Impact of number of realizations on the suitability of simulated weather data for hydrologic and environmental applications

Stochastic weather generators are widely used in hydrological, environmental, and agricultural applications to simulate weather time series. However, such stochastic models produce random outputs hence the question on how representative the generated data are if obtained from only one simulation run (realization) as is common practice. In this study, the impact of different numbers of realizations (1, 25, 50, and 100) on the suitability of generated weather data was investigated. Specifically, 50 years of daily precipitation, and maximum and minimum temperatures were generated for three weather stations in the Western Lake Erie Basin (WLEB), using three widely used weather generators, CLIGEN, LARSWG and WeaGETS. Generated results were compared with 50 years of observed data. For all three generators, the analyses showed that one realization of data for 50 years of daily precipitation, and maximum and minimum temperatures may not be representative enough to capture essential statistical characteristics of the climate. Results from the three generators captured the essential statistical characteristics of the climate when the number of realizations was increased from 1 to 25, 50 or 100. Performance did not improve substantially when realizations were increased above 25. Results suggest the need for more than a single realization when generating weather data and subsequently utilizing in other models, to obtain suitable representations of climate.     

Simulation of probability distributions commonly used in hydrological frequency analysis

Random variable simulation has been applied to many applications in hydrological modelling, flood risk analysis, environmental impact assessment, etc. However, computer codes for simulation of distributions commonly used in hydrological frequency analysis are not available in most software libraries. This paper presents a frequency‐factor‐based method for random number generation of five distributions (normal, log–normal, extreme‐value type I, Pearson type III and log‐Pearson type III) commonly used in hydrological frequency analysis. The proposed method is shown to produce random numbers of desired distributions through three means of validation: (1) graphical comparison of cumulative distribution functions (CDFs) and empirical CDFs derived from generated data; (2) properties of estimated parameters; (3) type I error of goodness‐of‐fit test. An advantage of the method is that it does not require CDF inversion, and frequency factors of the five commonly used distributions involves only the standard normal deviate. Copyright © 2006 John Wiley & Sons, Ltd.     

Simulation of historical temperatures using a multi‐site,multivariate block resampling algorithm with perturbation

Stochastic weather generators have evolved as tools for creating long time series of synthetic meteorological data at a site for risk assessments in hydrologic and agricultural applications. Recently, their use has been extended as downscaling tools for climate change impact assessments. Non‐parametric weather generators, which typically use a K‐nearest neighbour (K‐NN) resampling approach, require no statistical assumptions about probability distributions of variables and can be easily applied for multi‐site use. Two characteristics of traditional K‐NN models result from resampling daily values: (1) temporal correlation structure of daily temperatures may be lost, and (2) no values less than or exceeding historical observations can be simulated. Temporal correlation in simulated temperature data is important for hydrologic applications. Temperature is a major driver of many processes within the hydrologic cycle (for example, evaporation, snow melt, etc.) that may affect flood levels. As such, a new methodology for simulation of climate data using the K‐NN approach is presented (named KnnCAD Version 4). A block resampling scheme is introduced along with perturbation of the reshuffled daily temperature data to create 675 years of synthetic historical daily temperatures for the Upper Thames River basin in Ontario, Canada. The updated KnnCAD model is shown to adequately reproduce observed monthly temperature characteristics as well as temporal and spatial correlations while simulating reasonable values which can exceed the range of observations. Copyright © 2012 John Wiley & Sons, Ltd.     

