SEARCH · USGS Science

Results for “Environmental Modeling & Assessment”

Search indexed USGS publications on groundwater, aquifers, geologic maps, mineral resources and earthquakes. Explore source records by subject and place.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 721 records · Page 40Linked to original sources

USGS2017

Demographic modelling reveals a history of divergence with gene flow for a glacially tied stonefly in a changing post-Pleistocene landscape

Aim Climate warming is causing extensive loss of glaciers in mountainous regions, yet our understanding of how glacial recession influences evolutionary processes and genetic diversity is limited. Linking genetic structure with the influences shaping it can improve understanding of how species respond to environmental change. Here, we used genome-scale data and demographic modelling to resolve the evolutionary history of Lednia tumana , a rare, aquatic insect endemic to alpine streams. We also employed a range of widely used data filtering approaches to quantify how they influenced population structure results. Location Alpine streams in the Rocky Mountains of Glacier National Park, Montana, USA. Taxon Lednia tumana , a stonefly (Order Plecoptera) in the family Nemouridae. Methods We generated single nucleotide polymorphism data through restriction-site associated DNA sequencing to assess contemporary patterns of genetic structure for 11 L. tumana populations. Using identified clusters, we assessed demographic history through model selection and parameter estimation in a coalescent framework. During population structure analyses, we filtered our data to assess the influence of singletons, missing data and total number of markers on results. Results Contemporary patterns of population structure indicate that L. tumana exhibits a pattern of isolation-by-distance among populations within three genetic clusters that align with geography. Mean pairwise genetic differentiation ( F ST ) among populations was 0.033. Coalescent-based demographic modelling supported divergence with gene flow among genetic clusters since the end of the Pleistocene (~13-17 kya), likely reflecting the south-to-north recession of ice sheets that accumulated during the Wisconsin glaciation. Main conclusions We identified a link between glacial retreat, evolutionary history and patterns of genetic diversity for a range-restricted stonefly imperiled by climate change. This finding included a history of divergence with gene flow, an unexpected conclusion for a mountaintop species. Beyond L. tumana , this study demonstrates the complexity of assessing genetic structure for weakly differentiated species, shows the degree to which rare alleles and missing data may influence results, and highlights the usefulness of genome-scale data to extend population genetic inquiry in non-model species.

Montana↗

USGS2012

An economic value of remote-sensing information—Application to agricultural production and maintaining groundwater quality

Does remote-sensing information provide economic benefits to society, and can a value be assigned to those benefits? Can resource management and policy decisions be better informed by coupling past and present Earth observations with groundwater nitrate measurements? Using an integrated assessment approach, the U.S. Geological Survey (USGS) applied an established conceptual framework to answer these questions, as well as to estimate the value of information (VOI) for remote-sensing imagery. The approach uses moderate-resolution land-imagery (MRLI) data from the Landsat and Advanced Wide Field Sensor satellites that has been classified by the National Agricultural Statistics Service into the Cropland Data Layer (CDL). Within the constraint of the U.S. Environmental Protection Agency's public health threshold for potable groundwater resources, the USGS modeled the relation between a population of the CDL's land uses and dynamic nitrate (NO3-) contamination of aquifers in a case study region in northeastern Iowa. Employing various multiscaled, multitemporal geospatial datasets with MRLI to maximize the value of agricultural production, the approach develops and uses multiple environmental science models to address dynamic nitrogen loading and transport at specified distances from specific sites (wells) and at landscape scales (for example, across 35 counties and two aquifers). In addition to the ecosystem service of potable groundwater, this effort focuses on the use of MRLI for the management of the major land uses in the study region-the production of corn and soybeans, which can impact groundwater quality. Derived methods and results include (1) economic and dynamic nitrate-pollution models, (2) probabilities of the survival of groundwater, and (3) a VOI for remote sensing. For the northeastern Iowa study region, the marginal benefit of the MRLI VOI (in 2010 dollars) is $858 million ±$197 million annualized, which corresponds to a net present value of $38.1 billion ±$8.8 billion for that flow of benefits in perpetuity. Given that these economic estimates are derived from one case study in a part of only one State, the estimates provide a lower estimate related to the potential value of the Landsat Data Continuity Mission.

