River flow time series prediction with a range-dependent neural network

Abstract

Artificial neural networks provide a promising alternative to hydrological time series modelling. However, there are still many fundamental problems requiring further analyses, such as structure identification, parameter estimation, generalization, performance improvement, etc. Based on a proposed clustering algorithm for the training pairs, a new neural network, namely the range-dependent neural network (RDNN) has been developed for better accuracy in hydrological time series prediction. The applicability and potentials of the RDNN in daily streamflow and annual reservoir inflow prediction are examined using data from two watersheds in China. Empirical comparisons of the predictive accuracy, in terms of the model efficiency R² and absolute relative errors (ARE), between the RDNN, back-propagation (BP) networks and the threshold auto-regressive (TAR) model are made. The case studies demonstrated that the RDNN network performed significantly better than the BP network, especially for reproducing low-flow events.     

A wavelet-neural network hybrid modelling approach for estimating and predicting river monthly flows

Abstract

A wavelet-neural network (WNN) hybrid modelling approach for monthly river flow estimation and prediction is developed. This approach integrates discrete wavelet multi-resolution decomposition and a back-propagation (BP) feed-forward multilayer perceptron (FFML) artificial neural network (ANN). The Levenberg-Marquardt (LM) algorithm and the Bayesian regularization (BR) algorithm were employed to perform the network modelling. Monthly flow data from three gauges in the Weihe River in China were used for network training and testing for 48-month-ahead prediction. The comparison of results of the WNN hybrid model with those of the single ANN model show that the former is able to significantly increase the prediction accuracy.

Editor D. Koutsoyiannis; Associate editor H. Aksoy

Citation Wei, S., Yang, H., Song, J.X., Abbaspour, K., and Xu, Z.X., 2013. A wavelet-neural network hybrid modelling approach for estimating and predicting river monthly flows. Hydrological Sciences Journal, 58 (2), 374–389.     

Utility of high-resolution climate model simulations for water resources prediction over the Korean Peninsula: a sensitivity study / Utilité de simulations de modèle climatique à haute résolution pour la prévision des ressources en eau dans la Péninsule Coréenne: étude de sensibilité

Abstract

Abstract The utility of simulations of Global Climate Models (GCMs) for regional water resources prediction and management on the Korean Peninsula was assessed by a probabilistic measure. Global Climate Model simulations of an indicator variable (e.g. surface precipitation or temperature) were used for discriminating high vs low regional observations of a target variable (e.g. watershed precipitation or reservoir inflow). The formulation uses the significance probability of the Kolmogorov-Smirnov test for detecting differences between two distributions. High resolution Atmospheric Model Intercomparison Project-II (AMIP-II) type GCM simulations performed by the European Centre for Medium-Range Weather Forecasts (ECMWF) and AMIP-I type GCM simulations performed by the Korean Meteorological Research Institute (METRI) were used to obtain information for the indicator variables. Observed mean areal precipitation and temperature, and watershed-outlet discharge values for seven major river basins in Korea were used as the target variables. The results suggest that the use of the climate model nodal output from both climate models in the vicinity of the target basin with monthly resolution will be beneficial for water resources planning and management analysis that depends on watershed mean areal precipitation and temperature, and outlet discharge.     

Monthly suspended sediment load prediction using artificial intelligence: testing of a new random subspace method

ABSTRACT

The predictive capability of a new artificial intelligence method, random subspace (RS), for the prediction of suspended sediment load in rivers was compared with commonly used methods: random forest (RF) and two support vector machine (SVM) models using a radial basis function kernel (SVM-RBF) and a normalized polynomial kernel (SVM-NPK). Using river discharge, rainfall and river stage data from the Haraz River, Iran, the results revealed: (a) the RS model provided a superior predictive accuracy (NSE = 0.83) to SVM-RBF (NSE = 0.80), SVM-NPK (NSE = 0.78) and RF (NSE = 0.68), corresponding to very good, good, satisfactory and unsatisfactory accuracies in load prediction; (b) the RBF kernel outperformed the NPK kernel; (c) the predictive capability was most sensitive to gamma and epsilon in SVM models, maximum depth of a tree and the number of features in RF models, classifier type, number of trees and subspace size in RS models; and (d) suspended sediment loads were most closely correlated with river discharge (PCC = 0.76). Overall, the results show that RS models have great potential in data poor watersheds, such as that studied here, to produce strong predictions of suspended load based on monthly records of river discharge, rainfall depth and river stage alone.     

