Daily Streamflow Prediction Research Articles

Assessing the hydrologic performance of the EPA’s nonpoint source water quality assessment decision support tool using North American Land Data Assimilation System (NLDAS) products

The accuracy of streamflow predictions in the EPA’s BASINS (Better Assessment Science Integrating Point and Nonpoint Sources) decision support tool is affected by the sparse meteorological data contained in BASINS. The North American Land Data Assimilation System (NLDAS) data with high spatial and temporal resolutions provide an alternative to the NOAA National Climatic Data Center (NCDC)’s station data. This study assessed the improvement of streamflow prediction of the Hydrological Simulation Program-FORTRAN (HSPF) model contained within BASINS using the NLDAS 1/8 degree hourly precipitation and evapotranspiration estimates in seven watersheds of the Chesapeake Bay region. Our results demonstrated consistent improvements of daily streamflow predictions in five of the seven watersheds when NLDAS precipitation and evapotranspiration data was incorporated into BASINS. The improvement of using NLDAS data is significant when the watershed’s meteorological station is either far away or not in a similar climatic region. When the station is nearby, using NLDAS data produces similar results. The correlation coefficients of the analyses using NLDAS data were greater than 0.8, the Nash–Sutcliffe (NS) model fit efficiency greater than 0.6, and the error in the water balance was less than 5%. Our analyses also showed that the streamflow improvements were mainly contributed by NLDAS precipitation data and that the improvement from using NLDAS evapotranspiration data was not significant; partially due to the constraints of current BASINS-HSPF settings. However, NLDAS evapotranspiration data did improve the baseflow prediction. This study demonstrates NLDAS data has the potential to improve stream flow predictions, thus aid the water quality assessment in the EPA nonpoint water quality assessment decision tool.

GOES Solar Radiation for Evapotranspiration Estimation and Streamflow Prediction

This study explores the value of incoming solar radiation estimates derived from geostationary operational environmental satellites (GOES) as an alternative solar radiation data source to historical solar radiation estimates for use in rainfall – runoff simulations. At 26 sites in the United States, the GOES-based solar radiation (\iR²=0.79) and ET\dp (R²=0.94) estimates agreed well with the ground-based estimates. There was a slight positive bias for the GOES-based solar radiation (5.4%) and potential evapotranspiration (8.7%) estimates compared to the ground-based estimates, with the strong and significant biases noted during the warm seasons. However, individual stations biases ranged from – 12 to 27%. Daily streamflow predictions made using National Weather Service River Forecast System’s Sacramento Soil Moisture Accounting rainfall.runoff model for four NOAA Distributed Model Intercomparison Project watersheds show that although notable peak storm flow differences occurred occasionally during extended drying periods, on average, the GOES-based potential evapotranspiration estimates slightly outperformed the simulations using ground-based data on an annual basis and were relatively insensitive to biases. However, on seasonal time scales, differences in soil water storage are evident and can result in numerous moderate differences in streamflow predictions.

Daily streamflow prediction with uncertainty in ephemeral catchments using the GLUE methodology

The Generalised Likelihood Uncertainty Estimation (GLUE) approach is presented here as a tool for estimating the predictive uncertainty of a rainfall–runoff model. The GLUE methodology allows to recognise the possible equifinality of different parameter sets and assesses the likelihood of a parameters set being acceptable simulator when model predictions are compared to observed field data. The results of the GLUE methodology depend greatly on the choice of the likelihood measure and on the choice of the threshold which determines if a parameters set is behavioural or not. Moreover the sampling size has a strong influence on the uncertainty assessment of the response of a rainfall–runoff model. This is one of the most controversial and criticized aspect of the GLUE methodology, because it seems that this procedure does not learn from observations.Following these premises, this paper investigated first on the effect of different likelihood measures on the uncertainty analysis in the rainfall–runoff modelling of a mediterranean catchment. Performance of individual parameter sets has been assessed using three likelihood measures with a shaping factor. The acceptability threshold influence on the uncertainty analysis has been also evaluated. Finally it can be demonstrated how, using the GLUE, the predictive uncertainty grows with the streamflow variance while remains almost the same with the sampling size. In order to overcome these inconsistencies, a new simple likelihood measure, which explicitly takes into account the sample variance and extension, is here proposed.