Professional Paper↗

USGS1996

A spatial decision support system for coastal management: A research project at the National Wetlands Research Center of the U.S. Geological Survey

Environmental resource managers and scientists are being challenged in developing strategies to manage complex coastal systems. From an ecological perspective, there are myriad dynamic, interrelated natural and human-induced processes that affect the health and stability of coastal systems. However, the problems associated with managing coastal resources usually transcend purely ecological factors when one considers societal needs and expectations from these resources. For example, at least nine Federal, State, and local government agencies, often with widely varying responsibilities or interests, are charged with managing environmental resources and/or regulating human activities within Louisiana's coastal systems; that number may be higher in other coastal areas of the United States. In many coastal systems, a declining resource base and environmental quality combined with an expanding human population exert increased demands on those systems. This results in a number of conflicting resource management and environmental impact assessment issues. The issues include determining the most cost-effective strategies for restoring degraded natural systems, local and regional planning for future urban and commercial development in or near sensitive coastal habitats, predicting impacts from acute and chronic pollutant discharges (oil spills, fecal coliform contamination) as well as from natural hazard damages to coastal systems, and optimal partitioning of coastal resources among competing user groups (e.g., commercial and recreational fishermen). Solving such complex environmental resource management problems often involves a multidisciplinary approach, requires computerized analytical modeling abilities to manipulate large quantities of spatial-temporal data according to a defined set of objectives or constraints, and needs a mechanism to provide quick responses for dynamic resource and environmental issues. For this, a GIS-based multifunctional Spatial Decision Support System (SDSS) is being developed at the National Wetlands Research Center of the U.S. Geological Survey (formerly the Southern Science Center of the US. Fish and Wildlife Service/National Biological Service).

Louisiana↗

USGS2015

Modelling regional land change scenarios to assess land abandonment and reforestation dynamics in the Pyrenees (France)

Over the last decades and centuries, European mountain landscapes have experienced substantial transformations. Natural and anthropogenic LULC changes (land use and land cover changes), especially agro-pastoral activities, have directly influenced the spatial organization and composition of European mountain landscapes. For the past sixty years, natural reforestation has been occurring due to a decline in both agricultural production activities and rural population. Stakeholders, to better anticipate future changes, need spatially and temporally explicit models to identify areas at risk of land change and possible abandonment. This paper presents an integrated approach combining forecasting scenarios and a LULC changes simulation model to assess where LULC changes may occur in the Pyrenees Mountains, based on historical LULC trends and a range of future socio-economic drivers. The proposed methodology considers local specificities of the Pyrenean valleys, sub-regional climate and topographical properties, and regional economic policies. Results indicate that some regions are projected to face strong abandonment, regardless of the scenario conditions. Overall, high rates of change are associated with administrative regions where land productivity is highly dependent on socio-economic drivers and climatic and environmental conditions limit intensive (agricultural and/or pastoral) production and profitability. The combination of the results for the four scenarios allows assessments of where encroachment (e.g. colonization by shrublands) and reforestation are the most probable. This assessment intends to provide insight into the potential future development of the Pyrenees to help identify areas that are the most sensitive to change and to guide decision makers to help their management decisions.

Journal of Mountain Science↗

USGS2019

The evolving threat of rapid Ohia death (ROD) to Hawaii’s native ecosystems and rare plant species

Hawai‘i’s most widespread native tree, ‘ōhi‘a lehua ( Metrosideros polymorpha ), has been dying across large areas of Hawai‘i Island mainly due to two fungal pathogens ( Ceratocystis lukuohia and Ceratocystis huliohia ) that cause a disease collectively known as Rapid ‘Ōhi‘a Death (ROD). Here we examine patterns of positive detections of C. lukuohia as it has been linked to the larger mortality events across Hawai‘i Island. Our analysis compares the environmental range of C. lukuohia and its spread over time through the known climatic range and distribution of ‘ōhi‘a. Analyses show this fungal pathogen generally encompassed the core, but not the extremes of the climatic range of ‘ōhi‘a. We further modeled the potential distribution of C. lukuohia across the Hawaiian Archipelago to estimate the risk of ROD to other islands. Given the potential for C. lukuohia to alter the structure of ‘ōhi‘a dominated forests, we used our projected potential distribution of C. lukuohia to assess the risk of ROD to threatened and endangered plant species across Hawai‘i. Many native plants are likely vulnerable to these types of large ‘ōhi‘a mortality events: of 234 endangered native plant species considered, 147 (62.8%) have more than half of their range within current and expanding C. lukuohia suitable areas. We also found evidence that protecting habitat by fencing out introduced feral ungulates reduces the prevalence of the disease likely by reducing physical damage caused by these animals to ‘ōhi‘a trees, a precondition for Ceratocystis infection. Given the ongoing spread of C. lukuohia , we developed a dynamic web portal to host our results online, where models and analyses are updated with new lab-confirmed detections to provide managers with a useful tool to help monitor and assess the risk of C. lukuohia as it continues to spread.

