Keith Loague

STATISTICAL AND GRAPHICAL METHODS FOR EVALUATING SOLUTE TRANSPORT MODELS: OVERVIEW AND APPLICATION

Abstract Mathematical modeling is the major tool to predict the mobility and the persistence of pollutants to and within groundwater systems. Several comprehensive institutional models have been developed in recent years for this purpose. However, evaluation procedures are not well established for models of saturated-unsaturated soil-water flow and chemical transport. This paper consists of three parts: (1) an overview of various aspects of mathematical modeling focused upon solute transport models; (2) an introduction to statistical criteria and graphical displays that can be useful for model evaluation; and (3) an example of model evaluation for a mathematical model of pesticide leaching. The model testing example uses observed and predicted atrazine concentration profiles from a small catchment in Georgia. The model tested is the EPA pesticide root zone model (PRZM).

HYDROLOGIC RESPONSE OF A STEEP, UNCHANNELED VALLEY TO NATURAL AND APPLIED RAINFALL

Observations from natural rain storms and sprinkling experiments at a steep zero-order catchment in the Oregon Coast Range demonstrate the importance offlow through near-surface bedrock on runoff generation and pore pressure development in shallow colluvial soils. Sprinkling experiments, involving irrigation of the entire 860 m 2 catchment at average intensities of 1.5 and 3.0 mm/h, permitted detailed observation of runoff and the development of subsurface saturation under controlled conditions. A weir installed to collectflow through the colluvium at the base of the catchment recovered runoff equal to one third to one half of the precipitation rate during quasi-steady irrigation. Three key observations demonstrate that a significant proportion of storm runoffflows through near-surface bedrock and illustrate the importance of shallow bedrockflow in pore pressure development in the overlying colluvial soil: (1) greater discharge recovery during both the experiments and natural rainfall at a weir installed approximately 15 m downslope of the weir at the base of the catchment, (2) spatially discontinuous patterns of positive pressure head in the colluvium during steady sprinkling, and (3) local development of upward head gradients associated withflow from weathered rock into the overlying colluvium during high-intensity rainfall. Data from natural storms also show that smaller storms produce no significant runoff or piezometric response and point to a critical intensity-duration rainfall to overcome vadose zone storage. Together these observations highlight the role of interaction betweenflow in colluvium and near- surface bedrock in governing patterns of soil saturation, runoff production, and positive pore pressures.

Hydrologic‐Response simulations for the R‐5 catchment with a comprehensive physics‐based model

For approximately 20 years, there has been a concerted effort, by several different research groups, to simulate observed rainfall-runoff events from the well-known R-5 catchment, located near Chickasha, Oklahoma. These prior simulation efforts, with relatively simple models of Horton-type overland flow, have not been entirely successful, as the streamflow generation process for the R-5 catchment, as now recognized, may not be totally dominated by the Horton mechanism. In the effort reported here, a new fully coupled comprehensive physics-based hydrologic-response model, the Integrated Hydrology Model (InHM), is tested for two R-5 rainfall-runoff events. The InHM simulations in this study clearly show, in a hypothesis-testing mode, that both the Horton and Dunne overland flow mechanisms can be important streamflow generation processes for R-5 events. The InHM simulations reported here also suggest that accurate accounting of soil water storage can be as important as exhaustive characterization of spatial variations in near-surface permeability.

Unsaturated zone processes and the hydrologic response of a steep, unchanneled catchment

