成后的月NDVI数据建立与冬小麦产量之间的关系;其次,使用相关分析方法筛选出冬小麦关键期内与产量相关性较好的月份和旬,分别建立这些月份、旬与冬小麦产量之间的回归关系模型,最后对模型进行精度验证。结果显示,NDVI与产量之间有很好的相关性,其中4月上旬的相关性最好,其相关系数R2为0.605,即利用4月份上旬的数据预测冬小麦产量最好。在预测产量与实际产量对比的基础上,得出估产的相对误差为-5.2%—-7.9%,表明估产精度良好,可运用于冬小麦的实际估产。     

FOREST GROWTH IMPACT ON PEAK SNOW PACK MEASUREMENTS AT LONG TERM SNOW COURSES1

ABSTRACT: Snow course surveys in late winter provide stream‐flow forecasters with their best information for making water supply and flood forecasts for the subsequent spring and summer runoff period in mountainous regions of western North America. Snow survey data analyses are generally based on a 30‐year “normal” period. It is well documented that forest cover changes over time will affect snow accumulation on the ground within forests. This paper seeks to determine if forest cover changes over decades at long term snow courses decrease measured peak snow water equivalent (SWE) enough to affect runoff prediction. Annual peak SWE records were analyzed at four snow courses in two different forest types having at least 25 years of snowpack data to detect any decreases in SWE due to forest growth. No statistically significant decreases in annual peak SWE over time were found at any of these four snow courses. The wide range of annual winter precipitation and correspondingly highly variable peak snowpack accumulation, as well as many other weather and site variables, masked any minor trends in the data.     

Mid‐Range Streamflow Forecasts Based on Climate Modeling – Statistical Correction and Evaluation1

Abstract: Mid‐range streamflow predictions are extremely important for managing water resources. The ability to provide mid‐range (three to six months) streamflow forecasts enables considerable improvements in water resources system operations. The skill and economic value of such forecasts are of great interest. In this research, output from a general circulation model (GCM) is used to generate hydrologic input for mid‐range streamflow forecasts. Statistical procedures including: (1) transformation, (2) correction, (3) observation of ensemble average, (4) improvement of forecast, and (5) forecast skill test are conducted to minimize the error associated with different spatial resolution between the large‐scale GCM and the finer‐scale hydrologic model and to improve forecast skills. The accuracy of a streamflow forecast generated using a hydrologic model forced with GCM output for the basin was evaluated by forecast skill scores associated with the set of streamflow forecast values in a categorical forecast. Despite the generally low forecast skill score exhibited by the climate forecasting approach, precipitation forecast skill clearly improves when a conditional forecast is performed during the East Asia summer monsoon, June through August.     

DISTRIBUTION AND STOCHASTIC GENERATION OF ANNUAL AND MONTHLY PRECIPITATION ON A MOUNTAINOUS WATERSHED IN SOUTHWEST IDAHO1

ABSTRACT: The 18-year precipitation record from the dense gage network on the Reynolds Creek Experimental Watershed located in southwest Idaho was used to determine the spatial distribution of annual and monthly precipitation on a mountainous watershed. Analyses of these data showed a linear relationship between annual amounts and elevation. This relationship was best when the gages were grouped into downwind and upwind sites. This grouping was appropriate because most of the winter storms moved over the watershed from the west and southwest, and the heaviest precipitation was on the west (downwind) side of the watershed. Gage sites along the western and southern watershed borders were most representative of the upwind gages on the east side, because they measured the precipitation from the air moving upwind onto the watershed. The maximum annual precipitation on the watershed was just leeward of the western watershed boundary. The monthly precipitation and elevation relationship was also best represented by grouping the gage sites into upwind and downwind sites. However, during the summer when there are only small amounts of pre cipitation and thunderstorms are the source of most precipitation, one equation can be used to represent the elevation relationship. This study also showed that the log-normal distribution could be used to generate the annual synthetic series, and the cube-root-normal distribution could be used to generate monthly synthetic series for all locations on the watershed.     