Hawaii↗

USGS2015

Evaluating abundance and trends in a Hawaiian avian community using state-space analysis

Estimating population abundances and patterns of change over time are important in both ecology and conservation. Trend assessment typically entails fitting a regression to a time series of abundances to estimate population trajectory. However, changes in abundance estimates from year-to-year across time are due to both true variation in population size (process variation) and variation due to imperfect sampling and model fit. State-space models are a relatively new method that can be used to partition the error components and quantify trends based only on process variation. We compare a state-space modelling approach with a more traditional linear regression approach to assess trends in uncorrected raw counts and detection-corrected abundance estimates of forest birds at Hakalau Forest National Wildlife Refuge, Hawai‘i. Most species demonstrated similar trends using either method. In general, evidence for trends using state-space models was less strong than for linear regression, as measured by estimates of precision. However, while the state-space models may sacrifice precision, the expectation is that these estimates provide a better representation of the real world biological processes of interest because they are partitioning process variation (environmental and demographic variation) and observation variation (sampling and model variation). The state-space approach also provides annual estimates of abundance which can be used by managers to set conservation strategies, and can be linked to factors that vary by year, such as climate, to better understand processes that drive population trends.

Hawai'i↗

USGS2009

Web-based decision support and visualization tools for water quality management in the Chesapeake Bay watershed

Federal, State, and local water quality managers charged with restoring the Chesapeake Bay ecosystem require tools to maximize the impact of their limited resources. To address this need, the U.S. Geological Survey (USGS) and the Environmental Protection Agency's Chesapeake Bay Program (CBP) are developing a suite of Web-based tools called the Chesapeake Online Assessment Support Toolkit (COAST). The goal of COAST is to help CBP partners identify geographic areas where restoration activities would have the greatest effect, select the appropriate management strategies, and improve coordination and prioritization among partners. As part of the COAST suite of tools focused on environmental restoration, a water quality management visualization component called the Nutrient Yields Mapper (NYM) tool is being developed by USGS. The NYM tool is a web application that uses watershed yield estimates from USGS SPAtially Referenced Regressions On Watershed (SPARROW) attributes model (Schwarz et al., 2006) [6] to allow water quality managers to identify important sources of nitrogen and phosphorous within the Chesapeake Bay watershed. The NYM tool utilizes new open source technologies that have become popular in geospatial web development, including components such as OpenLayers and GeoServer. This paper presents examples of water quality data analysis based on nutrient type, source, yield, and area of interest using the NYM tool for the Chesapeake Bay watershed. In addition, we describe examples of map-based techniques for identifying high and low nutrient yield areas; web map engines; and data visualization and data management techniques.

Conference Paper↗

USGS2024

Using explainable machine learning methods to evaluate vulnerability and restoration potential of ecosystem state transitions

Ecosystem state transitions can be ecologically devastating or be a restoration success. State transitions are common within aquatic systems worldwide, especially considering human-mediated changes to land use and water use. We created a transferable conceptual framework to enable multiscale assessments of state resilience and early warnings of state transitions that can inform strategic restorations and avoid ecosystem collapse. The conceptual framework integrated machine learning predictions with ecosystem state concepts (e.g., state classification, gradients of vulnerability, and recovery potential leading to state transitions) and was devised to investigate possible environmental drivers. As an application of the framework, we generated prediction probabilities of submersed aquatic vegetation (SAV) presence at nearly 10,000 sites in the Upper Mississippi River (United States). Then, we used an interpretability method to explain model predictions to gain insights into possible environmental drivers and thresholds or linear responses of SAV presence and absence. Model accuracy was 89% without spatial bias. Average water depth, suspended solids, substrate, and distance to nearest SAV were the best predictors and likely environmental drivers of SAV habitat suitability. These environmental drivers exhibited nonlinear, threshold-type responses for SAV. All the results are also presented in an online dashboard to explore results at many spatial scales. The habitat suitability model outputs and prediction explanations from many spatial scales (4 m to 400 km of river reach) can inform research and restoration planning.

