H12E-01 INVITED
Conceptualizing, testing, and transferring watershed characteristics-runoff generation relationships
Despite ongoing efforts to understand watershed hydrology through qualitative and quantitative measures and analyses, the relationships between watershed characteristics and streamflow response are still poorly understood. Part of the difficulty lies in ascertaining which watershed characteristics are most important for streamflow generation from one watershed to the next. This difficulty is exacerbated by the temporal variability in the hydrologic-state of a watershed: it can depend on stormflow versus baseflow, dry versus wet antecedent conditions, annual versus event based analyses, and storm characteristics. Despite these difficulties, our task remains: to gain new insight in experimental watersheds for transfer to new gauged and ungauged watersheds. This presentation addresses three interrelated aspects of watershed characterization: 1) conceptualizing and testing upland-stream connectivity over space and time (where and when stormflow is generated along the stream network), 2) assessing spatial and temporal controls on stream baseflow generation and (where and why surface water gains and losses occur at the network scale), and 3) assessing and conceptualizing dominant response characteristics of landscape elements from headwaters to the meso-scale. We emphasize new insight gained in highly instrumented experimental watersheds, new tracer-based techniques for assessing streamflow gains and losses at the stream network scale, and the use of ALSM (airborne laser swath mapping) high resolution topography data for extending hydrological process understanding to the basin scale. Understanding spatially and temporally dynamic hydrological processes and their distribution is a necessary step toward improved model representation of complex systems.
H12E-02
Event to Multi-decadal Persistence in Spatial Rainfall and Rainfall and Runoff as a Function of Spatial and Temporal Scales
Spatial and temporal rainfall variability over watershed scales impacts the hydrologic response which in turn affects runoff, agricultural production, and soil water availability. Changes in the precipitation regime over decades may contribute to changes in vegetation, water supply and, over longer time scales, landscape evolution and geomorphology. This is especially important in the southwestern US, where urbanization is increasing pressure on limited water supplies. Daily, seasonal and annual precipitation volumes and intensities from the densely gaged network of raingages on the USDA-ARS Walnut Gulch Experimental Watershed (WGEW) are evaluated for multi-decadal trends in amount and intensity over a range of watershed scales (1.5 ha to 150 km2). The influence of teleconnections is tested for local-scale rainfall variability. Rainfall and runoff volume and rate variability are compared over the same spatial scales and over a 40 year period of high-quality runoff observations. The major findings of this study are that spatial variability of precipitation decreases exponentially with time scale; that long-term precipitation trends can be captured with a low spatial resolution; and that runoff from the 150 km2 watershed is not significantly affected by factors other than precipitation. High-resolution modeling studies on the WGEW are also discussed to further understand the processes and scales involved in hydrologic watershed response. http://www.tucson.ars.ag.gov
H12E-03
On the Link Between Streamflow Transit Time and Hydrograph Recession
In catchment hydrology, new interest has recently focused on estimation of transit time of water as a diagnostic characteristic of a watershed. Regarded as a fundamental control on water chemistry and a key descriptor of storage, estimates of transit time also offer an additional form of data with which to test watershed models. However, as identified in a recent literature review, many challenges in the estimation of a mean transit time (MTT) as well as its statistical distribution remain. While recent papers have addressed the assumptions implicit in the mathematical lumped parameter flow models for tracers like δ18O, few studies have tried to estimate MTT using other more easily available methods. As a result, measurement of MTT remains difficult with today's technologies. In this paper, we further test a method of baseflow hydrograph recession analysis to estimate MTT for the well characterized H.J. Andrews watersheds, a Long Term Ecological Research (LTER) station located in Oregon's Cascade Mountains, USA. The six H.J. Andrews watersheds range in size from 0.102 to 62.4 km2. We compared the simplified baseflow recession-based method and its assumptions against the stable isotope δ18O tracer-based convolution model previously applied to these catchments. The baseflow recession method estimates MTT values ranging from 1.6 to 2.4 years, within the evaluated uncertainty of the convolution models for five of the six watersheds. The analysis presented in this briefing provides encouraging evidence that the less costly baseflow hydrograph recession analysis may be potentially as useful in predicting this fundamental diagnostic of hydrologic behavior for some catchments.
