Search USGS⌕ Search

Geology topics

Andrew T. Leaf

Publications and source records attributed to Andrew T. Leaf.

27 records · Page 2Linked to original sources
USGS2019

Simulation of the regional groundwater-flow system in the St. Louis River basin, Minnesota

The St. Louis River Basin (SLRB) covers 3,600 square miles in northeastern Minnesota, with headwaters in the Mesabi Range and extensive wetlands and lakes throughout the basin. To better understand the regional groundwater system in the SLRB, a two-dimensional, steady-state groundwater-flow model of the SLRB was developed by the U.S. Geological Survey, in cooperation with the Minnesota Ojibwe Bands, using the analytic-element computer code GFLOW. The parameter-estimation software suite PEST was used to obtain a best fit of the modeled to measured groundwater levels and streamflows. The calibrated regional model was locally refined to create a smaller version of the model, the central SLRB model, that was used to evaluate hydrologic effects from extensive ditching in wetlands of the central SLRB. The refinements included adding ditches that were not represented in the regional model and modifying the aquifer base elevation to be more representative of the localized area. The central SLRB model was recalibrated to better match the distribution of mapped wetlands. Two scenarios were run of the central SLRB model: one with ditches and one without ditches. The model results were compared between the two scenarios to assess the effect of ditching on the groundwater system and potential changes to hydrologic conditions that support wetlands. Calibration of the regional SRLB model resulted in average horizontal hydraulic conductivity values of 6–39 feet per day for the glacial deposits and 3–4 feet per day for the uppermost fractured bedrock in the Biwabik Iron-Formation on the Mesabi Range. Average recharge across the calibrated model was 5.9 inches per year. Linesink resistance for the routed stream network was calibrated by using resistance categories based on the mapped soil hydrologic groups. The modeled regional groundwater-flow direction was generally to the south near the Mesabi Range topographic high and south or southwest across the rest of the basin. The updated calibration of the central SLRB model resulted in average horizontal hydraulic conductivity values of 5–36 feet per day for the glacial deposits and 3 feet per day for the uppermost fractured Biwabik Iron-Formation of the Mesabi Range. Average recharge across the ditch scenarios was 4.1 inches per year. Comparison of the preditch and postditch model scenarios showed that ditching reduced the area where the modeled water table was within 1 foot of the land surface (a wetland hydrology indicator) in as much as 40,000 acres, or 37 percent, of mapped permanent wetlands in the SLRB. An increase in the depth to the water table in wetland areas has the potential to degrade wetland persistence or function.

Minnesota↗
USGS2017

Groundwater flow model for the Little Plover River basin in Wisconsin’s Central Sands