Conservation Biology↗

USGS2011

Mortality of American alligators attributed to cannibalism

Mortality of juvenile (<122 cm total length; TL) American Alligators ( Alligator mississippiensis ) attributed to cannibalism on Orange Lake, Florida was examined. Alligator web tags used in mark–recapture studies were found in 12% of 267 stomachs sampled from alligators ≥168 cm TL. Captive alligators retained 76% of force-fed tags during a 588-d tag-retention trial. Models relating the probability of tag recovery to the annual probabilities of juvenile survival, cannibalism, tag retention, adult survival, and adult harvest suggested that cannibalism may on average remove 6–7% of the juvenile alligator population annually. Vulnerability continued to 140 cm TL (age 6–8 yr). Cannibalism of juveniles may serve to regulate the alligator population on Orange Lake. Alligator cannibalism may vary widely among populations, depending on demography and environmental conditions. The role and importance of cannibalism in alligator population dynamics should be more fully assessed and environmental and population factors that influence cannibalism identified to better evaluate management programs.

Florida↗

USGS2011

Land-cover change research at the U.S. Geological Survey-assessing our nation's dynamic land surface

The U.S. Geological Survey (USGS) recently completed an unprecedented, 27-year assessment of land-use and land-cover change for the conterminous United States. For the period 1973 to 2000, scientists generated estimates of change in major types of land use and land cover, such as development, mining, agriculture, forest, grasslands, and wetlands. To help provide the insight that our Nation will need to make land-use decisions in coming decades, the historical trends data is now being used by the USGS to help model potential future land use/land cover under different scenarios, including climate, environmental, economic, population, public policy, and technological change.

Fact Sheet↗

USGS2023

Refining capture-recapture recruitment estimation methods for Atlantic sturgeon

The Atlantic sturgeon Acipenser oxyrinchus oxyrinchus was once of great commercial importance in many coastal rivers of the eastern USA. Over the 19th and 20th centuries, most historical stocks of Atlantic sturgeon were depleted by human activities. Estimating recruitment for the remaining populations is challenging due to sampling constraints, limited age data, and natural variability. However, recruitment estimates could inform recovery efforts. The objectives of this study were to compare 2 modeling approaches to estimate recruitment of age-1 Atlantic sturgeon and provide an updated index of abundance across more than a decade of sampling in the Altamaha River, Georgia. First, we constructed capture histories of river-resident juveniles, using capture-mark-recapture data collected from 2008 to 2020, and assigned ages based on length-frequency analysis. Second, we compared more traditional Huggins closed population models and a recent nonlinear extension of Huggins models—vector generalized additive models (VGAMs)—to estimate abundance of age-1 fish. Both model types indicated similar yearly age-1 abundance estimates (Huggins: 163 in 2017 to 3839 in 2010; VGAM: 312 in 2020 to 4448 in 2010), but the VGAMs provided more direct interpretation for factors that might affect capture probability (e.g. sampling effort, temperature, fish length). This study indicates that the age-1 Altamaha River Atlantic sturgeon population has remained relatively stable over the past decade and provides a long-term baseline which will better enable managers to assess the effects of either future restoration actions or environmental disturbances on the population.

Georgia↗

USGS2023

Using high-frequency monitoring data to quantify city-wide suspended-sediment load and evaluate TMDL goals

Excess sediment is a common reason water bodies in the USA become listed as impaired resulting in total maximum daily loads (TMDL) that require municipalities to invest millions of dollars annually on management practices aimed at reducing suspended-sediment loads (SSLs), yet monitoring data are rarely used to quantify SSLs and track TMDL progress. A monitoring network was created to quantify the SSL from the City of Roanoke, Virginia, USA (CoR), to the Roanoke River and Tinker Creek and help guide TMDL assessment and implementation. Suspended-sediment concentrations were estimated between 2020 and 2022 from high-frequency turbidity data using surrogate linear-regression models. Sixty-one percent of the total three-year SSL resulted from five large storm events. The average suspended-sediment yield from the CoR (58.1 metric tons/km 2 /year) was similar to other urban watersheds in the Eastern United States; however, the yield was nearly five times larger than the TMDL allocation (12.2 metric tons/km 2 /year). The TMDL allocated load was modeled based on a predominantly forested reference watershed and may not be a practical target for highly impervious watersheds within the CoR. The TMDL model used daily input data which likely does not capture the full range of SSLs during storm events, particularly from flashy urban streams. The average SSL following the five large storm events doubled that of the CoR’s annual allocated load from the TMDL. The results of this study highlight the importance of using high-frequency monitoring data to accurately estimate SSLs and evaluate TMDLs in urban areas.