Prediction of daily flow duration curves and streamflow for ungauged catchments using regional flow duration curves

A method is presented to predict flow duration curves (FDCs) and streamflow for ungauged catchments in the Mid-Atlantic Region, USA. Twenty-nine catchments were selected from the Appalachian Plateau, Ridge and Valley, and Piedmont physiographic provinces to develop and test the proposed method. Using a step-wise multiple regression analysis, the dominant landscape and climate descriptors were identified and regional FDC models were developed for each province. Predictive performance of the proposed method was estimated using data from three evaluation sites that were not included in parameter estimation for the regional FDC models. The results of this study show that climate and geomorphological descriptors strongly influence the hydrology of the Appalachian Plateau and some Ridge and Valley catchments, whereas soil and geomorphological descriptors strongly influence the hydrology of the Piedmont and some Ridge and Valley catchments. Streamflow values calculated by the drainage area ratio method and those reconstructed by the FDC method were compared with the observed streamflow values. The results indicate that the FDC-based method shows great promise for predicting streamflow in ungauged basins.

Towards a generic neural network model for the prediction of daily streamflow in ungauged boreal plain watersheds

Most of the currently available streamflow neural network models are either recurrent neural networks or feed-forward multi-layer perceptron (FF-MLP) requiring past flow values for lead time prediction. These models cannot be used for modelling ungauged watersheds because past flow values are not available in such cases. This study proposes a FF-MLP algorithm that relies only on low-cost, readily available meteorological data and careful time series manipulation prior to model building, and thus, is suitable for modelling streamflow in ungauged watersheds. The proposed approach was tested on four watersheds (5 to 130 km2) in the Canadian Boreal forest and was found to provide an efficient modelling alternative for daily streamflow predictions. To assess the possibility of successful model transferability from a gauged watershed to a hydrologically similar ungauged watershed, a new remotely sensed hydrologic similarity measure — SWMIR_SI — was proposed and was found to provide a successful indicator of basin similarity. The square of Pearson’s correlation coefficient, r2, was evaluated to exceed 0.71 when SWMIR_SI was regressed to models’ “goodness-of-fit” statistics reflecting the usefulness of the approach.

Daily Stream Flow Prediction Capability of Artificial Neural Networks as influenced by Minimum Air Temperature Data

The minimum air temperature influence on training of artificial neural networks (ANNs) for daily stream flow prediction is investigated. For improved computed performances, efficiencies of the conjugate gradient (CG) and Levenberg–Marquardt (L–M) training algorithms are compared. The minimum air temperature influence is studied for a watershed in southern Iran, representing a continuous flow regime with peak flows occurring as a result of individual rainfall events. Peak flow occurrence generally coincides with freezing or near-freezing periods. Based on the results, the L–M algorithm is more efficient than the CG algorithm, so it is used to train four ANNs models for stream flow prediction at time step t+1 from time step t input. Model 1 uses enabled stream flow data as input dimension, Model 2 uses enabled stream flow and single raingauge, Model 3 uses enabled raingauge network and Model 4 uses enabled raingauge network data and minimum air temperature. Validation stage root mean square error (RMSE), root mean absolute error (RMAE) and efficiency (EF) measures are: 25·456, 0·220, 0·580 (Model 1); 17·401, 0·139, 0·813 (Model 2); 12·220, 0·112, 0·916 (Model 3) and 10·383, 0·069, 0·939 (Model 4), respectively. The influence of minimum air temperature on improved Model 4 performance is discussed.