Genetic algorithms for the classification and prediction of precipitation occurrence

Abstract

Using an approach similar to the biological processes of natural selection and evolution, the genetic algorithm (GA) is a nonconventional optimum search technique. Genetic algorithms have the ability to search large and complex decision spaces and handle nonconvexities. In this paper, the GA is applied for solving the optimum classification of rainy and non-rainy day occurrences based on vertical velocity, dewpoint depression, temperature and humidity data. The problem involves finding optimum classification based on known data, training the future prediction system and then making reliable predictions for rainfall occurrences which have significance in agricultural, transportation, water resources and tourism activities. Various statistical approaches require restrictive assumptions such as stationarity, homogeneity and normal probability distribution of the hydrological variables concerned. The GAs do not require any of these assumptions in their applications. The GA approach for the occurrence classifications and predictions is presented in steps and then the application of the methodology is shown for precipitation occurrence (non-occurrence) data. It has been shown that GAs give better results than classical approaches such as discriminant analysis. The application of the methodology is presented independently for the precipitation event occurrences and forecasting at the Lake Van station in eastern Turkey. Finally, the amounts of precipitation are predicted with a model similar to a third order Markov model whose parameters are estimated by the GA technique.     

Assessment of the effectiveness of a multi-site stochastic weather generator on hydrological modelling in the Red Deer River watershed,Canada

ABSTRACT

To improve the convergence of multiple-site weather generators (SWGs) based on the brute force algorithm (MBFA), a genetic algorithm (GA) is proposed to search the overall optimal correlation matrix. Precipitation series from weather generators are used as input to the hydrological model, the soil and water assessment tool (SWAT), to generate runoff over the Red Deer watershed, Canada for further runoff analysis. The results indicate that the SWAT model using SWG-generated data accurately represents the mean monthly streamflow for most of the months. The multi-site generators were capable of better representing the monthly streamflow variability, which was notably underestimated by the single-site version. In terms of extreme flows, the proposed method reproduced the observed extreme flow with smaller bias than MBFA, while the single-site generator significantly underestimated the annual maximum flows due to its poor capability in addressing partial precipitation correlations.     

An evaluation of numerical weather prediction based rainfall forecasts

ABSTRACT

Assessment of forecast precipitation is required before it can be used as input to hydrological models. Using radar observations in southeastern Australia, forecast rainfall from the Australian Community Climate Earth-System Simulator (ACCESS) was evaluated for 2010 and 2011. Radar rain intensities were first calibrated to gauge rainfall data from four research rainfall stations at hourly time steps. It is shown that the Australian ACCESS model (ACCESS-A) overestimated rainfall in low precipitation areas and underestimated elevated accumulations in high rainfall areas. The forecast errors were found to be dependent on the rainfall magnitude. Since the cumulative rainfall observations varied across the area and through the year, the relative error (RE) in the forecasts varied considerably with space and time, such that there was no consistent bias across the study area. Moreover, further analysis indicated that both location and magnitude errors were the main sources of forecast uncertainties on hourly accumulations, while magnitude was the dominant error on the daily time scale. Consequently, the precipitation output from ACCESS-A may not be useful for direct application in hydrological modelling, and pre-processing approaches such as bias correction or exceedance probability correction will likely be necessary for application of the numerical weather prediction (NWP) outputs.