Virginia↗

USGS2024

Unexpected effect of geographic origin on post-translocation survival in a long-lived reptile, the gopher tortoise

Mitigation translocations move wildlife from specific areas due to conflict with humans over land use at the site. A critical decision when carrying out mitigation translocation is the acceptable distance across which animals can be moved. This decision trades off logistical expediency of unrestricted translocation with the risk of reducing translocation success due to environmental mismatch between origin and translocation site conditions. In this study, we used a large dataset of 502 individually identifiable carcasses to examine the role of geographic origin and translocation distance in the relative survival of 2822 translocated subadult and adult gopher tortoises ( Gopherus polyphemus ), a species experiencing large-scale mitigation translocation, at a recipient site in the Florida panhandle, USA. We hypothesized that if climate or habitat differences between the origin and translocation site influenced survival, tortoises translocated from within the Florida panhandle would have the highest survival. To the contrary, we found that survival slightly increased with increasing climatic difference between origin and recipient site, driven by higher survival of tortoises coming from central Florida sites compared to those from the panhandle and north Florida. This suggests that environmental mismatch due to long-distance translocation is not a main driver of mortality. These models also indicated an effect of season, with a survival advantage to tortoises translocated in the spring and late fall, relative to summer translocations, and a negative effect of initial density on survival. Finally, we also estimated the upper bound on annual survival in three well-monitored groups to be quite low (92–95%) for several years following release, suggesting caution when considering large translocated populations to be viable without first assessing adult survival. Our unexpected results highlight the importance of investigating species-specific sensitivities to translocation distances and indicate the limitations of assumed linear effects of translocation distance on outcomes.

Florida↗

USGS2023

Crop water productivity from cloud-Based landsat helps assess California’s water savings

Demand for food and water are increasing while the extent of arable land and accessible fresh water are decreasing. This poses global challenges as economies continue to develop and the population grows. With agriculture as the leading consumer of water, better understanding how water is used to produce food may help support the increase of Crop Water Productivity (CWP; kg/m 3 ), the ratio of crop output per unit of water input (or crop per drop). Previous large-scale CWP studies have been useful for broad water use modeling at coarser resolutions. However, obtaining more precise CWP, especially for specific crop types in a particular area and growing season as outlined here are important for informing farm-scale water management decision making. Therefore, this study focused on California’s Central Valley utilizing high-spatial resolution satellite imagery of 30 m (0.09 hectares per pixel) to generate more precise CWP for commonly grown and water-intensive irrigated crops. First, two products were modeled and mapped. 1. Landsat based Actual Evapotranspiration (ET a ; mm/d) to determine Crop Water Use (CWU; m 3 /m 2 ), and 2. Crop Productivity (CP; kg/m 2 ) to estimate crop yield per growing season. Then, CWP was calculated by dividing CP by CWU and mapped. The amount of water that can be saved by increasing CWP of each crop was further calculated. For example, in the 434 million m 2 study area, a 10% increase in CWP across the 9 crops analyzed had a potential water savings of 31.5 million m 3 of water. An increase in CWP is widely considered the best approach for saving maximum quantities of water. This paper proposed, developed, and implemented a workflow of combined methods utilizing cloud computing based remote sensing data. The environmental implications of this work in assessing water savings for food and water security in the 21st century are expected to be significant.

California↗

USGS2013

Assessing factors affecting the thermal properties of a passive thermal refuge using three-dimensional hydrodynamic flow and transport modeling

Everglades restoration activities may cause changes to temperature and salinity stratification at the Port of the Islands (POI) marina, which could affect its suitability as a cold weather refuge for manatees. To better understand how the Picayune Strand Restoration Project (PSRP) may alter this important resource in Collier County in southwestern Florida, the USGS has developed a three-dimensional hydrodynamic model for the marina and canal system at POI. Empirical data suggest that manatees aggregate at the site during winter because of thermal inversions that provide warmer water near the bottom that appears to only occur in the presence of salinity stratification. To study these phenomena, the environmental fluid dynamics code simulator was used to represent temperature and salinity transport within POI. Boundary inputs were generated using a larger two-dimensional model constructed with the flow and transport in a linked overland-aquifer density-dependent system simulator. Model results for a representative winter period match observed trends in salinity and temperature fluctuations and produce temperature inversions similar to observed values. Modified boundary conditions, representing proposed PSRP alterations, were also tested to examine the possible effect on the salinity stratification and temperature inversion within POI. Results show that during some periods, salinity stratification is reduced resulting in a subsequent reduction in temperature inversion compared with the existing conditions simulation. This may have an effect on POI’s suitability as a passive thermal refuge for manatees and other temperature-sensitive species. Additional testing was completed to determine the important physical relationships affecting POI’s suitability as a refuge.

