Streamflow sensitivity analysis to climate change for a large water‐limited basin

Partitioning precipitation (P) between streamflow (Q) and actual evapotranspiration (E_a) on a basin scale is controlled by climate change in combination with catchment characteristics. Fu's formulation of the Budyko framework was used to estimate Q as a function of two meteorological variables, P and potential evaporation (E_p), and one adjustable parameter reflecting characteristics of catchment conditions (ω). Results show that ω reflects the impacts of catchment characteristics on the partitioning of P between Q and E_a for the different water yielding regions. As predicted, Q was more sensitive to P than to comparable changes in E_p for the whole of the Yellow River Basin (YRB), a water‐limited basin, while it was shown to be highly sensitive to changes in P, E_p, and ω in the low water yielding region (LWYR) of the basin, followed by YRB and the high water yielding region of the basin. The high sensitivity of Q to P, E_p, and ω in LWYR indicates that the management of catchments within these zones is critical to the management of overall basin flow, mitigating impacts of climate change on Q. The Budyko framework, incorporating the adjustable parameter ω, outlines interactions between Q, climate, and characteristics specific to different water yielding regions. It also provides a new approach in understanding hydrological process response to climate change. Due to the obscure physical attributes of ω, an explanation of the parameter using soil or vegetation characteristics will aid in the understanding of the eco‐hydrological behaviour of catchments and help to provide more detailed catchment management options for which to mitigate climate change with respect to concerns regarding agricultural water usage. Copyright © 2013 John Wiley & Sons, Ltd.     

Improving evapotranspiration estimation models through quantitative watershed parameter analysis and remote sensing applications

Numerous models had been developed to predict the annual evapotranspiration (ET) in vegetated lands across various spatial scales. Fu's (Scientia Atmospherica Sinica, 5, 23–31) and Zhang's (Water Resources Research, 37, 701–708) ET simulation models have emerged as highly effective and have been widely used. However, both formulas have the non-quantitative parameters (m in Fu's model and w in Zhang's model). Based on the collected 1789 samples from global long-term hydrological studies, this study discovered significant relations between m (or w) and vegetation coverage or greenness in collected catchments. Then, we used these relations to qualify the parameters in both Zhang's and Fu's models. Results show that the ET estimation accuracies of Fu's (or Zhang's) model are significantly improved by about 13.49 mm (or 6.74 mm) for grassland and cropland, 38.52 mm (or 29.84 mm) for forest and shrub land (coverage<40%), 19.74 mm (or 16.17 mm) for mixed land (coverage<40%), respectively. However, Zhang's model shows higher errors compared with Fu's model, especially in regions with high m (or w) values, such as those with dense vegetations or P/E₀ (annual precipitation to annual potential ET) smaller than 1.0. Additionally, this study also reveals that for regions with vegetation cover less than 40%, the annual ET is not only determined by vegetation types, but also relates to the sizes of vegetation-covered areas. Conversely, for regions with vegetation cover more than 40%, the annual ET is mainly determined by the vegetation density rather than vegetation types or vegetation coverage. Thus, linking m (or w) parameters with vegetation greenness allows leveraging remote sensing for forest management in data-scarce areas, safeguarding regional water resources. This study pioneers integrating vegetation-related indices with basin parameters, advocating for their crucial role in more effective hydrological modelling.     

Modelling interannual variations in catchment evapotranspiration considering vegetation and climate seasonality using the Budyko framework