As part of a larger, collaborative study, we conducted field experiments to investigate how rainfall signals propagate through an unsaturated soil profile, leading to a rapid pore pressure response and slope instability. We sprinkler-irrigated an entire, unchanneled headwater basin in the steep, humid Oregon Coast Range, and we drove the system to quasi steady state as indicated by tensiometers, piezometers, and discharge. During initial wetting some of the deeper tensiometers responded before the arrival of an advancing head gradient front. With continued irrigation most tensiometers attained near- zero pressure heads before most piezometers responded fully, and a stable unsaturated flow field preceded the development of a stable saturated flow field. Steady discharge occurred after the last piezometer reached steady state. With the onset of steady discharge the unsaturated zone, saturated zone, and discharge became delicately linked, and a spike increase in rain intensity led to a response in the saturated zone and discharge much faster than could have happened through advection alone. We propose that the rain spike produced a slight pressure wave that traveled relatively rapidly through the unsaturated zone, where it caused a large change in hydraulic conductivity and the rapid effusion of stored soil water. An important control on the hydrologic response of this catchment lies with the soil-water retention curve. In general, below pressure heads of about 20.05 m, soil-water contents change slightly with changes in pressure head, but above 20.05 m the soil-water content is highly variable. Minor rainstorms upon a wet soil can produce slight changes in pressure head and corresponding large changes in soil-water content, giving rise to the passage of pressure waves in response to increased rain intensity and a relatively rapid response in the unsaturated zone. This rapid unsaturated zone response led to a rapid rise in the saturated zone, and it may be the underlying mechanism enabling short bursts of rain to cause slope instability.

Subsurface flow paths in a steep, unchanneled catchment

Tracer studies during catchment-scale sprinkler experiments illuminate the pathways of subsurface flow in a small, steep catchment in the Oregon Coast Range. Bromide point injections into saturated materials showed rapid flow in bedrock to the catchment outlet. Bedrock flow returned to the colluvium, sustaining shallow subsurface flow there. The bromide peak velocity of ;10 23 ms 21 exceeded the saturated hydraulic conductivity of intact bedrock. This, and the peak shapes, verify that fractures provide important avenues for saturated flow in the catchment. Deuterium added to the sprinkler water moved through the vadose zone as plug flow controlled by rainfall rate and water content. Ninety-two percent of the labeled water remained in the vadose zone after 3 days (;140 mm) of sprinkling. Preferential flow of new water was not observed during either low-intensity irrigation or natural storms; however, labeled preevent water was mobile in shallow colluvium during a storm following our spiking experiment. In response to rainfall, waters from the deeper bedrock pathway, which have traveled through the catchment, exfiltrate into the colluvium mantle and mix with relatively young vadose zone water, derived locally, creating an area of subsurface saturation near the channel head. This effectively becomes a subsurface variable source area, which, depending on its size and the delivery of water from the vadose zone, dictates the apportioning of old and new water in the runoff and, correspondingly, the runoff chemistry. The slow movement of water through the vadose zone allows for chemical modification and limits the amount of new water in the runoff. Moreover, it suggests that travel time of new rain water does not control the timing of runoff generation.

Soil water content at R-5. Part 1. Spatial and temporal variability

Abstract This paper, the first part in a two-part series, is concerned with the interpretation of spatial and temporal variations in soil water content across a small rangeland catchment; two data sets are examined. The first data set is comprised of 25 728 soil water content measurements made at 34 sites over an 8 year period. The second data set consists of individual soil water content measurements made at 247 sites over a 6 day period. Geostatistical methods are used to describe variations in soil water content; general characterizations are made. In the companion paper the impact of antecedent soil water conditions is investigated for a suite of R-5 event simulations with a quasi-physically based rainfall-runoff model.

Near-surface hydrologic response for a steep, unchanneled catchment near Coos Bay, Oregon: 2. Physics-based simulations