The Little Plover River is a groundwater-fed stream in the sand plains region of central Wisconsin. In this region, sandy sediment deposited during or soon after the last glaciation forms an important unconfined sand and gravel aquifer. This aquifer supplies water for numerous high-capacity irrigation, municipal, and industrial wells that support a thriving agricultural industry. In recent years the addition of many new wells, combined with observed diminished flows in the Little Plover and other nearby rivers, has raised concerns about the impacts of the wells on groundwater levels and on water levels and flows in nearby lakes, streams, and wetlands. Diverse stakeholder groups, including well operators, Growers, environmentalists, local land owners, and regulatory and government officials have sought a better understanding of the local groundwater-surface water system and have a shared desire to balance the water needs of the he liagricultural, industrial, and urban users with the maintenance and protection of groundwater-dependent natural resources. To help address these issues, the Wisconsin Department of Natural Resources requested that the Wisconsin Geological and Natural History Survey and U.S. Geological Survey cooperatively develop a groundwater flow model that could be used to demonstrate the relationships among groundwater, surface water, and well withdrawals and also be a tool for testing and evaluating alternative water management strategies for the central sands region. Because of an abundance of previous studies, data availability, local interest, and existing regulatory constraints the model focuses on the Little Plover River watershed, but the modeling methodology developed during this study can apply to much of the larger central sands of Wisconsin. The Little Plover River groundwater flow model simulates three-dimensional groundwater movement in and around the Little Plover River basin under steady-state and transient conditions. This model explicitly includes all high-capacity wells in the model domain and simulates seasonal variations in recharge and well pumping. The model represents the Little Plover River, and other significant streams and drainage ditches in the model domain, as fully connected to the groundwater system, computes stream base flow resulting from groundwater discharge, and routes the flow along the stream channel. A separate soil-water-balance (SWB) model was used to develop groundwater recharge arrays as input for the groundwater flow model. The SWB model uses topography, soils, land use, and climatic data to estimate recharge as deep drainage from the soil zone. The SWB model explicitly includes recharge originating as irrigation water, and computes irrigation using techniques similar to those used by local irrigation operators. The groundwater flow model uses the U.S. Geological Survey’s MODFLOW modeling code which is freely available, widely accepted, and commonly used by the groundwater community. The groundwater flow model and the SWB model use identical high-resolution numerical grids having model cells 100 feet on a side, with physical properties assigned to each grid cell. This grid allows accurate geographic placement of wells, streams, and other model features. The 3-dimensional grid has three layers; layers 1 and 2 represent the sand and gravel aquifer and layer 3 represents the underlying sandstone. The distribution of material properties in the model (hydraulic conductivity, aquifer thickness, etc.) comes from previous published geologic studies of the region, updated by calibration to recent streamflow and groundwater level data. The SWB model operates on a daily time step. The groundwater flow model was calibrated to monthly stress periods with time steps ranging from 1 to 16 days. More detailed time discretization is possible. The groundwater model was calibrated to water-level and streamflow data collected during 2013 and 2014 by adjusting model parameters (primarily hydraulic conductivity, storage, and recharge) until the model produced a conditionally optimal fit between field observations and model output, subject to consistency with previously published geologic studies. Calibration was performed under both steady and transient conditions, and used a sophisticated parameter-estimation procedure (PEST) for the calibration process and to identify important model parameters. For the Little Plover River, the two most important parameters are the global recharge multiplier and the hydraulic conductivity of the stream bed. The calibrated model produces water-level and mass-balance results that are consistent with field observations and previous studies of the area. The completed model is a powerful tool for testing and demonstrating alternative water-management scenarios. Example model applications described in this report include simulating how the cumulative impacts of pumping and land-use change have affected average baseflow in the Little Plover River. Depletion-potential mapping represents a method for predicting which wells and well locations have the greatest impact on nearby surface-water resources. The completed model is publicly available, along with a companion user’s guide to assist with its operation, at http://wgnhs.org/littleplover- river-groundwater-model.

Wisconsin↗
USGS2016

Simulation of climate change effects on streamflow, groundwater, and stream temperature using GSFLOW and SNTEMP in the Black Earth Creek Watershed, Wisconsin