The Budyko framework is an efficient tool for investigating catchment water balance, focusing on the effects of seasonal changes in climate (S) and vegetation cover (M) on catchment evapotranspiration (ET). However, the effects of vegetation seasonality on ET remain largely unknown. The present study explored these effects by modelling interannual variations in ET considering vegetation and climate seasonality using the Budyko framework. Reconstructed 15-day GIMMS NDVI_3g timeseries data from 1982 to 2015 were used to estimate M and extract the relative duration of the vegetation growing season (GL) in the Yellow River Basin (YRB). To characterize S, seasonal variations in precipitation and potential ET were extracted using a Gaussian algorithm. Analysis of the observed datasets for 19 catchments revealed that interannual variation in the catchment parameter ϖ (in Fuh's equation) was significantly and positively correlated with M and GL. Conversely, ϖ was significantly but negatively correlated with S. Furthermore, stepwise linear regression was used to calibrate the empirical formula of ϖ for these three dimensionless parameters. Following validation, based on observations in the remaining 11 catchments, ϖ was integrated into Fuh's equation to accurately estimate annual ET. Over 79% subcatchments showed an upward trend (0.9 mm yr⁻¹), whereas fewer than 21% subcatchments showed a downward trend (−0.5 mm yr⁻¹) across YRB. In the central region of the middle reach, ET increased with increased M, prolonged GL, and decreased S, whereas in the source region of YRB, ET decreased with decreased M and shortened GL. Our study provides an alternative method to estimate interannual ET in ungauged catchments and offers a novel perspective to investigate hydrological responses to vegetation and climate seasonality in the long-term.     

Long-range forecasting of suspended sediment

This study examines the employment of two methods, multiple linear regression (MLR) and an artificial neural network (ANN), for multistep ahead forecasting of suspended sediment. The autoregressive integrated moving average (ARIMA) model is considered for one-step ahead forecasting of sediment series in order to provide a comparison with the MLR and ANN methods. For one- and two-step ahead forecasting, the ANN model performance is superior to that of the MLR model. For longer ranges, MLR models provide better accuracy, but there is an important assumption violation. The Durbin-Watson statistics of the MLR models show a noticeable decrease from 1.3 to 0.5, indicating that the residuals are not dependent over time. The scatterplots of the three methods (MLR, ARIMA and ANN) for one-step ahead forecasting for the validation period illustrate close fits with the regression line, with the ANN configuration having a slightly higher R² value.     

Diagnostic study and modeling of the annual positive water temperature onset

A data-driven model is designed using artificial neural networks (ANN) to predict the average onset for the annual water temperature cycle of North-American streams. The data base is composed of daily water temperature time series recorded at 48 hydrometric stations in Québec (Canada) and northern US, as well as the geographic and physiographic variables extracted from the 48 associated drainage basins. The impact of individual and combined drainage area characteristics on the stream annual temperature cycle starting date is investigated by testing different combinations of input variables. The best model allows to predict the average temperature onset for a site, given its geographical coordinates and vegetation and lake coverage characteristics, with a root mean square error (RMSE) of 5.6 days. The best ANN model was compared favourably with parametric approaches.     

Irrigation impact on annual water balance of the oases in Tarim Basin,Northwest China

By taking the sum of annual precipitation and lateral water input (in which irrigation water withdrawal is the main component) for water availability, the Budyko hypothesis and Fu's formula derived from it was extended to the study of oases in the Tarim Basin, Northwest China. For both long‐term (multi‐year) and annual values on water balances in the 26 oases subregions, the extended Fu's formula was confirmed. Regional patterns on water balance on the 26 oases subregions were related to change in land‐use types due to increased area for irrigation. Moreover, an empirical formula for the parameter was established to reflect the influences of change in land use on water balance. The extended Budyko framework was employed to evaluate the impact of irrigation variability on annual water balance. According to the multi‐year mean timescale, variabilities in actual evapotranspiration in the oases were mainly controlled by variability in irrigation water withdrawal rather than potential evapotranspiration. The influences of variability on potential evapotranspiration became increasingly apparent together with increases in irrigation water withdrawal. Copyright © 2010 John Wiley & Sons, Ltd.     