EDITOR M.C. Acreman ASSOCIATE EDITOR A. Viglione     

Estimation of monthly evaporative loss using relevance vector machine,extreme learning machine and multivariate adaptive regression spline models

The forecasting of evaporative loss (E) is vital for water resource management and understanding of hydrological process for farming practices, ecosystem management and hydrologic engineering. This study has developed three machine learning algorithms, namely the relevance vector machine (RVM), extreme learning machine (ELM) and multivariate adaptive regression spline (MARS) for the prediction of E using five predictor variables, incident solar radiation (S), maximum temperature (T _max), minimum temperature (T _min), atmospheric vapor pressure (VP) and precipitation (P). The RVM model is based on the Bayesian formulation of a linear model with appropriate prior that results in sparse representations. The ELM model is computationally efficient algorithm based on Single Layer Feedforward Neural Network with hidden neurons that randomly choose input weights and the MARS model is built on flexible regression algorithm that generally divides solution space into intervals of predictor variables and fits splines (basis functions) to each interval. By utilizing random sampling process, the predictor data were partitioned into the training phase (70 % of data) and testing phase (remainder 30 %). The equations for the prediction of monthly E were formulated. The RVM model was devised using the radial basis function, while the ELM model comprised of 5 inputs and 10 hidden neurons and used the radial basis activation function, and the MARS model utilized 15 basis functions. The decomposition of variance among the predictor dataset of the MARS model yielded the largest magnitude of the Generalized Cross Validation statistic (≈0.03) when the T _max was used as an input, followed by the relatively lower value (≈0.028, 0.019) for inputs defined by the S and VP. This confirmed that the prediction of E utilized the largest contributions of the predictive features from the T _max, verified emphatically by sensitivity analysis test. The model performance statistics yielded correlation coefficients of 0.979 (RVM), 0.977 (ELM) and 0.974 (MARS), Root-Mean-Square-Errors of 9.306, 9.714 and 10.457 and Mean-Absolute-Error of 0.034, 0.035 and 0.038. Despite the small differences in the overall prediction skill, the RVM model appeared to be more accurate in prediction of E. It is therefore advocated that the RVM model can be employed as a promising machine learning tool for the prediction of evaporative loss.     

动力和统计预测信息融合预测方法及对我国夏季降水预测的检验

短期气候预测中如何将气候模式和统计方法的预测结果科学、客观的集成起来,一直是非常重要的问题.本文针对动力模式和统计方法预测结果相结合的问题,引入资料同化中信息融合的思想,采用最优内插同化方法,实现了动力模式和统计季节降水预测结果的融合.检验表明,对1982-2015年我国夏季降水百分率的回报,融合预测结果与观测的平均空间相关系数可达0.44,分别较统计预测和CFSv2模式统计降尺度订正的技巧提高了0.1左右,而均方根误差较两者可以降低5%~20%.可见,该方法可以进一步提升对我国夏季降水的预测技巧,具有显著的业务应用价值.     

Analysis of trends of annual and seasonal precipitation from 1956 to 2000 in Guangdong Province,China

Abstract

The trends of annual, seasonal and monthly precipitation in southern China (Guangdong Province) for the period 1956–2000 are investigated, based on the data from 186 high-quality gauging stations. Statistical tests, including Mann-Kendall rank test and wavelet analysis, are employed to determine whether the precipitation series exhibit any regular trend and periodicity. The results indicate that the annual precipitation has a slightly decreasing trend in central Guangdong and slight increasing trends in the eastern and western areas of the province. However, all the annual trends are not statistically significant at the 95% confidence level. The average precipitation increases in the dry season in central Guangdong, but decreases in the wet season, meaning that the precipitation becomes more evenly distributed within the year. Furthermore, the analysis of monthly precipitation suggests that the distribution of intra-annual precipitation changes over time. The results of wavelet analysis show prominent precipitation with periods ranging from 10 to 12 years in every sub-region in Guangdong Province. Comparing with the sunspot cycle (11-year), the annual precipitation in every sub-region in Guangdong province correlates with Sunspot Number with a 3-year lag. The findings in this paper will be useful for water resources management.