MULTIPLICATIVE,SEASONAL ARIMA MODELS FOR LAKE ERIE AND LAKE ONTARIO WATER LEVELS1

ABSTRACT: Federal agencies in the U.S. and Canada continuously examine methods to improve understanding and forecasting of Great Lakes water level dynamics in an effort to reduce the negative impacts of fluctuating levels incurred by interests using the lakes. The short term, seasonal and long term water level dynamics of lakes Erie and Ontario are discussed. Multiplicative, seasonal ARIMA models are developed for lakes Erie and Ontario using standardized, monthly mean level data for the period 1900 to 1986. The most appropriate model identified for each lake had the general form: (1 0 1)(0 1 1)₁₂. The data for each lake were subdivided by time periods (1900 to 1942;1 943 to 1986) and the model coefficients estimated for the subdivided data were similar, indicating general model stability for the entire period of record. The models estimated for the full data sets were used to forecast levels 1,2,3, and 6 months ahead for a period of high levels (1984 to 1986). The average absolute forecast error for Lake Erie was 0.049m, 0.076m, 0.091 m and 0.128m for the 1, 2,3, and 6 month forecasts, respectively. The average absolute forecast error for Lake Ontario was 0.058m, 0.095m, 0.120m and 0.136m for the 1,2,3, and 6 month forecasts, respectively. The ARIMA models provide additional information on water level time series structure and dynamics. The models also could be coordinated with current forecasting methods, possibly improving forecasting accuracy.     

AN EVALUATION OF WEEKLY AND MONTHLY TIME SERIES FORECASTS OF MUNICIPAL WATER USE1

ABSTRACT: A cascade model for forecasting municipal water use one week or one month ahead, conditioned on rainfall estimates, is presented and evaluated. The model comprises four components: long term trend, seasonal cycle, autocorrelation and correlation with rainfall. The increased forecast accuracy obtained by the addition of each component is evaluated. The City of Deerfield Beach, Florida, is used as the application example with the calibration period from 1976–1980 and the forecast period the drought year of 1981. Forecast accuracy is measured by the average absolute relative error (AARE, the average absolute value of the difference between actual and forecasted use, divided by the actual use). A benchmark forecast is calculated by assuming that water use for a given week or month in 1981 is the same as the average for the corresponding period from 1976 to 1980. This method produces an AARE of 14.6 percent for one step ahead forecasts of monthly data and 15.8 percent for weekly data. A cascade model using trend, seasonality and autocorrelation produces forecasts with AARE of about 12 percent for both monthly and weekly data while adding a linear relationship of water use and rainfall reduces the AARE to 8 percent in both cases if it is assumed that rainfall is known during the forecast period. Simple rainfall predictions do not increase the forecast accuracy for water use so the major utility of relating water use and rainfall lies in forecasting various possible water use sequences conditioned on sequences of historical rainfall data.     

MONTHLY RUNOFF PREDICTIONS BASED ON RAINFALL FORECASTS IN A SMALL OKLAHOMA WATERSHED1

ABSTRACT: Conditions under which monthly rainfall forecasts translate into monthly runoff predictions that could support water resources planning and management activities were investigated on a small watershed in central Oklahoma. Runoff response to rainfall forecasts was simulated using the hydrologic model SWAT. Eighteen scenarios were examined that represented combinations of wet, average, and dry antecedent rainfall conditions, with wet, normal, and dry forecasted rainfall. Results suggest that for the climatic and physiographic conditions under consideration, rainfall forecasts could offer potential application opportunities in surface water resources but only under certain conditions. Pronounced wet and dry antecedent rainfall conditions were shown to have greater impact on runoff than forecasts, particularly in the first month of a forecast period. Large forecast impacts on runoff occurred under wet antecedent conditions, when the fraction of forecasted rainfall contributing to runoff was greatest. Under dry antecedent conditions, most of the forecasted rainfall was absorbed in the soil profile, with little immediate runoff response. Persistent three‐month forecasts produced stronger impacts on runoff than one‐month forecasts due to cumulative effects in the hydrologic system. Runoff response to antecedent conditions and forecasts suggest a highly asymmetric utility function for rainfall forecasts, with greatest decision‐support potential for persistent wet forecasts under wet antecedent conditions when the forecast signal is least dampened by soil‐storage effects. Under average and dry antecedent conditions, rainfall forecasts showed little potential value for practical applications in surface water resources assessments.     

