Implications For Hydrological Modeling Research Articles

Non-linearity in event runoff generation in a small agricultural catchment.

Understanding the role of soil moisture and other controls in runoff generation is important for predicting runoff across scales. This paper aims to identify the degree of non‐linearity of the relationship between event peak runoff and potential controls for different runoff generation mechanisms in a small agricultural catchment. The study is set in the 66 ha Hydrological Open Air Laboratory, Austria, where discharge was measured at the catchment outlet and for 11 sub‐catchments or hillslopes with different runoff generation mechanisms. Peak runoff of 73 events was related to three potential controls: event precipitation, soil moisture and groundwater levels. The results suggest that the hillslopes dominated by ephemeral overland flow exhibit the most non‐linear runoff generation behaviour for its controls; runoff is only generated above a threshold of 95% of the maximum soil moisture. Runoff generation through tile drains and in wetlands is more linear. The largest winter and spring events at the catchment outlet are caused by runoff from hillslopes with shallow flow paths (ephemeral overland flow and tile drainage mechanisms), while the largest summer events are caused by other hillslopes, those with deeper flow paths or with saturation areas throughout the year. Therefore, the response of the entire catchment is a mix of the various mechanisms, and the groundwater contribution makes the response more linear. The implications for hydrological modelling are discussed.

Extensibility of U-Net Neural Network Model for Hydrographic Feature Extraction and Implications for Hydrologic Modeling

Accurate maps of regional surface water features are integral for advancing ecologic, atmospheric and land development studies. The only comprehensive surface water feature map of Alaska is the National Hydrography Dataset (NHD). NHD features are often digitized representations of historic topographic map blue lines and may be outdated. Here we test deep learning methods to automatically extract surface water features from airborne interferometric synthetic aperture radar (IfSAR) data to update and validate Alaska hydrographic databases. U-net artificial neural networks (ANN) and high-performance computing (HPC) are used for supervised hydrographic feature extraction within a study area comprised of 50 contiguous watersheds in Alaska. Surface water features derived from elevation through automated flow-routing and manual editing are used as training data. Model extensibility is tested with a series of 16 U-net models trained with increasing percentages of the study area, from about 3 to 35 percent. Hydrography is predicted by each of the models for all watersheds not used in training. Input raster layers are derived from digital terrain models, digital surface models, and intensity images from the IfSAR data. Results indicate about 15 percent of the study area is required to optimally train the ANN to extract hydrography when F1-scores for tested watersheds average between 66 and 68. Little benefit is gained by training beyond 15 percent of the study area. Fully connected hydrographic networks are generated for the U-net predictions using a novel approach that constrains a D-8 flow-routing approach to follow U-net predictions. This work demonstrates the ability of deep learning to derive surface water feature maps from complex terrain over a broad area.

K in an Urban World: New Contexts for Hydraulic Conductivity.

Hydraulic conductivity (K) is a key hydrologic parameter widely recognized to be difficult to estimate and constrain, with little consistent assessment in disturbed, urbanized soils. To estimate K, it is either measured, or simulated by pedotransfer functions, which relate K to easily measured soil properties. We measured K in urbanized soils by double-ring infiltrometer (K dring), near-saturated tension infiltrometry (K minidisk), and constant head borehole permeametry (K borehole), along with other soil properties across the major soil orders in 12 United States cities. We compared measured K with that predicted from the pedotransfer function, ROSETTA. We found that regardless of soil texture, K dring was consistently larger than K minidisk; with the latter having slightly less sample variance. K borehole was dependent upon specific subsurface conditions, and contrary to common expectations, did not always decrease with depth. Based on either soil textural class, or percent textural separates (sand, silt clay), ROSETTA did not accurately predict measured K for surface nor subsurface soils. We go on to discuss how K varies in urban landscapes, the role of measurement methods and artifacts in the perception of this metric, and implications for hydrologic modeling. Overall, we aim to inspire consistency and coherence when addressing K-related challenges in sustainable urban water management.