Editor Z.W. Kundzewicz; Associate editor Sheng Yue

Citation Dedi Liu, Shenglian Guo, Xiaohong Chen and Quanxi Shao, 2012. Analysis of trends of annual and seasonal precipitation from 1956 to 2000 in Guangdong Province, China. Hydrological Sciences Journal, 57 (2), 358–369.     

General Secretary's Notes

Abstract

The following paper presents two different snowmelt submodels requiring a minimum input and suitable for estimating the monthly snowmelt-runoff in seasonally snowcovered watersheds. The first submodel discussed is based on the degree-day method, relating the variation of the degree-day factor in the melt season to temperature range as an index for the available radiation energy. The second submodel estimates the melt rate according to each of the components of the total energy contributing to the melt process of the snowpack. The submodels were used as a subroutine of a monthly water yield model in conjunction with a nonlinear optimization algorithm for identification of the parameters. The inputs used are the monthly precipitation, the average of daily air temperatures over a given month, and the extra-terrestrial radiation.     

Delineation of precipitation regions in two Canadian study areas: the role of the temporal resolution of the precipitation data

The identification of homogeneous precipitation regions has value in many water resources engineering applications (infrastructure planning, design, operations; climate forecasting, modelling). The objective of this paper is to assess the sensitivity of precipitation regions to the temporal resolution (monthly, seasonal, annual and the annual maximum series) of the data. The presented method uses the fuzzy c-means clustering algorithm to partition climate sites into statistically homogeneous precipitation regions. The regions are validated using an approach based on L-moment statistics. The method is conducted in two climatically different study areas in western and eastern Canada. There does not appear to be a relationship between the spatial distributions of the regions formed using different temporal resolutions of the precipitation data. It is recommended to delineate precipitation regions that are specific to the task at hand, and to select a temporal resolution that is consistent with the final application of the regional precipitation dataset.

EDITOR A. Castellarin; ASSOCIATE EDITOR T. Kjeldsen     

Evaluation of satellite-based rainfall products for hydrological modelling in Morocco

ABSTRACT

Several satellite-based precipitation estimates are becoming available at a global scale, providing new possibilities for water resources modelling, particularly in data-sparse regions and developing countries. This work provides a first validation of five different satellite-based precipitation products (TRMM-3B42 v6 and v7, RFE 2.0, PERSIANN-CDR, CMORPH1.0 version 0.x) in the 1785 km² Makhazine catchment (Morocco). Precipitation products are first compared against ground observations. Ten raingauges and four different interpolation methods (inverse distance, nearest neighbour, ordinary kriging and residual kriging with altitude) were used to compute a set of interpolated precipitation reference fields. Second, a parsimonious conceptual hydrological model is considered, with a simulation approach based on the random generation of model parameters drawn from existing parameter set libraries, to compare the different precipitation inputs. The results indicate that (1) all four interpolation methods, except the nearest neighbour approach, give similar and valid precipitation estimates at the catchment scale; (2) among the different satellite-based precipitation estimates verified, the TRMM-3B42 v7 product is the closest to observed precipitation, and (3) despite poor performance at the daily time step when used in the hydrological model, TRMM-3B42 v7 estimates are found adequate to reproduce monthly dynamics of discharge in the catchment. The results provide valuable perspectives for water resources modelling of data-scarce catchments with satellite-based rainfall data in this region.