H12E-04
Combining Pedotransfer Functions and Detailed Geomorphic Mapping to Characterize Runoff Potential on an Arid Alluvial Fan Complex
A common practice for predicting surface runoff from hydrographic basins is to assign curve numbers (CN), which relate to hydrologic soil groups, land use, and vegetative cover, and that can be used in rainfall-runoff models to generate hydrographs. The purposes of this study were to characterize the soils of a remote basin in southern Nevada, specifically runoff potential and soil hydraulic conductivity (Ks), and to examine the use of a site- specific pedotransfer function (PTF) that would relate soil texture and bulk density to CNs. Geomorphic mapping, soil sampling and analysis (n=79), rainfall simulation (N=19), and tension infiltrometer (n=47) tests were all used as characterization techniques on the five distinct geomorphic surfaces identified on this 19.5 km2 site. The field- measured CNs for the older Qf5 and Qf6 surfaces (87 and 83, respectively) were close to default CN values for similar surfaces classified as hydrologic soil group C. The higher CN values obtained for the Qf4 and QTt surfaces (92.5 and 94, respectively) reflect the well-developed desert pavements covering these surfaces. The field values of Ks from the infiltrometer were higher on surfaces younger in age or otherwise eroded or incised, and lower on desert pavement surfaces and old terrace deposits. Higher CN values correspond to lower Ks. The use of detailed geomorphic mapping significantly reduced the variance in Ks across the watershed resulting in statistically distinct hydrologic groups, which could be scaled to the watershed. When average Ks was regressed onto field-measured CNs, a linear relationship was found with R2=0.928. The results show that measurement of Ks may be used to directly estimate CN in this watershed. The site-specific PTF method was also tested using the entire database of 79 soil texture and bulk density data, in combination with field measured Ks and a multiple linear regression approach. The results showed that this simple PTF approach, using only soil texture and bulk density, provided an excellent estimate of field Ks (R2=0.890). This finding means that, at this field site, an easy way to characterize the hydrographic basin would be to analyze soil samples collected from across the site, and use the site-specific relationship obtained from the field work to estimate CN, or any other hydraulic properties. Though other relationships may exist for different watersheds, the method appears to be robust for this site, making rapid assessments of flood potential possible.
H12E-05
Model-Data Integration to Provide a Probabilistic Assessment of the Role of Snowmelt in Runoff Production in the Pacific Northwest
We utilize an integrated combination of spatially distributed hydrologic modeling with remotely-sensed and ground based measurements to evaluate the contribution of snowmelt to runoff over a range of flow conditions and geographic locations in the Pacific Northwest. The contribution of snowmelt to runoff varies widely across the region depending on geographic location, elevation, land cover, and local meteorological conditions. Rain-on- snow events are relatively common within the snow transition zone and have contributed to a number of large floods. Snotel sites provide information on precipitation and changes in snow water equivalent at higher elevations, however, ground-based observation are generally lacking in the transition zone. We utilize an approach that allows the hydrologic model to be updated with remotely sensed spatial snow properties and measured streamflow using an Ensemble Kalman-based data assimilation strategy that accounts for uncertainty in meteorology, model parameters, and the observations used for updating. Although designed for ensemble streamflow forecasting, in this application the model provides a process based method to integrate multiple data sets across various spatial and temporal scales, allowing a probabilistic assessment of the role of snowmelt in runoff production.
H12E-06
Diagnostic Evaluation of the abcd Monthly Water Balance Model for the Conterminous United States
Watershed classification systems should provide sufficient information to select proper conceptual and numerical formulations in the modeling of hydrological processes. This ideal situation is limited by our ability to sample the fluxes and characteristics of the watersheds, and the lack of theoretical frameworks that consider watershed heterogeneities across different locations and scales. This work uses several analytic tools to study relationships between watershed characteristics and dominant hydrologic processes at the monthly level by diagnosing the reasons for different levels of performance of the abcd Monthly Water Balance Model at more than 700 watersheds within the conterminous United States using data from the Hydro-Climatic Data Network (HCDN) dataset. To facilitate analysis of the data, cluster analysis and Artificial Neural Networks (ANN) were used to identify groups of watersheds with similar input/output relationship and streamflow patterns. These watershed groups were used to examine the discriminatory power of a number of physical variables, indices and signatures of hydrologic behavior with regard to their dominant processes and the performance of the abcd Model and other benchmark models of reduced complexity. The diagnostic analysis resulted in a description of the problems encountered in modeling the water balance for each watershed group, and suggests a road map for future model and data improvements.
H12E-07
Reduction of Uncertainty in Water Mass Balances
Two novel approaches that reduce uncertainty in lake, watershed, and basin water balances are presented and applied in the Lake Tahoe basin. A novel residual redistribution technique reduces random error in water balance component estimates. This technique is well-grounded in standard statistical methods, and is simple, robust, and of broad general applicability. Reduction of random error in areal precipitation and streamflow estimates is validated using independent data. Remaining random error variance in areal precipitation estimates is markedly small. Reduction of random error in annual areal precipitation estimates resolves watershed ‘memory' of precipitation from prior water-years (WY). The signal of precipitation from prior WY is often obscured in random error noise associated with established methods for estimating inter-annual variations in total annual areal precipitation. It is shown that the relationship of eastern Tahoe sub-basin annual streamflow to precipitation from prior WY can be inferred in the absence of gage data, using noise-filtered precipitation data and whole basin water yield data. Limited stream gage records from eastern Tahoe sub-basins confirm the inferred dependence on precipitation from prior WY, and thus that watershed moisture storage changes are significant to the water mass balance over time scales of several years. Such long time scales for storage change effects on streamflow are typically not accurately accounted for in watershed hydrology models. Inter-annual changes in watershed moisture storage are readily distinguishable from inter-annual variations in watershed ET. Application of a novel precipitation-decorrelation technique yields an estimate of Lake Tahoe mean annual evaporation with associated rigorously quantified uncertainty. This novel estimate agrees closely with several independent standard measurement-based evaporation estimates; and has uncertainty comparable to that of a high-quality energy balance approach. The two novel techniques used together yield robust lower bounds on the magnitudes of mean annual areal precipitation and atmospheric loss (ET, sublimation) in the Tahoe Basin. These lower bounds indicate that several previously published investigations have likely underestimated actual areal precipitation and atmospheric loss.