Evaluation of gridded climate datasets over Canada using univariate and bivariate approaches: Implications for hydrological modelling

This study evaluates the univariate and bivariate characteristics of five gridded products with respect to the Adjusted and Homogenized Canadian Climate Data for 1980–2010. Spatial variations of temperature and precipitation biases including the ones corresponding to the magnitude and frequency of extremes are assessed over Canada. The dependence structure between the two variables is analysed across each dataset using a goodness of fit test based on copulas. The propagation of univariate and bivariate biases into streamflow simulations is then investigated by driving a semi-distributed hydrological model for three watersheds with distinct characteristics in western Canada. The input data include original (unadjusted) and bias-corrected climate data based on the univariate quantile delta mapping and multivariate bias correction approaches. The univariate analyses show that all datasets have relatively significant cold and wet biases to the west of the Rocky Mountains and hot and dry biases over the Prairies. Bivariate evaluations using copulas show that all products fail to capture the dependence structure between temperature and precipitation at the majority of the locations, which can undermine their suitability for compound event assessments. The differences between univariate and multivariate bias correction approaches are highlighted by the significant differences in the interrelationships between precipitation and temperature. Hydrological modelling results show major improvements in the detection of extremes after correcting the bivariate biases of the input datasets.

Interpolated or satellite-based precipitation? Implications for hydrological modeling in a meso-scale mountainous watershed on the Qinghai-Tibet Plateau

The interpolation algorithms and the satellite-based precipitation products (SPPs) are two major approaches to estimating spatially distributed precipitation. This would inevitably raise the questions: whether the interpolation algorithms could outperform the SPPs or vice versa? And what’s the implications for hydrological modeling? These questions, however, have received little attention in the literature. This study compared the performances of two quasi-physically based interpolation algorithms (i.e., MicroMet and PrecLaps) with the widely used SPPs, i.e., Tropical Rainfall Measuring Mission Multisatellite Precipitation Analysis (TMPA) and Integrated Multi-Satellite Retrievals for Global Precipitation Measurement (IMERG), in and around the Babao River Basin (BRB), a meso-scale mountainous watershed on the Qinghai-Tibet Plateau. Meanwhile, the hydrological utilities of the interpolated and the satellite-based precipitation were evaluated by using the Distributed Hydrology Soil Vegetation Model (DHSVM). Results indicate that TMPA and IMERG perform worse than the interpolation algorithms in estimating daily precipitation, while they show comparable and even higher performance in reproducing monthly precipitation. Both the interpolation algorithms and the SPPs have an obvious lower performance in winter than the other seasons. DHSVM with the TMPA-based or IMERG-based precipitation, when not subjected to additional calibration, performs worse than those with the interpolated precipitation in simulating streamflow. Nevertheless, interestingly, the ET simulations consistently match well with the independent remote sensing (RS)-based ET product, as indicated by the higher coefficient of determination (R2 ≥ 0.85) and Nash-Sutcliffe efficiency (NSE ≥ 0.72). The additional calibration of DHSVM with the satellite-based precipitation could enhance the streamflow simulation accuracy substantially, with the NSE increasing by 70.59–132% in the validation period. This, however, would bring about larger discrepancies between the simulated and the RS-based ET in summer. The study further discussed the implications of these findings for hydrological modeling over the data-scarce mountainous watersheds, and revealed the uncertainties associated with the rain gauge density.

Evaluating infiltration models and pedotransfer functions: Implications for hydrologic modeling

Soil is a critical medium that governs hydrological processes in a landscape; however, soil data quality and resolution are not given much emphasis during watershed modeling and developing practical management practices, especially in data scarce regions such as in the Ethiopian highlands. This is primarily because measuring most soil parameters is costly, laborious, and at times impractical. Infiltration models and Pedotransfer functions (PTFs) may bridge this gap by estimating such complex soil parameters from easily measurable ones. However, most infiltration models and PTFs were developed for temperate regions and there is no comprehensive experimental or modeling work done to evaluate the robustness of such approaches in the Ethiopian highlands. Therefore, this study aims to i) evaluate the performance of three infiltration models, namely Horton, Koistakov, and Philip in simulating soil infiltration rates and ii) test the reliability of nine PTFs in estimating saturated hydraulic conductivity and soil moisture contents at field capacity and permanent wilting point in a case study in the Anjeni watershed. Soils in the Anjeni watershed on average contain higher percentages of clay and silt with 42 and 34%, respectively; and have acidic pH and low organic matter. Steady infiltration rate (fs) and saturated hydraulic conductivity (Ks) ranged between 0.19 and 30.6 mm h−1 and 0.19 and 3.3 mm h−1, respectively. All the three infiltration models fitted observed data very well with R2 and Nash-Sutcliffe efficiency (NS) values between 0.95 and 0.97 each. While the Horton and Koistakov models had the same R2 and NS values, the Horton provided smaller Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) values. Some of the tested PTFs showed promising results in estimating saturated hydraulic conductivity and soil moisture content; however, all PTFs showed limitations which warrant further evaluation using observed data. This study showed that soil hydrologic parameters vary significantly even within a small area, and widely available soil datasets (e.g., Africa Soil Information Service (AfSIS) and FAO Harmonized World Soil Database) fail to reflect actual field conditions. This suggests that emphasis should be given to soil data quality and resolution during watershed modeling, and land and water management planning.