Stochastic Transfer Function Model and Neural Networks to Estimate Soil Moisture1

Abstract: A practical methodology is proposed to estimate the three‐dimensional variability of soil moisture based on a stochastic transfer function model, which is an approximation of the Richard’s equation. Satellite, radar and in situ observations are the major sources of information to develop a model that represents the dynamic water content in the soil. The soil‐moisture observations were collected from 17 stations located in Puerto Rico (PR), and a sequential quadratic programming algorithm was used to estimate the parameters of the transfer function (TF) at each station. Soil texture information, terrain elevation, vegetation index, surface temperature, and accumulated rainfall for every grid cell were input into a self‐organized artificial neural network to identify similarities on terrain spatial variability and to determine the TF that best resembles the properties of a particular grid point. Soil moisture observed at 20 cm depth, soil texture, and cumulative rainfall were also used to train a feedforward artificial neural network to estimate soil moisture at 5, 10, 50, and 100 cm depth. A validation procedure was implemented to measure the horizontal and vertical estimation accuracy of soil moisture. Validation results from spatial and temporal variation of volumetric water content (vwc) showed that the proposed algorithm estimated soil moisture with a root mean squared error (RMSE) of 2.31% vwc, and the vertical profile shows a RMSE of 2.50% vwc. The algorithm estimates soil moisture in an hourly basis at 1 km spatial resolution, and up to 1 m depth, and was successfully applied under PR climate conditions.     

WATER RESOURCE MANAGEMENT AND DROUGHT FORECASTS IN SOUTH CAROLINA1

ABSTRACT: Despite potential benefits for resource planning, community water systems managers have not used seasonal climate forecasts extensively. Obstacles to forecast use include a lack of awareness of their existence, distrust of their accuracy, perceived irrelevance to management decisions, and competition from other technological innovations. In this paper, ways in which seasonal forecasts might be extended to address more directly some concerns of South Carolina community water systems managers are explored. From May 1998 through August 2002, this group experienced drought conditions that threatened water quality and supply and required restrictions on water consumption. Methods for incorporating long lead forecasts with joint probabilities of monthly temperature and precipitation to produce drought forecasts are demonstrated. When tailored to specific places, such forecasts show the likelihood of exceeding drought thresholds that would trigger water use restrictions. The methods illustrate how long lead forecasts can be extended and customized into secondary products that address issues of greater relevance to water resource managers.     

THE DEVELOPMENT OF THE NILE HYDROMETEOROLOGICAL FORECAST SYSTEM1

ABSTRACT: The National Oceanic and Atmospheric Administration is developing a river forecast system for the Nile River in Egypt. The river forecast system operates on scientific work stations using hydrometeorological models and software to predict inflows into the high Aswan Dam and forecast flow hydrographs at selected gaging locations above the dam The Nile Forecasting System (NFS) utilizes satellite imagery from the METEOSAT satellite as the input to the forecast system. Satellite imagery is used to estimate precipitation over the Blue Nile Basin using five different techniques. Observed precipitation data and climatic statistics are used to improve precipitation estimation. Precipitation data for grid locations are input to a distributed water balance model, a hill slope routing model, and a channel routing model. A customized Geographic Information System (GIS) was developed to show political boundaries, rivers, terrain elevation, and gaging network. The GIS was used to develop hydrologic parameters for the basin and is used for multiple display features.     