Editor M.C. Acreman; Associate editor N. Verhoest     

A two-stage time series model for monthly hydrological projections under climate change in the Lim River basin (southeast Europe)

ABSTRACT

Climate change projections of precipitation and temperature suggest that Serbia could be one of the most affected regions in southeastern Europe. To prepare adaptation measures, the impact of climate changes on water resources needs to be assessed. Pilot research is carried out for the Lim River basin, in southeastern Europe, to predict monthly flows under different climate scenarios. For estimation of future water availability, an alternative approach of developing a deterministic-stochastic time series model is chosen. The proposed two-stage time series model consists of several components: trend, long-term periodicity, seasonality and the stochastic component. The latter is based on a transfer function model with two input variables, precipitation and temperature, as climatic drivers. The Nash-Sutcliffe model efficiency for the observed period 1950–2012 is 0.829. The model is applied for the long-term hydrological prediction under the representative concentration pathway (RCP) emissions scenarios for the future time frame 2013–2070.     

SPI drought class prediction using log-linear models applied to wet and dry seasons

A log-linear modelling for 3-dimensional contingency tables was used with categorical time series of SPI drought class transitions for prediction of monthly drought severity. Standardized Precipitation Index (SPI) time series in 12- and 6-month time scales were computed for 10 precipitation time series relative to GPCC datasets with 2.5° spatial resolution located over Portugal and with 112 years length (1902–2014). The aim was modelling two-month step class transitions for the wet and dry seasons of the year and then obtain probability ratios – Odds – as well as their respective confidence intervals to estimate how probable a transition is compared to another. The prediction results produced by the modelling applied to wet and dry season separately, for the 6- and the 12-month SPI time scale, were compared with the results produced by the same modelling without the split, using skill scores computed for the entire time series length. Results point to good prediction performances ranging from 70 to 80% in the percentage of corrects (PC) and 50–70% in the Heidke skill score (HSS), with the highest scores obtained when the modelling is applied to the SPI12. The adding up of the wet and dry seasons introduced in the modelling brought improvements in the predictions, of about 0.9–4% in the PC and 1.3–6.8% in the HSS, being the highest improvements obtained in the SPI6 application.     

Improvement of Local Categorical Precipitation Forecasts from an NWP Model by Various Statistical Postprocessing Methods

Several statistical postprocessing methods are applied to results from a numerical weather prediction (NWP) model to test the potential for increasing the accuracy of its local precipitation forecasts. Categorical (Yes/No) forecasts for 12hr precipitation sums equalling or exceeding 0.1, 2.0 and 5.0 mm are selected for improvement. The two 12hr periods 0600-1800 UTC and 1800-0600 UTC are treated separately based on NWP model initial times 0000 UTC and 1200 UTC, respectively. Input data are taken from three successive summer seasons, April-September, 1994-96. The forecasts are prepared and verified for five synoptic stations, four located in the western Czech Republic, and one in Germany near the Czech-German border. Two approaches to statistical postprocessing are tested. The first uses Model Output Statistics (MOS) and the second modifies the MOS approach by applying a successive learning technique (SLT). For each approach several statistical models for the relationship between NWP model predictors and predictand were studied. An independent data set is used for forecast verification with the skill measured by a True Skill Score. The results of the statistical postprocessing are compared with the direct model precipitation forecasts from gridpoints nearest the stations, and they show that both postprocessing approaches provide substantially better forecasts than the direct NWP model output. The relative improvement increases with increasing precipitation amount and there is no significant difference in performance between the two 12hr periods. The skill of the SLT does not depend significantly on the size of the initial learning sample, but its results are nevertheless comparable with the results obtained from the MOS approach, which requires larger developmental samples.     