Global estimation of effective plant rooting depth: Implications for hydrological modeling

Plant rooting depth (Zr) is a key parameter in hydrological and biogeochemical models, yet the global spatial distribution of Zr is largely unknown due to the difficulties in its direct measurement. Additionally, Zr observations are usually only representative of a single plant or several plants, which can differ greatly from the effective Zr over a modelling unit (e.g., catchment or grid-box). Here, we provide a global parameterization of an analytical Zr model that balances the marginal carbon cost and benefit of deeper roots, and produce a climatological (i.e., 1982-2010 average) global Zr map. To test the Zr estimates, we apply the estimated Zr in a highly transparent hydrological model (i.e., the Budyko-Choudhury-Porporato (BCP) model) to estimate mean annual actual evapotranspiration (E) across the globe. We then compare the estimated E with both water balance-based E observations at 32 major catchments and satellite grid-box retrievals across the globe. Our results show that the BCP model, when implemented with Zr estimated herein, optimally reproduced the spatial pattern of E at both scales (i.e., R2=0.94, RMSD=74 mm yr−1 for catchments, and R2=0.90, RMSD=125 mm yr−1 for grid-boxes) and provides improved model outputs when compared to BCP model results from two already existing global Zr datasets. These results suggest that our Zr estimates can be effectively used in state-of-the-art hydrological models, and potentially biogeochemical models, where the determination of Zr currently largely relies on biome type-based look-up tables. This article is protected by copyright. All rights reserved.

Putting the cat in the box: why our models should consider subsurface heterogeneity at all scales

Subsurface heterogeneity is found across the whole hydrological cycle. It causes preferential flow in the soil, the vadose zone, and the bedrock. Aside of this, it has many other consequences for water flow and storage. This opinion paper discusses the implications of subsurface heterogeneity across the hydrological cycle and its implications for hydrological modeling. It focuses on the representation (and misrepresentation) of subsurface heterogeneity within hydrological models at different scales. Possible limitations to the prediction performance of hydrological models that disregard subsurface heterogeneity are introduced and discussed. The paper ends with practical recommendations to include subsurface heterogeneity into hydrological models, to parameterize them realistically, and to present possible directions for future research. WIREs Water 2016, 3:478–486. doi: 10.1002/wat2.1146This article is categorized under: Science of Water > Hydrological Processes

Dynamically downscaled winter precipitation over complex terrain of the Central Rockies of Western Montana, USA

[1] We evaluate the results of dynamically downscaled winter precipitation over Western Montana using the weather research and forecasting (WRF) model through comparison with estimates from the observationally based parameter-elevation regressions on independent slopes model (PRISM). Seven years (six winters) from 2000 to 2006 are simulated at 4 km resolution to assess the similarities and differences between the two models as well as the implications for hydrologic modeling. Inherent biases in both approaches are apparent, highlighting the difficulty in climate model validation. Results show general agreement between the two models in the spatial distribution of winter precipitation. A principal component analysis shows similar spatial patterns between models in the leading six components suggesting that the main processes that drive the spatial distribution of precipitation were properly captured. The first component explains almost 70% of total variance, and the first three components explain more than 85% in both data sets. The largest differences between the two data sets exist in areas at high elevation and upstream of the continental divide where observations are sparse. In these areas, WRF consistently predicts higher amounts of precipitation and larger interannual variability than PRISM. We suggest that these results are realistic for impingement of moist air masses on topography and, if correct, could have significant implications in flood forecasting, water resource management, and climate change studies.