A stochastic model and synthesis for near‐fault impulsive ground motions

The orientations of ground motions are paramount when the pulse‐like motions and their unfavorable seismic responses are considered. This paper addresses the stochastic modeling and synthesizing of near‐fault impulsive ground motions with forward directivity effect taking the orientation of the strongest pulses into account. First, a statistical parametric analysis of velocity time histories in the orientation of the strongest pulse with a specified magnitude and various fault distances is performed. A new stochastic model is established consisting of a velocity pulse model with random parameters and a stochastic approach to synthesize high‐frequency velocity time history. The high‐frequency velocity history is achieved by integrating a stochastic high‐frequency accelerogram, which is generated via the modified K‐T spectrum of residual acceleration histories and then modulated by the specific envelope function. Next, the associated parameters of pulse model, envelope function, and power spectral density are estimated by the least‐square fitting. Some chosen parameters in the stochastic model of near‐fault motions based on correlation analysis are regarded as random variables, which are validated to follow the normal or lognormal distribution. Moreover, the number theoretical method is suggested to select efficiently representative points, for generating artificial near‐fault impulsive ground motions with the feature of the strongest pulse, which can be used to the seismic response and reliability analysis of critical structures conveniently. Finally, the simulated ground motions demonstrate that the synthetic ground motions generated by the proposed stochastic model can represent the impulsive characteristic of near‐fault ground motions. Copyright © 2014 John Wiley & Sons, Ltd.     

一种模拟随机数字地层模型的方法

介绍了一种利用计算机生成随机数字地层模型的方法。此方法根据给定的地层厚度及参数统计特性．按截尾正态分布进行随机抽样构建随机分层结构,并对各分层参数随机赋值,完成随机模型的构筑。用此方法可以很容易地在计算机上生成大量符合某种统计特性的随机数字层状地层模型,从而可以在这些模型上进行感兴趣的仿真模拟研究。为展示这一方法的应用,生成了10个第四系随机层状模型,并在这些模型上进行了地震SH波的地震动放大效应数字模拟实验。结果发现随机模型的响应无论在形态特点还是幅值上都与均值模型的响应显著不同,表明用不完全准确的参数模型模拟估计场址地震动响应时必须充分考虑到参数不准导致的误差。     

Driving a lumped hydrological model with precipitation output from weather generators of different complexity

ABSTRACT

This paper deals with the question of whether a lumped hydrological model driven with lumped daily precipitation time series from a univariate single-site weather generator can produce equally good results compared to using a multivariate multi-site weather generator, where synthetic precipitation is first generated at multiple sites and subsequently lumped. Three different weather generators were tested: a univariate “Richardson type” model, an adapted univariate Richardson type model with an improved reproduction of the autocorrelation of precipitation amounts and a semi-parametric multi-site weather generator. The three modelling systems were evaluated in two Alpine study areas by comparing the hydrological output with respect to monthly and daily statistics as well as extreme design flows. The application of a univariate Richardson type weather generator to lumped precipitation time series requires additional attention. Established parametric distribution functions for single-site precipitation turned out to be unsuitable for lumped precipitation time series and led to a large bias in the hydrological simulations. Combining a multi-site weather generator with a hydrological model produced the least bias.     

Generating probabilistic estimates of hydrological response for Irish catchments using a weather generator and probabilistic climate change scenarios

In accounting for uncertainties in future simulations of hydrological response of a catchment, two approaches have come to the fore: deterministic scenario‐based approaches and stochastic probabilistic approaches. As scenario‐based approaches result in a wide range of outcomes, the role of probabilistic‐based estimates of climate change impacts for policy formulation has been increasingly advocated by researchers and policy makers. This study evaluates the impact of climate change on seasonal river flows by propagating daily climate time series, derived from probabilistic‐based climate scenarios using a weather generator (WGEN), through a set of conceptual hydrological models. Probabilistic scenarios are generated using two different techniques. The first technique used probabilistic climate scenarios developed from statistically downscaled scenarios for Ireland, hereafter called SDprob. The second technique used output from 17 global climate models (GCMs), all of which participated in CMIP3, to generate change factors (hereafter called CF). Outputs from both the SDprob and the CF approach were then used in combination with WGEN to generate daily climate scenarios for use in the hydrological models. The range of simulated flow derived with the CF method is in general larger than those estimated with the SDprob method in winter and vice versa because of the strong seasonality in the precipitation signal for the 17 GCMs. Despite this, the simulated probability density function of seasonal mean streamflow estimated with both methods is similar. This indicates the usefulness of the SDprob or probabilistic approach derived from regional scenarios compared with the CF method that relies on sampling a diversity of response from the GCMs. Irrespective of technique used, the probability density functions of seasonal mean flow produced for four selected basins is wide indicating considerable modelling uncertainties. Such a finding has important implications for developing adaptation strategies at the catchment level in Ireland. Copyright © 2011 John Wiley & Sons, Ltd.     