Analysis of hydrological seasonality across northern catchments using monthly precipitation–runoff polygon metrics

Abstract

Seasonality is an important hydrological signature for catchment comparison. Here, the relevance of monthly precipitation–runoff polygons (defined as scatter points of 12 monthly average precipitation–runoff value pairs connected in the chronological monthly sequence) for characterizing seasonality patterns was investigated to describe the hydrological behaviour of 10 catchments spanning a climatic gradient across the northern temperate region. Specifically, the research objectives were to: (a) discuss the extent to which monthly precipitation–runoff polygons can be used to infer active hydrological processes in contrasting catchments; (b) test the ability of quantitative metrics describing the shape, orientation and surface area of monthly precipitation–runoff polygons to discriminate between different seasonality patterns; and (c) examine the value of precipitation–runoff polygons as a basis for catchment grouping and comparison. This study showed that some polygon metrics were as effective as monthly average runoff coefficients for illustrating differences between the 10 catchments. The use of precipitation–runoff polygons was especially helpful to look at the dynamics prevailing in specific months and better assess the coupling between precipitation and runoff and their relative degree of seasonality. This polygon methodology, linked with a range of quantitative metrics, could therefore provide a new simple tool for understanding and comparing seasonality among catchments.

Editor Z.W. Kundzewicz; Associate editor K. Heal

Citation Ali, G., Tetzlaff, D., Kruitbos, L., Soulsby, C., Carey, S., McDonnell, J., Buttle, J., Laudon, H., Seibert, J., McGuire, K., and Shanley, J., 2013. Analysis of hydrological seasonality across northern catchments using monthly precipitation–runoff polygon metrics. Hydrological Sciences Journal, 59 (1), 56–72.     

Hydrological model comparison and assessment: criteria from catchment scales and temporal resolution

ABSTRACT

This study examines the performance of three hydrological models, namely the artificial neural network (ANN) model, the Hydrologiska Byråns Vattenbalansavdelning-D (HBV-D) model, and the Soil and Water Integrated Model (SWIM) over the upper reaches of the Huai River basin. The assessment is done by using databases of different temporal resolution and by further examining the applicability of SWIM for different catchment sizes. The results show that at monthly scale the performance of the ANN model is better than that of HBV-D and SWIM. The ANN model can be applied at any temporal scale as it establishes an artificial precipitation–runoff relationship for various time scales by only using monthly precipitation, temperature and runoff data. However, at daily scale the performance of both HBV-D and SWIM are similar or even better than the ANN model. In addition, the performance of SWIM at a small catchment size (less than 10 000 km²) is much better than at a larger catchment size. In view of climate change modelling, HBV-D and SWIM might be integrated in a dynamical atmosphere-water-cycle modelling rather than the ANN model due to their use of observed physical links instead of artificial relations within a black box.

Editor D. Koutsoyiannis; Associate editor D. Hughes     

Precipitation,temperature and runoff analysis from 1950 to 2002 in the Yangtze basin,China / Analyse des précipitations,températures et débits de 1950 à 2002 dans le bassin du Yangtze,en Chine

Abstract

Abstract Monthly precipitation and temperature trends of 51 stations in the Yangtze basin from 1950–2002 were analysed and interpolated. The Mann-Kendall trend test was applied to examine the monthly precipitation and temperature data. Significant positive and negative trends at the 90, 95 and 99% significance levels were detected. The monthly mean temperature, precipitation, summer precipitation and monthly mean runoff at Yichang, Hankou and Datong stations were analysed. The results indicate that spatial distribution of precipitation and temperature trends is different. The middle and lower Yangtze basin is dominated by upward precipitation trend but by somewhat downward temperature trend; while downward precipitation trend and upward temperature trend occur in the upper Yangtze basin. This is because increasing precipitation leads to increasing cloud coverage and, hence, results in decreasing ground surface temperature. Average monthly precipitation and temperature analysis for the upper, middle and lower Yangtze basin, respectively, further corroborate this viewpoint. Analysis of precipitation trend for these three regions and of runoff trends for the Yichang, Hankou and Datong stations indicated that runoff trends respond well to the precipitation trends. Historical flood trend analysis also shows that floods in the middle and lower Yangtze basin are in upward trend. The above findings indicate that the middle and lower Yangtze basin is likely to face more serious flood disasters. The research results help in further understanding the influence of climatic changes on floods in the Yangtze basin, providing scientific background for the flood control activities in large catchments in Asia.