Spatial-Scale Characteristics of Precipitation Simulated by Regional Climate Models and the Implications for Hydrological Modeling

Abstract Precipitation simulated by regional climate models (RCMs) is generally biased with respect to observations, especially at the local scale of a few tens of kilometers. This study investigates how well two different RCMs are able to reproduce the spatial correlation patterns of observed summer precipitation for the central United States. On local scales, gridded precipitation observations and simulated precipitation are compared for the period of the 1987 First International Satellite Land Surface Climatological Project Field Experiment (FIFE) campaign. The results show that spatial correlation length scales on the order of 130 km are found in both observed data and RCM simulations. When simulations and observations are aggregated to different grid sizes, the pattern correlation significantly decreases when the aggregation length is less than roughly 100 km. Furthermore, the intermodel standard deviation between simulations with different domains or resolutions increases for aggregation lengths below ~130 km. Below this length scale there is a high level of randomness in the precise location of precipitation events. Conversely, spatial correlation values increase above this length scale, reflecting larger predictive certainty of the RCMs at larger scales. The findings on aggregated grid scales are shown to be largely independent of the underlying RCMs grid resolutions but not of the overall size of RCM domain. With regard to hydrological modeling applications, these findings indicate that precipitation extracted from the present RCM simulations at a catchment scale below the intermodel standard deviation length cannot be expected to accurately match observations.

Characteristics of concentrated flow hydraulics for rangeland ecosystems: implications for hydrologic modeling

ABSTRACTConcentrated flow is often the dominant source of water erosion following disturbance on rangelands. Because of the lack of studies that explain the hydraulics of concentrated flow on rangelands, cropland‐based equations have typically been used for rangeland hydrology and erosion modeling, leading to less accurate predictions due to different soil and vegetation cover characteristics. This study investigates the hydraulics of concentrated flow using unconfined field experimental data over diverse rangeland landscapes within the Great Basin Region, United States. The results imply that the overall hydraulics of concentrated flow on rangelands differ significantly from those of cropland rills. Concentrated flow hydraulics on rangelands are largely controlled by the amount of cover or bare soil and hillslope angle. New predictive equations for concentrated flow velocity (R2 = 0·47), hydraulic friction (R2 = 0·52), and width (R2 = 0·4) representing a diverse set of rangeland environments were developed. The resulting equations are applicable across a wide span of ecological sites, soils, slopes, and vegetation and ground cover conditions and can be used by physically‐based rangeland hydrology and erosion models to estimate rangeland concentrated flow hydraulic parameters. Published in 2011. This article is a US Government work and is in the public domain in the USA.

The influence of elevation error on the morphometrics of channel networks extracted from DEMs and the implications for hydrological modelling

AbstractStream network morphometrics have been used frequently in environmental applications and are embedded in several hydrological models. This is because channel network geometry partly controls the runoff response of a basin. Network indices are often measured from channels that are mapped from digital elevation models (DEMs) using automated procedures. Simulations were used in this paper to study the influence of elevation error on the reliability of estimates of several common morphometrics, including stream order, the bifurcation, length, area and slope ratios, stream magnitude, network diameter, the flood magnitude and timing parameters of the geomorphological instantaneous unit hydrograph (GIUH) and the network width function. DEMs of three UK basins, ranging from high to low relief, were used for the analyses. The findings showed that moderate elevation error (RMSE of 1·8 m) can result in significant uncertainty in DEM‐mapped network morphometrics and that this uncertainty can be expressed in complex ways. For example, estimates of the bifurcation, length and area ratios and the flood magnitude and timing parameters of the GIUH each displayed multimodal frequency distributions, i.e. two or more estimated values were highly likely. Furthermore, these preferential estimates were wide ranging relative to the ranges typically observed for these indices. The wide‐ranging estimates of the two GIUH parameters represented significant uncertainty in the shape of the unit hydrograph. Stream magnitude, network diameter and the network width function were found to be highly sensitive to elevation error because of the difficulty in mapping low‐magnitude links. Uncertainties in the width function were found to increase with distance from outlet, implying that hydrological models that use network width contain greater uncertainty in the shape of the falling limb of the hydrograph. In light of these findings, care should be exercised when interpreting the results of analyses based on DEM‐mapped stream networks. Copyright © 2007 John Wiley & Sons, Ltd.