Evaluating the transferability of monthly water balance models under changing climate conditions

ABSTRACT

Monthly water balance models (MWBMs) are often used for making flow projections under climate change. As such, these models should provide accurate flow simulations; however, they are seldom evaluated in this regard. This paper presents a comprehensive framework intended for the evaluation of the applicability of MWBMs under changing climatic conditions. The framework consists of analyses of consistency in model performance, parameter estimates and simulated water balance components, and a subjective assessment of model transferability. Four MWBMs – abcd, Budyko, GR2M and WASMOD – are used to simulate runoff in the Wimmera catchment affected by the Millennium drought. Although abcd and Budyko slightly outperformed GR2M and WASMOD, none of the models performed well in transfer to the driest period. The greatest variability is detected in simulated groundwater storage and baseflow; thus, these model components should be improved and/or enhanced calibration strategies should be employed to advance the transferability of MWBMs under changing climate.     

SNOW WETNESS ESTIMATES OF VEGETATED TERRAIN FROM SATELLITE PASSIVE MICROWAVE DATA

The Special Sensor Microwave/Imager (SSM/I) radiometer is a useful tool for monitoring snow wetness on a large scale because water content has a significant effect on the microwave emissions at the snowpack surface. To date, SSM/I snow wetness algorithms, based on statistical regression analysis, have been developed only for specific regions. Inadequate ground-based snow wetness measurements and the non-linearity between SSM/I brightness temperatures (T_Bs) and snow wetness over varied vegetation covered terrain has impeded the development of a general model. In this study, we used a previously developed linear relationship between snowpack surface wetness (% by volume) and concurrent air temperature (°C) to estimate the snow wetness at ground weather stations. The snow condition (snow free, dry, wet or refrozen snow) of each SSM/I pixel (a 37 × 29 km area at 37.0 GHz) was determined from ground-measured weather data and the T_B signature. SSM/I T_Bs of wet snow were then linked with the snow wetness estimates as an input/output relationship. A single-hidden-layer back-propagation (backprop) artificial neural network (ANN) was designed to learn the relationships. After training, the snow wetness values estimated by the ANN were compared with those derived by regression models. Results show that the ANN performed better than the existing regression models in estimating snow wetness from SSM/I data over terrain with different amounts of vegetation cover.     

Integrating lysimeter drainage and eddy covariance flux measurements in a groundwater recharge model

Abstract

Field-scale water balance is difficult to characterize because controls exerted by soils and vegetation are mostly inferred from local-scale measurements with relatively small support volumes. Eddy covariance flux and lysimeters have been used to infer and evaluate field-scale water balances because they have larger footprint areas than local soil moisture measurements. This study quantifies heterogeneity of soil deep drainage (D) in four 12.5-m² repacked lysimeters, compares evapotranspiration from eddy covariance (ET_EC) and mass balance residuals of lysimeters (ETwb_Lys), and models D to estimate groundwater recharge. Variation in measured D was attributed to redirection of snowmelt infiltration and differences in lysimeter hydraulic properties caused by surface soil treatment. During the growing seasons of 2010, 2011 and 2012, ETwb_Lys (278, 289 and 269 mm, respectively) was in good agreement with ET_EC (298, 301 and 335 mm). Annual recharge estimated from modelled D was 486, 624 and 613 mm for three calendar years 2010, 2011 and 2012, respectively. In summary, lysimeter D and ET_EC can be integrated to estimate and model groundwater recharge.

Editor D. Koutsoyiannis     

Quantification of cloud water interception in the canopy vegetation from fog gauge measurements

With changes in climate looming, quantifying often‐overlooked components of the canopy water budget, such as cloud water interception (CWI), is increasingly important. Commonly, CWI quantification requires detailed continuous measurements, which is extremely challenging, especially when throughfall is included. In this study, we propose a simplified approach to estimate CWI using the Rutter‐type interception model, where CWI inputs in the canopy vegetation are proportional to fog interception measured by an artificial fog gauge. The model requires the continuous acquisition of meteorological variables as input and calibration datasets. Throughfall measurements below the forest are used only for calibration and validation of the model; thus, CWI estimates can be provided even after the cessation of throughfall monitoring. This approach provides an indirect and undemanding way to quantify CWI by vegetation and allows the identification of its controlling factors, which could be useful to the comparison of CWI in contrasting land covers. The method is applied on a 2‐year dataset collected in an endemic highland forest of San Cristobal Island (Galapagos). Our results show that CWI reaches 21% ± 6% of the total water input during the first year, and 9% ± 2% during the second one. These values represent 32% ± 10% and 17% ± 5% of water inputs during the cool foggy season of the first and second year, respectively. The difference between seasons is attributed to a lower fog liquid water during the second season.     

