Adrian P. Butler

Modeling irrigation behavior in groundwater systems

Integrated hydro-economic models have been widely applied to water management problems in regions of intensive groundwater-fed irrigation. However, policy interpretations may be limited as most existing models do not explicitly consider two important aspects of observed irrigation decision making, namely the limits on instantaneous irrigation rates imposed by well yield and the intraseasonal structure of irrigation planning. We develop a new modeling approach for determining irrigation demand that is based on observed farmer behavior and captures the impacts on production and water use of both well yield and climate. Through a case study of irrigated corn production in the Texas High Plains region of the United States we predict optimal irrigation strategies under variable levels of groundwater supply, and assess the limits of existing models for predicting land and groundwater use decisions by farmers. Our results show that irrigation behavior exhibits complex nonlinear responses to changes in groundwater availability. Declining well yields induce large reductions in the optimal size of irrigated area and irrigation use as constraints on instantaneous application rates limit the ability to maintain sufficient soil moisture to avoid negative impacts on crop yield. We demonstrate that this important behavioral response to limited groundwater availability is not captured by existing modeling approaches, which therefore may be unreliable predictors of irrigation demand, agricultural profitability, and resilience to climate change and aquifer depletion.

A multivariate spatial-temporal model of rainfall in southwest Saudi Arabia. I. Spatial rainfall characteristics and model formulation

Abstract Hourly data from a network of 100 autographic raingauges have been collected as part of an extensive hydrological investigation of five wadis in southwest Saudi Arabia. The general spatial and temporal characteristics of these data are presented. Seasonality of catchment rainday occurrence is illustrated and gauge rainday occurrence shown to be strongly associated with elevation. Most point rainfalls are of 1–2 h duration, initiated in mid to late afternoon. Spatial distribution is highly localised so that for the minimum intergauge distances of 8–20 km, probability of joint occurrence and cross-correlation of hourly rainfall are small. On the basis of these data, a stochastic model of the spatial distribution of hourly rainfall has been formulated. Catchment and gauge rainday occurrence are well reproduced, as are point rainfall depths and durations. Spatial performance is reasonable for daily rainfall. However, the weak spatial dependence in the data leads to some clustering of occurrence at the hourly level, which the model is unable to reproduce.

Recovering tracer test input functions from fluid electrical conductivity logging in fractured porous rocks

A radially convergent tracer test was carried out in an unconfined Chalk aquifer of Berkshire, United Kingdom. Fluorescent tracers were injected into two boreholes lying 32 m (PL10A) and 54 m (PL10B) from the abstraction hole. The tracers were also mixed with an NaCl solution so that vertical distributions of tracer within the injection wells could be monitored using fluid electrical conductivity (FEC) logging. The breakthrough curve (BTC) from PL10A was unimodal and had a first arrival time of 14 min. The BTC from PL10B exhibited two distinct peaks and a first arrival time of just 4 min. The tracer test input functions were derived by numerically modeling the observed FEC logs of the injection wells. These were then convoluted with a conventional, Fickian matrix diffusion dual-porosity model. The results suggested that the multiple peaks were due to the way in which the tracers left the injection wells and migrated into the aquifer. FEC log inversion proved to be an effective method for predicting borehole flow data obtained by flowmeters and recovering tracer test input functions for radially convergent tracer tests.

Recent advances in modelling nitrate transport in the Chalk unsaturated zone

The Chalk unsaturated zone is crucial in controlling the delivery of nitrate to Chalk streams and groundwater abstraction wells. In this paper, results from a dual-permeability numerical model of the Chalk unsaturated zone are used to illustrate the relative roles of matrix and fracture flow. A major challenge arises in representing the Chalk unsaturated zone within catchment-scale models for nutrient management. These have generally been based on simple conceptual reservoirs or compartments to represent soils and groundwater. A more appropriate conceptualization has recently been developed and applied to the Lambourn catchment within a catchment-scale nutrient model (INCA-Chalk). Preliminary results from this work are discussed, which clearly illustrate the decadal time scales that need to be considered in the context of nutrient management and the Water Framework Directive.

Measurements of spontaneous potential in chalk with application to aquifer characterization in the southern UK

We report the first measured values of the streaming potential coupling coefficient in chalk samples saturated with natural groundwater, and preliminary field measurements of the spontaneous potential (SP), at both ambient and pumped conditions, at a test site in the Berkshire Chalk aquifer in the southern UK. The ultimate aim of the work is to use measurements of SP, in conjunction with borehole data, to characterize groundwater flow and aquifer properties. Laboratory measurements yield a value of the streaming potential coupling coefficient of −60 ± 4 mV MPa−1 and a corresponding zeta potential of −13 ± 1 mV. A negative zeta potential contrasts with previous published open-system measurements on artificial calcite, and may reflect the presence of organic material in the natural chalk samples or HCO3 and SO4 ions in the groundwater. Field measurements at ambient conditions show temporal variations in SP consistent with flow processes within the aquifer, but no coherent spatial variations. Measurements during water abstraction demonstrate that voltages at the ground surface and in monitoring boreholes become more positive during pressure drawdown and more negative during pressure build-up, consistent with the negative values of streaming potential coupling coefficient and zeta potential observed in the laboratory. Moreover, the magnitude of the change in voltage is similar to that estimated using the laboratory value of the coupling coefficient. Our results suggest that measurements of SP may make a valuable contribution to characterizing groundwater flow in the UK Chalk aquifer.