The comprehensive physics-based hydrologic-response model InHM was used to simulate 3D variably-saturated flow and solute transport for three controlled sprinkling experiments at the Coos Bay 1 (CB1) experimental catchment in the Oregon Coast Range. The InHM-simulated hydrologic-response was evaluated against observed discharge, pressure head, total head, soil-water content, and deuterium concentration records. Runoff generation, tensiometric/piezometric response in the soil, pore-water pressure generation, and solute (tracer) transport were all simulated well, based on statistical and graphical model performance evaluation. The InHM simulations reported herein indicate that the 3D geometry and hydraulic characteristics of the layered geologic interfaces at CB1 can control the development of saturation and pore-water pressures at the soil-saprolite interface. The weathered bedrock piezometric response and runoff contribution were not simulated well with InHM in this study, most likely as a result of the uncertainty in the weathered bedrock layer geometry and fractured-rock hydraulic properties that preclude accurate fracture flow representation. Sensitivity analyses for the CB1 boundary-value problem indicate that: (i) hysteretic unsaturated flow in the CB1 soil is important for accurate hydrologic-response simulation, (ii) using an impermeable boundary condition to represent layered geologic interfaces leads to large errors in simulated magnitudes of runoff generation and pore-water pressure development, and (iii) field-based retention curve measurements can dramatically improve variably-saturated hydrologic-response simulation at sites with steep soil-water retention curves. The near-surface CB1 simulations reported herein demonstrate that physics-based models like InHM are useful for characterizing detailed spatio-temporal hydrologic-response, developing process-based concepts, and identifying information shortfalls for the next generation of field experiments. The field-based observations and hydrologic-response simulations from CB1 highlight the challenges in characterizing/simulating fractured bedrock flow at small catchments, which has important consequences for hydrologic response and landslide initiation.

Uncertainty in a pesticide leaching assessment for Hawaii

Abstract In this paper we report the predictive uncertainty associated with using the retardation factor ( RF ) as an index to evaluate pesticide leaching in Hawaii soils when uncertainty exists within the soil and chemical data used to excite the RF model. Our analysis takes two separate but ultimately related tacks. First, we assess the uncertainty caused by extrapolating soil properties between taxonomic categories. Second, we characterize the amount of uncertainty for calculated RF values due to data uncertainty by using first-order uncertainty analysis. Our results indicate that the RF index should only be used with soil information from the lowest taxonomic category and that even then considerable uncertainty will exist in the predicted RF values used to screen and rank chemicals.

Impact of uncertainty in soil, climatic, and chemical information in a pesticide leaching assessment.

A simple mobility index, when combined with a geographic information system, can be used to generate rating maps which indicate qualitatively the potential for various organic chemicals to leach to groundwater. In this paper we investigate the magnitude of uncertainty associated with pesticide mobility estimates as a result of data uncertainties. Our example is for the Pearl Harbor Basin, Oahu, Hawaii. The two pesticides included in our analysis are atrazine (2-chloro-4-ethylamino-6-isopropylamino-s-triazine) and diuron [3-(3,4-dichlorophenyul)-1,1-dimethylarea]. The mobility index used here is known as the Attenuation Factor (AF); it requires soil, hydrogeologic, climatic and chemical information as input data. We employ first-order uncertainty analysis to characterize the uncertainty in estimates of AF resulting from uncertainties in the various input data. Soils in the Pearl Harbor Basin are delineated at the order taxonomic category for this study. Our results show that there can be a significant amount of uncertainty in estimates of pesticide mobility for the Pearl Harbor Basin. This information needs to be considered if future decisions concerning chemical regulation are to be based on estimates of pesticide mobility determined from simple indices.

Spatial and temporal variability in the R-5 infiltration data set: Déjà vu and rainfall-runoff simulations

This paper is a continuation of the event-based rainfall-runoff model evaluation study reported by Loague and Freeze [1985[. Here we reevaluate the performance of a quasi-physically based rainfall-runoff model for three large events from the well-known R-5 catchment. Five different statistical criteria are used to quantitatively judge model performance. Temporal variability in the large R-5 infiltration data set [Loague and Gander, 1990] is filtered by working in terms of permeability. The transformed data set is reanalyzed via geostatistical methods to model the spatial distribution of permeability across the R-5 catchment. We present new estimates of the spatial distribution of infiltration that are in turn used in our rainfall-runoff simulations with the Horton rainfall-runoff model. The new rainfall-runoff simulations, complicated by reinfiltration impacts at the smaller scales of characterization, indicate that the near-surface hydrologic response of the R-5 catchment is most probably dominated by a combination of the Horton and Dunne overland flow mechanisms.