A groundwater/surface-water model was constructed and calibrated for the Black Earth Creek watershed in south-central Wisconsin. The model was then run to simulate scenarios representing common societal concerns in the basin, focusing on maintaining a cold-water resource in an urbanizing fringe near its upper stream reaches and minimizing downstream flooding. Although groundwater and surface water are considered a single resource, many hydrologic models simplistically simulate feedback loops between the groundwater system and other hydrologic processes. These feedbacks include timing and rates of evapotranspiration, surface runoff, soil-zone flow, and interactions with the groundwater system; however, computer models can now routinely and iteratively couple the surface-water and groundwater systems—albeit with longer model run times. In this study, preliminary calibrations of uncoupled transient surface-water and steady-state groundwater models were used to form the starting point for final calibration of one transient computer simulation that iteratively couples groundwater and surface water. The computer code GSFLOW (Groundwater/Surface-water FLOW) was used to simulate the coupled hydrologic system; a surface-water model represented hydrologic processes in the atmosphere, at land surface, and within the soil zone, and a groundwater-flow model represented the unsaturated zone, saturated zone, and streams. The coupled GSFLOW model was run on a daily time step during water years 1985–2007. Early simulation times (1985–2000) were used for spin-up to make the simulation results less sensitive to initial conditions specified; the spin-up period was not included in the model calibration. Model calibration used observed heads, streamflows, solar radiation, and snowpack measurements from 2000 to 2007 for history matching. Calibration was performed by using the PEST parameter estimation software suite. Simulated streamflows from the calibrated GSFLOW model and other basin characteristics were used as input to the one-dimensional SNTEMP (Stream-Network TEMPerature) model. SNTEMP was used to simulate daily stream temperature in selected stream reaches in the watershed. The temperature model was calibrated to high-resolution stream temperature time-series data measured in 2005. The calibrated GSFLOW and SNTEMP models were then used to simulate effects of potential climate change for the years 2010 through 2100. An ensemble of climate models and emission scenarios was evaluated. Downscaled climate drivers for the simulation period showed increases in maximum and minimum air temperature. Scenarios of future precipitation, however, did not show a monotonic trend like temperature. Uncertainty in the climate drivers increased with time for both temperature and precipitation. Forecasts of potential climate change scenarios showed growing season length increasing by weeks, and both potential and actual evapotranspiration rates increasing appreciably, in response to increasing air temperature. Simulated actual evapotranspiration rates increased less than simulated potential evapotranspiration rates as a result of water limitation in the root zone during the summer high-evapotranspiration period. The hydrologic-system response to climate change was characterized by a reduction in the importance of the snowmelt pulse and an increase in the importance of fall and winter groundwater recharge. The less dynamic hydrologic regime is likely to result in drier soil conditions, with relatively less drying expected in groundwater-fed systems. Groundwater discharge in the current upper cold-water reaches of Black Earth Creek is expected to decrease; flooding in downstream reaches may appreciably increase. The magnitude of changes in forecasted flow and associated groundwater/surface-water interaction is dependent on the General Circulation Model and emission scenario chosen. Potential future changes in air temperature drivers were consistently upward regardless of General Circulation Model and emission scenario selected; thus, simulated stream temperatures are forecast to increase appreciably with future climate. However, the amount of temperature increase was variable. Such uncertainty is reflected in temperature model results, along with uncertainty in the groundwater/surface-water interaction itself. The estimated increase in annual average temperature ranged from approximately 3 to 6 degrees Celsius by 2100 in the upper reaches of Black Earth Creek and 2 to 4 degrees Celsius in reaches farther downstream. As with all forecasts that rely on projections of an unknowable future, the results are best considered to approximate potential outcomes of climate change given the underlying uncertainty.