Florida↗

USGS2016

Predicting animal home-range structure and transitions using a multistate Ornstein-Uhlenbeck biased random walk

The home‐range concept is central in animal ecology and behavior, and numerous mechanistic models have been developed to understand home range formation and maintenance. These mechanistic models usually assume a single, contiguous home range. Here we describe and implement a simple home‐range model that can accommodate multiple home‐range centers, form complex shapes, allow discontinuities in use patterns, and infer how external and internal variables affect movement and use patterns. The model assumes individuals associate with two or more home‐range centers and move among them with some estimable probability. Movement in and around home‐range centers is governed by a two‐dimensional Ornstein‐Uhlenbeck process, while transitions between centers are modeled as a stochastic state‐switching process. We augmented this base model by introducing environmental and demographic covariates that modify transition probabilities between home‐range centers and can be estimated to provide insight into the movement process. We demonstrate the model using telemetry data from sea otters ( Enhydra lutris ) in California. The model was fit using a Bayesian Markov Chain Monte Carlo method, which estimated transition probabilities, as well as unique Ornstein‐Uhlenbeck diffusion and centralizing tendency parameters. Estimated parameters could then be used to simulate movement and space use that was virtually indistinguishable from real data. We used Deviance Information Criterion (DIC) scores to assess model fit and determined that both wind and reproductive status were predictive of transitions between home‐range centers. Females were less likely to move between home‐range centers on windy days, less likely to move between centers when tending pups, and much more likely to move between centers just after weaning a pup. These tendencies are predicted by theoretical movement rules but were not previously known and show that our model can extract meaningful behavioral insight from complex movement data.

Ecology↗

USGS2023

No evidence for cottonwood forest decline along a flow-augmented western U.S. river

In contrast to many other arid region rivers, streamflow in the South Platte River is heavily augmented by trans-basin water imports and irrigation return flows. Hydrological changes began in the 1880s, resulting in channel narrowing and the development of a continuous Populus-Salix forest by the mid-twentieth century. We assessed the composition, structure and regeneration status of the riparian forest and identified environmental variables affecting annual Populus deltoides tree growth. We sampled forest structure at four sites in 2015, and conducted dendroecological analysis at seven additional sites in 2019. The riparian forest was dominated by P. deltoides , which occurred at all sites, comprising 79% of total tree basal area and 62% of total tree density. Age structure data indicated ongoing though episodic recruitment of P. deltoides , at least over the past ~130 years. We tested 14 linear mixed effects models to describe the effect of climate and streamflow on individual tree growth (modeled as the log of BAI, n = 237 trees). The most parsimonious model selected with AICc explained 28.6% of BAI variability, and included hydrology and climate factors during the growing season (i.e., June–August streamflow, June–July PDSI), some aspects of off-season (i.e., previous November and March) streamflow, along with tree age and study site effects. The riparian forest developed in response to, and has been maintained by, current climate conditions and water management regimes. It may be negatively affected by future climate change and increased urban water demand in the basin.

Colorado↗

USGS2013

A comparison of data-driven groundwater vulnerability assessment methods

Increasing availability of geo-environmental data has promoted the use of statistical methods to assess groundwater vulnerability. Nitrate is a widespread anthropogenic contaminant in groundwater and its occurrence can be used to identify aquifer settings vulnerable to contamination. In this study, multivariate Weights of Evidence (WofE) and Logistic Regression (LR) methods, where the response variable is binary, were used to evaluate the role and importance of a number of explanatory variables associated with nitrate sources and occurrence in groundwater in the Milan District (central part of the Po Plain, Italy). The results of these models have been used to map the spatial variation of groundwater vulnerability to nitrate in the region, and we compare the similarities and differences of their spatial patterns and associated explanatory variables. We modify the standard WofE method used in previous groundwater vulnerability studies to a form analogous to that used in LR; this provides a framework to compare the results of both models and reduces the effect of sampling bias on the results of the standard WofE model. In addition, a nonlinear Generalized Additive Model has been used to extend the LR analysis. Both approaches improved discrimination of the standard WofE and LR models, as measured by the c -statistic. Groundwater vulnerability probability outputs, based on rank-order classification of the respective model results, were similar in spatial patterns and identified similar strong explanatory variables associated with nitrate source (population density as a proxy for sewage systems and septic sources) and nitrate occurrence (groundwater depth).

Milan District↗