Delineating Source Areas for Runoff in Depressional Landscapes: Implications for Hydrologic Modeling

A digital terrain analysis of the Huron River watershed, using a new spatial analytic procedure, suggests that many areas do not contribute surface runoff directly to the Huron River. These areas are depressions and internally drained valleys of glacial origin. Understanding gained from this analysis should significantly improve the design of lumped-parameter hydrologic models of undulatory glacial terrains. The analysis also shows that man-made drainage features can greatly increase runoff by connecting parts of the watershed that were previously topographically isolated. Consequently, the current use of contemporary subbasin divides is an inaccurate means of calculating surface runoff in southeast Michigan, because of the incorporation of areas that are not directly contributing runoff to the main river system. For example, 63.1% of the Huron River watershed is unconnected. An example is presented for Mallets Creek, showing how the new procedure can automate definitions of overland flow planes and catchments in a hydrological model. This model is then used to evaluate the sensitivity of runoff to the introduction of storm sewers that change the potential contributing source area. Results suggest that anthropogenic drainage features have doubled the effective size of Mallets Creek watershed.

The effects of land use on soil moisture variation in the Danangou catchment of the Loess Plateau, China

Soil moisture plays a critical role in both crop growth and vegetation restoration in semiarid environments. Its spatial and temporal variability results from topography, soils, vegetation and land uses. However, little knowledge exists about the effects of land use structure on soil moisture variability. In order to analyze soil moisture variations in relation to land use patterns, five land use structures and seven land use types were selected to monitor soil moisture. Soil moisture measurements were performed biweekly at 26 locations in a small catchment on the Loess Plateau of China from May to October 1998. The measurements were taken using a Theta-Probe at five depths of soil profile (0–5, 15–20, 25–30, 45–50 and 70–75 cm). These data were analyzed for soil moisture variations in space and time for seven land use types. Three peaks and three troughs of soil moisture variations during the growing season were found. The influence of shrubland on mean soil moisture within 0–70 cm was significantly different from that of cropland, orchard and intercropping land. Three types of soil moisture changes in profile were classified as increasing, decreasing and waving types. An analysis of the differences in soil moisture for five land use structures indicated that the influences of land use patterns on soil moisture were complex. This study provides an insight into the implications for hydrological modeling, runoff and erosion control in this area.

On the physical geometry concept at the basis of space/time geostatistical hydrology

The objective of this paper is to show that the structure of the spatiotemporal continuum has important implications in practical stochastic hydrology (e.g., geostatistical analysis of hydrologic sites) and is not merely an abstract mathematical concept. We propose that the concept of physical geometry as a spatiotemporal continuum with properties that are empirically defined is important in hydrologic analyses, and that the elements of the spatiotemporal geometry (e.g., coordinate system and space/time metric) should be selected based on the physical properties of the hydrologic processes. We investigate the concept of space/time distance (metric) in various physical spaces, and its implications for hydrologic modeling. More specifically, we demonstrate that physical geometry plays a crucial role in the determination of appropriate spatiotemporal covariance models, and it can affect the results of geostatistical operations involved in spatiotemporal hydrologic mapping.

Preferred states in spatial soil moisture patterns: Local and nonlocal controls

In this paper we develop a conceptual and observational case in which soil water patterns in temperate regions of Australia switch between two preferred states. The wet state is dominated by lateral water movement through both surface and subsurface paths, with catchment terrain leading to organization of wet areas along drainage lines. We denote this as nonlocal control. The dry state is dominated by vertical fluxes, with soil properties and only local terrain (areas of high convergence) influencing spatial patterns. We denote this as local control. The switch is described in terms of the dominance of lateral over vertical water fluxes and vice versa. When evapotranspiration exceeds rainfall, the soil dries to the point where hydraulic conductivity is low and any rainfall that occurs essentially wets up the soil uniformly and is evapotranspired before any significant lateral redistribution takes place. As evapotranspiration decreases and/or rainfall increases, areas of high local convergence become wet, and runoff that is generated moves downslope, rapidly wetting up the drainage lines. In the wet to dry transitional period a rapid increase in potential evapotranspiration (and possibly a decrease in rainfall) causes drying of the soil and “shutting down” of lateral flow. Vertical fluxes dominate and the “dry” pattern is established. Three data sets from two catchments are presented to support the notion of preferred states in soil moisture, and the results of a modeling exercise on catchments from a range of climatic conditions illustrate that the conclusions from the field studies may apply to other areas. The implications for hydrological modeling are discussed in relation to methods for establishing antecedent moisture conditions for event models, for distribution models, and for spatially distributing bulk estimates of catchment soil moisture using indices.