Wisconsin↗
USGS2015

Groundwater/surface-water interactions in the Bad River Watershed, Wisconsin

A groundwater-flow model was developed for the Bad River Watershed and surrounding area by using the U.S. Geological Survey (USGS) finite-difference code MODFLOW-NWT. The model simulates steady-state groundwater-flow and base flow in streams by using the streamflow routing (SFR) package. The objectives of this study were to: (1) develop an improved understanding of the groundwater-flow system in the Bad River Watershed at the regional scale, including the sources of water to the Bad River Band of Lake Superior Chippewa Reservation (Reservation) and groundwater/surface-water interactions; (2) provide a quantitative platform for evaluating future impacts to the watershed, which can be used as a starting point for more detailed investigations at the local scale; and (3) identify areas where more data are needed. This report describes the construction and calibration of the groundwater-flow model that was subsequently used for analyzing potential locations for the collection of additional field data, including new observations of water-table elevation for refining the conceptualization and corresponding numerical model of the hydrogeologic system. The study area can be conceptually divided into three primary hydrogeologic environments. The first encompasses the southern uplands with relatively low topographic relief, where groundwater-flow is unconfined and occurs primarily in sandy till and glacial outwash overlying Archean-aged crystalline bedrock. The second includes a transitional area of higher topographic relief and shallow depth to bedrock, in the vicinity of ridges formed by steeply dipping, early-Proterozoic aged metasedimentary units of the Marquette Range Supergroup (including the Ironwood Formation), and late-Proterozoic igneous units associated with the Midcontinent Rift System (MRS). Groundwater-flow in this area likely occurs primarily through connected networks of bedrock fractures that are not well characterized, and also in isolated pockets of Quaternary deposits. The third and last hydrogeologic environment includes lowlands along Lake Superior where a deep sandstone aquifer is confined by thick deposits of clay-rich till. Model input was compiled by using both published and unpublished data. Constant flux boundary conditions for the model perimeter were developed from a regional analytic element model described in appendix 1 of this report. Pumping from 26 high-capacity wells within the model area was included. The SFR stream network was developed from the National Hydrography Dataset (NHDPlus Version 2) and hydrography from the Wisconsin Department of Natural Resources (WDNR). Hydraulic conductivity values were determined for each model cell by interpolation from a network of pilot points, within zones representing major hydrogeologic units. Recharge to the groundwater system was estimated on a cell-by-cell basis by using the Soil Water Balance code (SWB), with gridded daily temperature and precipitation data for the period 1980–2011, and GIS coverages of soil and land-surface conditions. Estimated recharge varies considerably, following spatial patterns in the precipitation and soil hydrologic group inputs. The lowest recharge values occur in the Superior lowlands, whereas the highest values occur in the upland areas, especially those underlain by sandy soils, and in the vicinity of bedrock hills. The model was calibrated to groundwater-levels and base flows obtained from the USGS National Water Information System (NWIS) database, and groundwater-levels obtained from the WDNR and Band River Band well-construction databases. Calibration was performed via nonlinear regression by using the parameter-estimation software suite PEST. Groundwater levels and base-flow observations in the calibration dataset were well simulated by the calibrated model, with reasonable values of hydraulic conductivity. The pilot-point parameters that were most constrained by observations during model calibration coincided with the locations containing the most wells (head observations)—especially the population centers of Ashland, Mellen, and other communities along the major highway corridors. Results from the calibrated model illustrate differences in the nature of groundwater-flow within the watershed. In the southern part of the watershed, where bedrock is shallow, groundwater flow paths are relatively short, extending from local recharge areas to adjacent first and second-order streams. In contrast, laterally continuous deposits of clay-rich till covering the Superior Lowlands isolate most smaller streams from the sandstone aquifer, allowing for longer flow paths toward larger streams such as the Bad, Marengo, and White Rivers. Approximately three-quarters of all first-order stream cells were dry in the Superior Lowlands, compared to only half of first-order stream cells in the southern bedrock uplands. The model was used to delineate the groundwatershed for the Bad and Kakagon Rivers. “Groundwatershed” is defined as the area contributing groundwater discharge to one of these streams and their tributaries. The groundwatershed was found to align closely with the surface-watershed, with the most notable exception occurring along the southwestern half of Birch Hill, where surface water drains southwest towards the Potato River, and groundwater flows north and east towards Lake Superior. Similarly, the contributing area of groundwater-flow to the Reservation was delineated. Results indicate the off-Reservation groundwater contributing area to be limited in comparison to the extent of the watershed, extending southward into the highlands underlain by MRS igneous rock units, but not further into the area underlain by the Marquette Range Supergroup. Stable isotope samples were collected from 54 wells within the watershed, to investigate sources of groundwater. Oxygen-18 (δ 18 O) values lower than -13.0 per mil were documented in the sampling, and likely indicate the presence of recharge water from the last glacial period (>9,500 years old) beneath the northern portion of the Reservation, in the vicinity of Odanah, Wisconsin. Finally, a new data-worth analysis of potential new monitoring-well locations was performed by using the model. The relative worth of new measurements was evaluated based on their ability to increase confidence in model predictions of groundwater levels and base flows at 35 locations, under the condition of a proposed open-pit iron mine. Results of the new data-worth analysis, and other inputs and outputs from the Bad River model, are available through an online dynamic web mapping service at ( http://wim.usgs.gov/badriver/ ).

Wisconsin↗
USGS2015

PESTools – A Python toolkit for processing PEST-related information

PESTools is an open-source Python package for processing and visualizing information associated with the parameter estimation software PEST and PEST++. While PEST output can be reformatted for post- processing in spreadsheets or other menu-driven software packages, that approach can be error-prone and time-consuming. Managing information from highly parameterized models with thousands of parameters and observations presents additional challenges. PESTools consists of a set of Python object classes to facilitate efficient processing and visualization of PEST-related information. Processing and visualization of observation residuals, objective function contributions, parameter and observation sensitivities, parameter correlation and identifiability, and other common PEST outputs have been implemented. PESTools is integrated with the pyemu software package for linear-based computer model uncertainty analyses, allowing for efficient computations using the Jacobian Matrix without any external utilities or files. The use of dataframe objects (pandas Python package) facilitates rapid subsetting and querying of large datasets, as well as the incorporation of ancillary information such as observation locations, times, measurement types, and other associated information. PESTools’ object methods can be easily scripted with concise code, or alternatively, the use of IPython notebooks allows for live interaction with the information. PESTools is designed to streamline workflows and provide deeper insight into model behavior, enhance troubleshooting, and improve transparency in the calibration process.

Conference Paper↗