A multivariate spatial-temporal model of rainfall in southwest Saudi Arabia. II. Regional analysis and long-term performance

Abstract In the preceding paper a stochastic multivariate model of the spatial distribution of hourly rainfall, developed for five basins in southwest Saudi Arabia, is described and tested with respect to 2 years of detailed spatial data. In this paper, long-term performance is evaluated in the context of regional analysis, as, in general, long-term gauges lie outside the study basins. The calibration data are shown to have lower than average rainday occurrence and the model is empirically adjusted. Extreme value characteristics of hourly rainfall are reasonably reproduced in the regional context. Some under-estimation of daily rainfall is shown to be associated with a small but distinct population of very high occurrences.

Effects of model complexity on uncertainty estimates

Abstract In the Model Complexity working group of BIOMOVS II, models of varying complexity have been applied to a theoretical problem concerning downward transport of radionuclides in soils. The purpose was to study how uncertainty in model predictions varies with model complexity and how model simplifications can suitably be made. A scenario describing a case of surface contamination of a pasture soil was defined. Three different radionuclides with different environmental behavior and radioactive half-lives were considered: 137 Cs, 90 Sr and 129 I. A detailed specification of the parameters required by different kinds of models was given, together with reasonable values for the parameter uncertainty. A total of seven modelling teams participated in the study using 13 different models. Four of the modelling groups performed uncertainty calculations using nine different modelling approaches. The models ranged in complexity from analytical solutions of a 2-box model using annual average data to numerical models coupling hydrology and transport using data varying on a daily basis.

Advances in modelling groundwater behaviour in Chalk catchments

Abstract Groundwater in Chalk catchments is a major resource that also helps support internationally important habitats and ecosystems. Its dual porosity and dual permeability properties, coupled with large-scale structural features (such as hard rock layers and marls), produce a highly complex hydrogeological system. Recent impacts from groundwater flooding as well as vulnerability to drought have raised questions over the ability of traditional approaches to model these aquifers. Current work on near-surface hydrological processes has highlighted the importance of the soil and weathered zone for controlling recharge rates. In addition, karst-like features, sedimentary deposits and valley bottom processes govern stream–aquifer interaction and present a challenge in their representation in any modelling system. Methods that have, and are being, developed to incorporate these features, and their use in modelling Chalk catchments, are described. These are required in order to address major challenges, such as groundwater flooding and drought impacts, both of which could become more frequent and intense as a result of climate change.

A parameter sensitivity analysis of two Chalk tracer tests

Abstract As with most fractured rock formations, Chalk is highly heterogeneous. Therefore, meaningful estimates of model parameters must be obtained at a scale comparable with the process of concern. These are frequently obtained by calibrating an appropriate model to observed concentration–time data from radially convergent tracer tests. Arguably, an appropriate model should consider radially convergent dispersion and Fickian matrix diffusion. Such a model requires the estimation of at least four parameters. A question arises as to whether this level of model complexity is supported by the information contained within the calibration data. A four-parameter model was developed for the analysis of two Chalk aquifer radially convergent tracer tests. The parameters included an advective travel time, ta, a characteristic fracture diffusion time, tcf, a characteristic matrix block diffusion time, tcb and a Peclet number, P. Because the tracer test duration was less than 500 h, tcb was impossible to identify. Further analysis showed that a large set of correlated values of P, ta and tcf would lead to equally good model fits. To resolve this ambiguity, more and better quality data are needed at the very start of the breakthrough curve, to constrain the mechanical dispersion parameter, P.

Medical Hydrogeology of Asian Deltas: Status of Groundwater Toxicants and Nutrients, and Implications for Human Health

Drinking water, a fluid primarily for human hydration, is also a source of mineral nutrients. Groundwater, a drinking water source for more than 70% of inhabitants living in Asian deltas, has received much attention because of its naturally occurring arsenic, but the linkage of arsenic toxicity with other water constituents has not been studied. In addition, although nutrients are generally provided by food, in under developed rural settings, where people subsist on low nutrient diets, drinking-water-nutrients may supply quantities vital to human health thereby preventing diseases. Here, we show, using augmented datasets from three Asian deltas (Bengal, Mekong, and Red River), that the chemical content of groundwater is such that in some areas individuals obtain up to 50% or more of the recommended daily intake (RDI) of some nutrients (e.g., calcium, magnesium, iron) from just two litres of drinking water. We also show some indications of a spatial association of groundwater nutrients and health outcome using demographic health data from Bangladesh. We therefore suggest that an understanding of the association of non-communicable disease and poor nutrition cannot be developed, particularly in areas with high levels of dissolved solids in water sources, without considering the contribution of drinking water to nutrient and mineral supply.