On the utility of operational precipitation forecasts to served as input for streamflow forecasting

This article studies the utility of quantitative forecast precipitation for the prediction of daily streamflow. Application is made over the Rhone basin, which was included in the Gewex–Rhone program. The precipitation forecasts of the two numerical weather prediction models operationally used in France, ARPEGE and ALADIN, are tested. The riverflow forecast is made using the precipitation forecast as input to the one-way atmosphere–hydrology coupled model SAFRAN–ISBA–MODCOU (SIM). Such a forecast is very sensitive to the initialisation of soil moisture and snow-pack. Therefore, two kinds of streamflow forecast were made: first, a plain forecast, for which the initial conditions are taken from the guess, and second, a re-initialised forecast, for which the initial conditions are set according to a reference run. This reference run is obtained using 1200 daily observed precipitation, interpolated in time and space by SAFRAN. First, the quality of the precipitation forecasts is checked over the Rhone basin for the period August 1997–July 1998 using the SAFRAN analysis as a reference. Then, the SIM system used to forecast riverflow is briefly presented. The predicted riverflows are compared both to the observations at 22 streamgage locations and to the reference run. The results show that the annual average of the average discharge errors at the 22 streamgages can reach 20% in the forecast without re-initialisation, but that this error is reduced significantly when the model is properly initialised. This is due to the fact the re-initialisation of the soil moisture and snow-pack according to the reference run minimizes influences of the previous precipitation forecast error. It is shown that the use of precipitation forecast as input of the SIM system can be of interest to forecast the progress of the long-duration floods of the main stations of the Rhone basin.

On the utility of operational precipitation forecasts to served as input for streamflow forecasting

This article studies the utility of quantitative forecast precipitation for the prediction of daily streamflow. Application is made over the Rhone basin, which was included in the Gewex–Rhone program. The precipitation forecasts of the two numerical weather prediction models operationally used in France, ARPEGE and ALADIN, are tested. The riverflow forecast is made using the precipitation forecast as input to the one-way atmosphere–hydrology coupled model SAFRAN–ISBA–MODCOU (SIM). Such a forecast is very sensitive to the initialisation of soil moisture and snow-pack. Therefore, two kinds of streamflow forecast were made: first, a plain forecast, for which the initial conditions are taken from the guess, and second, a re-initialised forecast, for which the initial conditions are set according to a reference run. This reference run is obtained using 1200 daily observed precipitation, interpolated in time and space by SAFRAN. First, the quality of the precipitation forecasts is checked over the Rhone basin for the period August 1997–July 1998 using the SAFRAN analysis as a reference. Then, the SIM system used to forecast riverflow is briefly presented. The predicted riverflows are compared both to the observations at 22 streamgage locations and to the reference run. The results show that the annual average of the average discharge errors at the 22 streamgages can reach 20% in the forecast without re-initialisation, but that this error is reduced significantly when the model is properly initialised. This is due to the fact the re-initialisation of the soil moisture and snow-pack according to the reference run minimizes influences of the previous precipitation forecast error. It is shown that the use of precipitation forecast as input of the SIM system can be of interest to forecast the progress of the long-duration floods of the main stations of the Rhone basin.

Bayesian estimation of parameters in a regional hydrological model

Abstract. This study evaluates the applicability of the distributed, process-oriented Ecomag model for prediction of daily streamflow in ungauged basins. The Ecomag model is applied as a regional model to nine catchments in the NOPEX area, using Bayesian statistics to estimate the posterior distribution of the model parameters conditioned on the observed streamflow. The distribution is calculated by Markov Chain Monte Carlo (MCMC) analysis. The Bayesian method requires formulation of a likelihood function for the parameters and three alternative formulations are used. The first is a subjectively chosen objective function that describes the goodness of fit between the simulated and observed streamflow, as defined in the GLUE framework. The second and third formulations are more statistically correct likelihood models that describe the simulation errors. The full statistical likelihood model describes the simulation errors as an AR(1) process, whereas the simple model excludes the auto-regressive part. The statistical parameters depend on the catchments and the hydrological processes and the statistical and the hydrological parameters are estimated simultaneously. The results show that the simple likelihood model gives the most robust parameter estimates. The simulation error may be explained to a large extent by the catchment characteristics and climatic conditions, so it is possible to transfer knowledge about them to ungauged catchments. The statistical models for the simulation errors indicate that structural errors in the model are more important than parameter uncertainties. Keywords: regional hydrological model, model uncertainty, Bayesian analysis, Markov Chain Monte Carlo analysis