REGIONAL RUNOFF SIMULATION FOR SOUTHEAST MICHIGAN1

ABSTRACT: The feasibility of simulating monthly runoff for southeast Michigan, which comprises four major river basins, was evaluated with the Streamflow Synthesis and Reservoir Regulation watershed model. The evaluation covered a 13-year period (1961–73), which encompassed a complete runoff cycle. Results indicate it is feasible to simulate monthly runoff volumes on a regional scale with a single equivalent watershed by using daily precipitation and temperature data. Simulation of regional flows appears particularly attractive for the Great Lakes basin, since the basin consists of many relatively small watersheds. This method also appears promising for development of monthly runoff forecasts by employing average monthly meteorological data distributed on a daily basis. Tests of six-month runoff forecasts show relatively small deterioration with time and offer considerable improvement over climatology.     

Hydrologic Calibration and Validation of SWAT in a Snow‐Dominated Rocky Mountain Watershed,Montana, U.S.A.1

Abstract: The Soil and Water Assessment Tool (SWAT) has been applied successfully in temperate environments but little is known about its performance in the snow‐dominated, forested, mountainous watersheds that provide much of the water supply in western North America. To address this knowledge gap, we configured SWAT to simulate the streamflow of Tenderfoot Creek (TCSWAT). Located in central Montana, TCSWAT represents a high‐elevation watershed with ～85% coniferous forest cover where more than 70% of the annual precipitation falls as snow, and runoff comes primarily from spring snowmelt. Model calibration using four years of measured daily streamflow, temperature, and precipitation data resulted in a relative error (RE) of 2% for annual water yield estimates, and mean paired deviations (Dv) of 36 and 31% and Nash‐Sutcliffe (NS) efficiencies of 0.90 and 0.86 for monthly and daily streamflow, respectively. Model validation was conducted using an additional four years of data and the performance was similar to the calibration period, with RE of 4% for annual water yields, Dv of 43% and 32%, and NS efficiencies of 0.90 and 0.76 for monthly and daily streamflow, respectively. An objective, regression‐based model invalidation procedure also indicated that the model was validated for the overall simulation period. Seasonally, SWAT performed well during the spring and early summer snowmelt runoff period, but was a poor predictor of late summer and winter base flow. The calibrated model was most sensitive to snowmelt parameters, followed in decreasing order of influence by the surface runoff lag, ground water, soil, and SCS Curve Number parameter sets. Model sensitivity to the surface runoff lag parameter reflected the influence of frozen soils on runoff processes. Results indicated that SWAT can provide reasonable predictions of annual, monthly, and daily streamflow from forested montane watersheds, but further model refinements could improve representation of snowmelt runoff processes and performance during the base flow period in this environment.     

INFLUENCE OF DIFFERENT TEMPORAL SAMPLING STRATEGIES ON ESTIMATING TOTAL PHOSPHORUS AND SUSPENDED SEDIMENT CONCENTRATION AND TRANSPORT IN SMALL STREAMS1

ABSTRACT: Various temporal sampling strategies are used to monitor water quality in small streams. To determine how various strategies influence the estimated water quality, frequently collected water quality data from eight small streams (14 to 110 km²) in Wisconsin were systematically subsampled to simulate typically used strategies. These subsets of data were then used to estimate mean, median, and maximum concentrations, and with continuous daily flows used to estimate annual loads (using the regression method) and volumetrically weighted mean concentrations. For each strategy, accuracy and precision in each summary statistic were evaluated by comparison with concentrations and loads of total phosphorus and suspended sediment estimated from all available data. The most effective sampling strategy depends on the statistic of interest and study duration. For mean and median concentrations, the most frequent fixed period sampling economically feasible is best. For maximum concentrations, any strategy with samples at or prior to peak flow is best. The best sampling strategy to estimate loads depends on the study duration. For one‐year studies, fixed period monthly sampling supplemented with storm chasing was best, even though loads were overestimated by 25 to 50 percent. For two to three‐year load studies and estimating volumetrically weighted mean concentrations, fixed period semimonthly sampling was best.