The measurement and modelling of stemflow in an alpine Myricaria squamosa community

Stemflow is an important hydrological process of rainfall partitioning, but it has rarely been studied in the alpine riparian shrub Myricaria squamosa in the Qinghai–Tibet Plateau. This study aimed to measure and model the stemflow of the unstudied M. squamosa and to identify the key controlling factors of stemflow yield. Correlations and stepwise regression analysis between stemflow and five meteorological and ten biological factors indicated that the rainfall amount and the aboveground biomass were the best variables for modelling and predicting stemflow. We used the best model to estimate annual and stand stemflow, as well as rainfall threshold for stemflow generation. Annual stemflow accounted for 2.3 to 10.2% of the annual rainfall amount, varying with different vegetation coverage and leaf area index. The annual stemflow percentage increased linearly with the annual total rainfall amount of events > 7.3 mm. For M. squamosa stands, branches with diameters of 10 to 25 mm were less frequent but contributed much more stemflow than branches with diameters smaller than 10 mm. The stemflow percentage increased sharply with increasing rainfall amounts when the rainfall amounts were less than 4, 8 or 13 mm for the M. squamosa stands with coverage of 32.6, 47.6 or 56.1%, respectively, but increased gently when the rainfall amounts were greater than these values. The rainfall threshold for stemflow generation decreased as the branch aboveground biomass increased, and the estimated median value of the rainfall threshold was 0.8 mm for M. squamosa stands, with a range of 3.0 to 0.4 mm for branches weighing 10 to 300 g, respectively. Copyright © 2014 John Wiley & Sons, Ltd.     

The effects of scene heterogeneity on soil moisture retrieval from passive microwave data

The τ–ω model of microwave emission from soil and vegetation layers is widely used to estimate soil moisture content from passive microwave observations. Its application to prospective satellite-based observations aggregating several thousand square kilometres requires understanding of the effects of scene heterogeneity. The effects of heterogeneity in soil surface roughness, soil moisture, water area and vegetation density on the retrieval of soil moisture from simulated single- and multi-angle observing systems were tested. Uncertainty in water area proved the most serious problem for both systems, causing errors of a few percent in soil moisture retrieval. Single-angle retrieval was largely unaffected by the other factors studied here. Multiple-angle retrievals errors around one percent arose from heterogeneity in either soil roughness or soil moisture. Errors of a few percent were caused by vegetation heterogeneity. A simple extension of the model vegetation representation was shown to reduce this error substantially for scenes containing a range of vegetation types.     

Developing an empirical model of canopy water flux describing the common response of transpiration to solar radiation and VPD across five contrasting woodlands and forests