Hydrological functions of a peatland in a Boreal Plains catchment

Streamflow response in Boreal Plains catchments depends on hydrological connectivity between forested uplands, lakes, and peatlands, and their hydrogeomorphic setting. Expected future drying of the Boreal Plains ecozone is expected to reduce hydrological connectivity of landscape units. To better understand run‐off generation during dry periods, we determined whether peatland and groundwater connectivity can dampen expected future water deficits in forests and lakes. We studied Pine Fen Creek catchment in the Boreal Plains ecozone of central Saskatchewan, Canada, which has a large, valley‐bottom, terminally positioned peatland, two lakes, and forested uplands. A shorter intensive study permitted a more detailed partitioning of water inputs and outputs within the catchment during the low flow period, and an assessment of a 10‐year data set provided insight into the function of the peatland over a range of climate conditions. Using a water balance approach, we learned that two key processes regulate flow of Pine Fen Creek. The cumulative impact of landscape unit hydrological connectivity and the peatland's hydrological functional state were needed to understand catchment response. There was evidence of a run‐off threshold which, when crossed, changed the peatland's hydrological function from transmission to run‐off generation. Results also suggest the peatland should behave more often as a transmitter of groundwater than as a generator of run‐off under a drier climate future, owing to a reduced water supply.     

Issues in generating stochastic observables for hydrological models

This paper provides a historical review and critique of stochastic generating models for hydrological observables, from early generation of monthly discharge series, through flood frequency estimation by continuous simulation, to current weather generators. There are a number of issues that arise in such models, from uncertainties in the observational data on which such models must be based, to the potential persistence effects in hydroclimatic systems, the proper representation of tail behaviour in the underlying distributions, and the interpretation of future scenarios.     

A simple method for effective multi-site generation of stochastic hydrologic time series

This paper presents an algorithm for generating stationary stochastic hydrologic time series at multiple sites. The ideas in this paper constitute a radical departure from commonly accepted methodologies. The approach relies on the recent advances in statistical science for simulating random variables with arbitrary marginal distributions and a given covariance structure, and on an algorithm for re-ordering the generated sub-sets of each synthetic year of data such that the annual auto-correlation of desired lag is maintained, along with the autocorrelations between the end of the preceding year and the beginning of the current year. The main features of the proposed algorithm are simplicity and ease of implementation. A numerical test is presented containing the generation of 1000 years of weekly stochastic series for four sites based on the 84 years of historical natural weekly flows from Southern Alberta in Canada.     

Hydrological footprints of urban developments in the Lake Simcoe watershed,Canada: a combined paired‐catchment and change detection modelling approach

Urban sprawl and regional climate variability are major stresses on surface water resources in many places. The Lake Simcoe watershed (LSW) Ontario, Canada, is no exception. The LSW is predominantly agricultural but is experiencing rapid population growth because of its proximity to the Greater Toronto area. This has led to extensive land use changes that have impacted its water resources and altered run‐off patterns in some rivers draining to the lake. Here, we use a paired‐catchment approach, hydrological change detection modelling and remote sensing analysis of satellite images to evaluate the impacts of land use change on the hydrology of the LSW (1994 to 2008). Results show that urbanization increased up to 16% in Lovers Creek, the most urban‐impacted catchment. Annual run‐off from Lovers Creek increased from 239 to 442 mm/year in contrast to the reference catchment (Black River at Washago) where run‐off was relatively stable with an annual mean of 474 mm/year. Increased annual run‐off from Lovers Creek was not accompanied by an increase in annual precipitation. Discriminant function analysis suggests that early (1992–1997; pre‐major development) and late (2004–2009; fully urbanized) periods for Lovers Creek separated mainly based on model parameter sets related to run‐off flashiness and evapotranspiration. As a result, parameterization in either period cannot be used interchangeably to produce credible run‐off simulations in Lovers Creek because of greater scatter between the parameters in canonical space. Separation of early and late‐period parameter sets for the reference catchment was based on climate and snowmelt‐related processes. This suggests that regional climatic variability could be influencing hydrologic change in the reference catchment, whereas urbanization amplified the regional natural hydrologic changes in urbanizing catchments of the LSW. Copyright © 2014 John Wiley & Sons, Ltd.     