Scale and the Nature of Spatial Variability: Field Examples Having Implications for Hydrologic Modeling

In this paper we examine how the nature of spatial variability affects hydrologic response over a range of scales using five field studies as examples. The nature of variability was characterized as either stochastic, when random, or deterministic, when due to known, nonrandom sources. We have emphasized how that characterization may change with the scale of hydrologic model. The five field examples, along with corresponding sources of variability, were (1) infiltration and surface runoff affected by shrub canopy, (2) groundwater recharge affected by soil depth, (3) groundwater recharge and streamflow affected by small‐scale topography, (4) frozen soil runoff affected by elevation, and (5) snowfall distribution affected by large‐scale topography. In each example there was a scale, the deterministic length scale, over which the hydrologic response was strongly dependent upon the specific, location‐dependent ecosystem properties. Smaller‐scale variability may be represented as either stochastic or homogeneous with nonspatial data. In addition, changes in scale or location sometimes resulted in the introduction of larger‐scale sources of variability that subsume smaller‐scale sources. Thus recognition of the nature and sources of variability can reduce data requirements by focusing on important sources of variability and using nonspatial data to characterize variability at scales smaller than the deterministic length scale. All the sources of variability described are present in the same watershed and affect hydrologic response simultaneously. Physically based models should therefore utilize both spatial and stochastic data where scale appropriate. Other implications for physically based modeling are that modeling algorithms should reflect larger‐scale variability which generally has greater impact and that model and measurement grids should be consistent with the nature of variability.

Variability of unsaturated zone water transport parameters: implications for hydrological modelling. 2. Predicted vs. in situ measurements and evaluation of methods

Two methods to estimate the saturated and unsaturated hydraulic conductivity of soil were compared in a catchment in south-west England. The catchment consists of three land use domains — arable, permanent grass and woodland. The methods used were in situ measurements using a Guelph permeameter and the predictive model of Rawls and Brakensiek, which uses percentage of sand, percentage of clay and porosity as input. In grassland and woodland no significant differences were found between the mean saturated and unsaturated hydraulic conductivity estimated by the two methods. It is suggested that a significant difference found for the arable land is due to the effect of farming operations on soil permeability. On a catchment scale there were no significant differences between the methods. The results obtained using the predictive model are encouraging and worth further investigation. The advantages and disadvantages of the two methods are evaluated.

Variability of unsaturated zone water transport parameters: implications for hydrological modelling. 1. In situ measurements

Physically based hydrological models require values of saturated and unsaturated hydraulic conductivity to solve the equations used to describe subsurface flow. Variability of these parameters within and between various land uses and soil types is an important consideration if hydrological behaviour is to be described realistically. A catchment in south-west England has been chosen to validate the European Hydrological System (Système Hydrologique Européen — SHE). The catchment includes three land use domains — arable (including temporary grass), permanent grassland and woodland. For the unsaturated zone component of the SHE both saturated hydraulic conductivity, K s, and unsaturated hydraulic conductivity as a function of matric suction, K(ψ), are required. The aim of this investigation was to measure the parameters for input to the model, to assess the variability of these parameters and to determine whether or not parameters should be used for each domain independently. A Guelph permeameter was used to obtain K s and α, which describes the slope of the logarithmic unsaturated hydraulic conductivity function ( K( ψ) = K s exp( αψ)). Measurements at four depths were carried out. Domains were chosen according to land use — arable land (including temporary grass), permanent grassland and woodland. Both K s and α were found to be lognormally distributed. The K s values were highest in the grassland and lowest in arable land. The difference between means was significant for K s but not for α in the three domains. High values of K s were attributed to preferential flow along slate faces and root-induced channels in the grassland and woodland, respectively, as indicated by dye tests. Finally, it was concluded that, as K s and K(ψ) differ significantly between domains, different hydrological behaviour would be expected. It is therefore recommended that different parameter sets be used for modelling the individual domains.