A modified Jarvis–Stewart model of canopy transpiration (E_c) was tested over five ecosystems differing in climate, soil type and species composition. The aims of this study were to investigate the model's applicability over multiple ecosystems; to determine whether the number of model parameters could be reduced by assuming that site‐specific responses of E_c to solar radiation, vapour pressure deficit and soil moisture content vary little between sites; and to examine convergence of behaviour of canopy water‐use across multiple sites. This was accomplished by the following: (i) calibrating the model for each site to determine a set of site‐specific (SS) parameters, and (ii) calibrating the model for all sites simultaneously to determine a set of combined sites (CS) parameters. The performance of both models was compared with measured E_c data and a statistical benchmark using an artificial neural network (ANN). Both the CS and SS models performed well, explaining hourly and daily variation in E_c. The SS model produced slightly better model statistics [R² = 0.75–0.91; model efficiency (ME) = 0.53–0.81; root mean square error (RMSE) = 0.0015–0.0280 mm h^‐1] than the CS model (R² = 0.68–0.87; ME = 0.45–0.72; RMSE = 0.0023–0.0164 mm h^‐1). Both were highly comparable with the ANN (R² = 0.77–0.90; ME = 0.58–0.80; RMSE = 0.0007–0.0122 mm h^‐1). These results indicate that the response of canopy water‐use to abiotic drivers displayed significant convergence across sites, but the absolute magnitude of E_c was site specific. Period totals estimated with the modified Jarvis–Stewart model provided close approximations of observed totals, demonstrating the effectiveness of this model as a tool aiding water resource management. Analysis of the measured diel patterns of water use revealed significant nocturnal transpiration (9–18% of total water use by the canopy), but no Jarvis–Stewart formulations are able to capture this because of the dependence of water‐use on solar radiation, which is zero at night. Copyright © 2012 John Wiley & Sons, Ltd.     

Appraisal of Budyko formula in calculating long‐term water balance in humid watersheds of southern China

The Budyko formula for estimating the long‐term average annual evaporation is applied to calculate the long‐term water balance in 29 humid watersheds of southern China. As a result of overestimation of evaporation, the long‐term average annual runoff is underestimated, with the Nash‐Sutcliffe efficiency (NSE) at just ? 17%. A one‐variable linear regression model is employed to find that the Budyko scatter and the relative errors of Budyko runoff and evaporation estimates are all closely related to the long‐term aridity index. Through combining the original Budyko formula with the different linear regression models for estimating the Budyko estimation errors, three forms of revised Budyko equation for estimating the long‐term average annual runoff are derived, with all their NSE values to be around 66%. After calibration, both one‐parameter Turc‐Pike and one‐parameter Fu equations lead to the NSE value of 60% in estimating long‐term average annual runoff. Two conclusions are made, with the first one being that, the nonparametric Budyko formula, although very intuitive and very simple, does not apply well in calculating long‐term water balance in 29 humid watersheds in southern China. The second one is that, the parametric evaporation formulae, with locally optimized parameter values, can achieve better accuracy in estimating long‐term average annual evaporation and runoff than the nonparametric Budyko evaporation formula. Copyright © 2011 John Wiley & Sons, Ltd.     

Comparison of transfer functions in statistical downscaling models for daily temperature and precipitation over Canada

This study compares three linear models and one non-linear model, specifically multiple linear regression (MLR) with ordinary least squares (OLS) estimates, robust regression, ridge regression, and artificial neural networks (ANNs), to identify an appropriate transfer function in statistical downscaling (SD) models for the daily maximum and minimum temperatures (T_max and T_min) and daily precipitation occurrence and amounts (P_occ and P_amount). This comparison was made over twenty-five observation sites located in five different Canadian provinces (British Columbia, Saskatchewan, Manitoba, Ontario, and Québec). Reanalysis data were employed as atmospheric predictor variables of SD models. Predictors of linear transfer functions and ANN were selected by linear correlations coefficient and mutual information, respectively. For each downscaled case, annual and monthly models were developed and analysed. The monthly MLR, annual ANN, annual ANN, and annual MLR yielded the best performance for T_max, T_min, P_occ and P_amont according to the modified Akaike information criterion (AICu). A monthly MLR is recommended for the transfer functions of the four predictands because it can provide a better performance for the T_max and as good performance as the annual MLR for the T_min, P_occ, and P_amount. Furthermore, a monthly MLR can provide a slightly better performance than an annual MLR for extreme events. An annual MLR approach is also equivalently recommended for the transfer functions of the four predictands because it showed as good a performance as monthly MLR in spite of its mathematical simplicity. Robust and ridge regressions are not recommended because the data used in this study are not greatly affected by outlier data and multicollinearity problems. An annual ANN is recommended only for the T_min, based on the best performance among the models in terms of both the RMSE and AICu.     