The sensitivity of runoff generation to spatial snowpack uniformity in an alpine watershed: Green Lakes Valley,Niwot Ridge Long-Term Ecological Research station

Seasonal water storage in high-elevation alpine catchments are critical sources of water for mountainous regions like the western U.S. The spatial distribution of snow in these topographically complex catchments is primarily governed by orography, solar radiation, and wind redistribution. While the effect of solar shading is relatively consistent from year-to-year, the redistribution of snow due to wind is more variable – capable of producing snowpacks that have varying degrees of uniformity across these hydrologically-important catchments. A reasonable hypothesis is that a warmer climate will cause snowfall to become more dense (i.e. wetter and heavier), possibly leading to less wind redistribution and thus produce a more uniformly distributed snowpack across the landscape. In this study, we investigate the role of increasingly uniform spatial snowpack distributions on streamflow generation in the Green Lakes Valley Niwot Ridge Long Term Ecological Research station, within the headwaters of the Boulder Creek watershed in Colorado. A set of idealized hydrologic simulation experiments driven by reconstructed snowpacks spanning 2001–2014 show that more a more uniform spatial snowpack distribution leads to an earlier melt-out of 31 days on average and tends to produce less total streamflow, with maximum decreases as large as 7.5%. Isolating the role of snowpack heterogeneity from melt-season precipitation, we find that snowpack uniformity reduces total streamflow by as much as 13.2%. Reductions in streamflow are largely explained by greater exposure to solar radiation in the uniformly distributed case relative to a more heterogeneous snowpack, with this exposure driving shifts towards earlier snowmelt and changes in soil water storage. Overall, we find that the runoff efficiency from shallower snowpacks is more sensitive to the effects of uniformity than deeper snowpacks, which has potential implications for a warming climate where shallower snowpacks and enhanced sensitivities may be present.     

Assessment of transmission loss in a Mediterranean karstic watershed (Wadi Natuf,West Bank)

Run‐off transmission loss into karstified consolidated aquifer bedrock below ephemeral streams (wadis) has rarely been described nor quantified. This study presents unique data of long‐term high‐resolution field measurements and field observations in a semiarid to subhumid Mediterranean carbonatic mountainshed. The catchment with a 103 km² surface area is subdivided into 5 subcatchments. Coupled run‐off measurements were made in the different stream sections (reaches), and transmission loss calculated from differences in discharge. Rainfall and run‐off observations from 9 automated precipitation gauging stations and 5 pressure transducers for automatic water level recording are complemented by manual measurements during 34 run‐off events covering a total measurement period of 8 consecutive years. Run‐off generation is strongly event based depending on rainfall intensities and depths. Both, run‐off generation and transmission losses are related to spatial patterns of bedrock lithologies (and hydrostratigraphy). Transmission losses range between 62% and 80% of generated run‐off, with most of the smaller events showing 100% transmission loss. Therefore, although event run‐off coefficients in the mountains can reach up to 22%, only 0.11% of total annual precipitation leaves the catchment as run‐off. Most run‐off infiltrates directly into the regional karst aquifers (Upper Cretaceous carbonate series), with transmission loss intensities of up to 40 mm/h below the stream channels. The factors determining run‐off—such as geology, pedology, vegetation cover and land use, relief and morphology, the semiarid to subhumid Mediterranean climate with a strong elevation gradient, and the patchiness of individual storm events distributed over the winter seasons—as well as the lithology and epikarst features of the bedrock are all characteristic for larger areas in the Mediterranean region. Therefore, we expect that our findings can be generalized to a large extent.     

Multisite generalization of a daily stochastic precipitation generation model

The familiar chain-dependent-process stochastic model of daily precipitation, consisting of a two-state, first-order Markov chain for occurrences and a mixed exponential distribution for nonzero amounts, is extended to simultaneous simulation at multiple locations by driving a collection of individual models with serially independent but spatially correlated random numbers. The procedure is illustrated for a network of 25 locations in New York state, with interstation separations ranging approximately from 10 to 500 km. The resulting process reasonably reproduces various aspects of the joint distribution of daily precipitation observations at the modeled locations. The mixed exponential distributions, in addition to providing substantially better fits than the more conventional gamma distributions, are convenient for representing the tendency for smaller amounts at locations near the edges of wet areas. Means, variances, and interstation correlations of monthly precipitation totals are also well reproduced. In addition, the use of mixed exponential rather than gamma distributions yields interannual variability in the synthetic series that is much closer to the observed.     