Prediction of Density Flow Plunging Depth in Dam Reservoirs: An Artificial Neural Network Approach

Experimental findings and observations indicate that plunging flow is related to the formation of bed load deposition in dam reservoirs. The sediment delta begins to form in the plunging region where the inflow river water meets the ambient reservoir water. Correct estimation of dam reservoir flow, plunging point, and plunging depth is crucial for dam reservoir sedimentation and water quality issues. In this study, artificial neural network (ANN), multi‐linear regression (MLR), and two‐dimensional hydrodynamic model approaches are used for modeling the plunging point and depth. A multi layer perceptron (MLP) is used as the ANN structure. A two‐dimensional model is adapted to simulate density plunging flow through a reservoir with a sloping bottom. In the model, nonlinear and unsteady continuity, momentum, energy, and k–ε turbulence equations are formulated in the Cartesian coordinates. Density flow parameters such as velocity, plunging points, and plunging depths are determined from the simulation and model results, and these are compared with previous experimental and model works. The results show that the ANN model forecasts are much closer to the experimental data than the MLR and mathematical model forecasts.     

Dynamics of seasonal recharge beneath a semiarid vegetation on the gnangara mound,Western Australia

To estimate seasonal changes in recharge to the underlying sandy aquifer, the soil water dynamics of the unsaturated zone was monitored down to a depth of 20 m over a period of three years (1985 to 1987). The measurements were made by a neutron probe at eight locations beneath a native vegetation in a semiarid region, Western Australia, receiving precipitation of 775 mm yr^?1. A relatively simple method, based on the analyses of sequentially measured soil water profiles involving utilization of zero flux plane in the unsaturated zone, is presented and used to compute seasonal recharge rates. Drainage fluxes (recharge rates) below two specified depths were estimated. These were: R₁ (water flux at a depth of 10 m, just below the maximum rooting depth) and R₂ (water flux at a depth of 18 m, just above the water table). These two estimates were significantly different both on a seasonal and annual basis, but their cumulative values for the three year period were very similar. While the annual precipitation varied from 525 to 850 mm yr^?1, the corresponding spatially averaged R₁ varied from 34 to 149 mm yr^?1, and R₂ varied from 65 to 80 mm yr^?1. A significant difference in recharge between the upslope and downslope positions on a hillslope was ascribed to differences in vegetation density of the understorey and differences in hydraulic properties of subsoils. For the three year period, the average R₁ and R₂ were 13 per cent and 10 per cent of the precipitation respectively. These values compare favourably with a long-term estimate based on an environmental tracer technique.     

Comparison of an artificial neural network and a conceptual rainfall–runoff model in the simulation of ephemeral streamflow

ABSTRACT

The rainfall–runoff process is governed by parameters that can seldom be measured directly for use with distributed models, but are rather inferred by expert judgment and calibrated against historical records. Here, a comparison is made between a conceptual model (CM) and an artificial neural network (ANN) for their ability to efficiently model complex hydrological processes. The Sacramento soil moisture accounting model (SAC-SMA) is calibrated using a scheme based on genetic algorithms and an input delay neural network (IDNN) is trained for variable delays and hidden layer neurons which are thoroughly discussed. The models are tested for 15 ephemeral catchments in Crete, Greece, using monthly rainfall, streamflow and potential evapotranspiration input. SAC-SMA performs well for most basins and acceptably for the entire sample with R² of 0.59–0.92, while scoring better for high than low flows. For the entire dataset, the IDNN improves simulation fit to R² of 0.70–0.96 and performs better for high flows while being outmatched in low flows. Results show that the ANN models can be superior to the conventional CMs, as parameter sensitivity is unclear, but CMs may be more robust in extrapolating beyond historical record limits and scenario building.

EDITOR M.C. Acreman; ASSOCIATE EDITOR not assigned