Time‐varying suspended sediment‐discharge rating curves to estimate climate impacts on fluvial sediment transport

This study presents time‐varying suspended sediment‐discharge rating curves to model suspended‐sediment concentrations (SSCs) under alternative climate scenarios. The proposed models account for hysteresis at multiple time scales, with particular attention given to systematic shifts in sediment transport following large floods (long‐term hysteresis). A series of nested formulations are tested to evaluate the elements embedded in the proposed models in a case study watershed that supplies drinking water to New York City. To maximize available data for model development, a dynamic regression model is used to estimate SSC based on denser records of turbidity, where the parameters of this regression are allowed to vary over time to account for potential changes in the turbidity‐SSC relationship. After validating the proposed rating curves, we compare simulations of SSC among a subset of models in a climate change impact assessment using an ensemble of flow simulations generated using a stochastic weather generator and hydrologic model. We also examine SSC estimates under synthetic floods generated using a peaks‐over‐threshold model. Our results indicate that estimates of extreme SSC under new climate and hydrologic scenarios can vary widely depending on the selected model and may be significantly underestimated if long‐term hysteresis is ignored when simulating impacts under sequences of large storm event. Based on the climate change scenarios explored here, average annual maximum SSC could increase by as much as 2.45 times over historical values.     

Implications of heterogeneous flood‐frequency distributions on traditional stream‐discharge prediction techniques

Traditional flood‐frequency analysis involves the assumption of homogeneity of the flood distribution. However, floods are often generated by heterogeneous distributions composed of a mixture of two or more populations. Differences between the populations may be the result of a number of factors, including seasonal variations in the flood‐producing mechanisms, changes in weather patterns resulting from low‐frequency climate shifts and/or El Niño/La Nina oscillations, changes in channel routing owing to the dominance of within‐channel or floodplain flow, and basin variability resulting from changes in antecedent soil moisture. Not recognizing these physical processes in conventional flood‐frequency analysis probably is the main reason that many frequency distributions do not provide an acceptable fit to flood data. In this paper, we use long‐term hydroclimatic records from the Gila River basin of south‐east and central Arizona in the USA to explore the extent and significance of mixed populations. First, we discuss the probable causes of heterogeneity in the frequency distribution of annual flood and present evidence of its occurrence. Second, we investigate the implications of using various popular homogeneous distributions for predicting peak flows for basins that exhibit mixed population characteristics. Third, we demonstrate how alternative frequency models that explicitly account for floods generated by a mixture of two or more populations are both hydrologically and statistically more appropriate. We illustrate how the selection of the most plausible distribution for flood‐frequency analysis also should be based on hydrological reasoning as opposed to the sole application of the traditional statistical goodness‐of‐fit tests. Copyright © 2002 John Wiley & Sons, Ltd.     

Suitability of CLIGEN precipitation estimates based on an updated database and their impacts on urban runoff: a case study of the Great Lakes Region,USA

ABSTRACT

Weather generators rely on historical meteorological records to simulate time series of synthetic weather sequences, the quality of which has direct influence on model applications. The climate generator CLIGEN’s database has recently been updated to comprise consistent historical records from 1974 to 2013 (updated CLIGEN database, UCD) compared to the current database in which records are of different lengths. In this study, CLIGEN’s performance in estimating precipitation using UCD (eight stations) and the subsequent impact on urban runoff simulations (371 stations) were evaluated in the Great Lakes Region, USA. Generally, UCD-based precipitation could replicate observed daily precipitation up to the 99.5th percentile, but maximum precipitation was underestimated. Results from the Long-Term Hydrologic Impact Assessment model using UCD-based precipitation showed about 0.57 billion cubic meters more (14.9%) average annual runoff being simulated compared with simulations based on the current CLIGEN database. Overall, CLIGEN with the updated database was found suitable for providing precipitation estimates and for use with modeling urban runoff or urbanization effects.     

Finding the most appropriate precipitation probability distribution for stochastic weather generation and hydrological modelling in Nordic watersheds

Six precipitation probability distributions (exponential, Gamma, Weibull, skewed normal, mixed exponential and hybrid exponential/Pareto distributions) are evaluated on their ability to reproduce the statistics of the original observed time series. Each probability distribution is also indirectly assessed by looking at its ability to reproduce key hydrological variables after being used as inputs to a lumped hydrological model. Data from 24 weather stations and two watersheds (Chute‐du‐Diable and Yamaska watersheds) in the province of Quebec (Canada) were used for this assessment. Various indices or statistics, such as the mean, variance, frequency distribution and extreme values are used to quantify the performance in simulating the precipitation and discharge. Performance in reproducing key statistics of the precipitation time series is well correlated to the number of parameters of the distribution function, and the three‐parameter precipitation models outperform the other models, with the mixed exponential distribution being the best at simulating daily precipitation. The advantage of using more complex precipitation distributions is not as clear‐cut when the simulated time series are used to drive a hydrological model. Although the advantage of using functions with more parameters is not nearly as obvious, the mixed exponential distribution appears nonetheless as the best candidate for hydrological modelling. The implications of choosing a distribution function with respect to hydrological modelling and climate change impact studies are also discussed. Copyright © 2012 John Wiley & Sons, Ltd.     

Weather types,runoff and sediment yield in a Mediterranean mountain landscape

The relationship between the synoptic weather types (WTs), runoff and sediments in a Mediterranean mountain landscape was analysed. The study was performed between 2005 and 2012 using one of the most complete and extensive daily databases of rainfall, runoff and sediment recorded in the Spanish Pyrenees, coupled with WTs defined from the National Meteorological Center/National Center for Atmospheric Research (NMC/NCAR) 40‐Year Reanalysis Project ? ? Correction added on 18 December 2013, after online publication: EMULATE project was replaced with NMC/NCAR 40‐Year Reanalysis Project.
. The results show that the three wettest WTs accounted for 30% of rainy days and 46% of rainfall, but comprised only 13% of total daily records. To obtain a much more robust association between WTs and rainfall an analysis was carried out using a longer rainfall record (1989–2011). The analyses confirmed that the results obtained from the Araguás catchment are representative of a longer time span. The cyclonic, north‐westerly and westerly WTs play an important role in runoff generation, coinciding with the wettest WTs. Extreme floods are commonly associated also with south‐westerly and westerly airflows, whereas less flooding was generated under cyclonic circulations. Sediment transport was concentrated in 2.9% of total time mainly related to westerly WTs. Seasonal differences exist in WT frequency. In winter and spring north‐west and West are the most prominent WTs related to rainfall, water and sediment yield, although in spring cyclonic frequency was higher. During autumn north‐west and south‐west were the most frequent, but sediment yield was produced nearby under south‐west flow. In summer the WTs that produce sediment are north and west. A magnitude–frequency analysis shows the different behaviour of WTs according to their efficiency in producing runoff and sediment. A study with different monitored areas around the Iberian Peninsula and the Mediterranean basin would be very valuable in providing information for hydrological and sediment behaviour under the current conditions of global climate change. Copyright © 2013 John Wiley & Sons, Ltd.     

A multivariate non‐parametric model for synthetic generation of daily streamflow

A p‐order multivariate kernel density model based on kernel density theory has been developed for synthetic generation of multivariate variables. It belongs to a kind of data‐driven approach and is able to avoid prior assumptions as to the form of probability distribution (normal or Pearson III) and the form of dependence (linear or non‐linear). The p‐order multivariate kernel density model is a non‐parametric method for synthesis of streamflow. The model is more flexible than conventional parametric models used in stochastic hydrology. The effectiveness and satisfactoriness of this model are illustrated through its application to the simultaneous synthetic generation of daily streamflow from Pingshan station and Yibin‐Pingshan region (Yi‐Ping region) of the Jinsha River in China. Copyright © 2007 John Wiley & Sons, Ltd.