Megan Mehaffey

Assessing Landscape Condition Relative to Water Resources in the Western United States: A Strategic Approach

The Environmental Monitoring and Assessment Program (EMAP) is proposing an ambitious agenda to assess the status of streams and estuaries in a 12-State area of the western United States by the end of 2003. Additionally, EMAP is proposing to access landscape conditions as they relate to stream and estuary conditions across the west. The goal of this landscape project is to develop a landscape model that can be used to identify the relative risks of streams and estuaries to potential declines due to watershed-scale, landscape conditions across the west. To do so, requires an understanding of quantitative relationships between landscape composition and pattern metrics and parameters of stream and estuary conditions. This paper describes a strategic approach for evaluating the degree to which landscape composition and pattern influence stream and estuary condition, and the development and implementation of a spatially-distributed, landscape analysis approach.

Developing a dataset to assess ecosystem services in the Midwest United States

The Midwest United States produces around one-quarter of the worlds grain supply. The demand for corn ethanol is likely to cause a shift toward greater corn planting. To be prepared for the potential impacts of increased corn production, we need a better understanding of the current state of ecosystem services in this region. In this article, we describe a unique procedure for developing a dataset containing multiple variables useful in modeling ecological responses and tradeoffs. We demonstrate how to construct a detailed land cover classification and link it to yield and agricultural practices. We used the 2001 National Land Cover Database (NLCD) to spatially constrain the datasets during overlay analysis. With this method, we found that the percent agreement between classifications was frequently greater than 80%, indicating little change to the original base layer accuracies. Using three different land cover datasets, we were able to add 18 classes for agriculture and 155 classes for natural cover. We then linked variables for yield, fertilizer, and pesticide application rates, field residue, irrigation percentages, and tillage practices to the land cover data. The final Midwest dataset contained 15.5 million grid values and 15 variables. Capturing the land cover and land management information at the 30-m grid scale allows for aggregation and modeling of the ecosystem services at a variety of spatial scales. As a final step, we demonstrate a tradeoff evaluation between corn yield and nitrogen loadings using our dataset. The effort required to develop the Midwest dataset was greater than initially anticipated; however, the benefit of being able to calculate derivative variables and add new variables justifies the time expenditure needed to create such a detailed database.

Sediment and total phosphorous contributors in Rock River watershed

Total phosphorous (TP) and total suspended sediment (TSS) pollution is a problem in the US Midwest and is of particular concern in the Great Lakes region where many water bodies are already eutrophic. Increases in monoculture corn planting to feed ethanol based biofuel production could exacerbate these already stressed water bodies. In this study we expand on the previous studies relating landscape variables such as land cover, soil type and slope with changes in pollutant concentrations and loading in the Great Lakes region. The Rock River watershed in Wisconsin, USA was chosen due to its diverse land use, numerous lakes and reservoirs susceptible to TSS and TP pollution, and the availability of long-term streamflow, TSS and TP data. Eight independent subwatersheds in the Rock River watershed were identified using United States Geological Survey (USGS) monitoring sites that monitor flow, TSS and TP. For each subwatershed, we calculated land use, soil type, and terrain slope metrics or variables. TSS and TP from the different subwatersheds were compared using Analysis of Variance (ANOVA), and associations and relationships between landscape metrics and water quality (TSS and TP) were evaluated using the partial least square (PLS) regression. Results show that urban land use and agricultural land growing corn rotated with non-leguminous crops are associated with TSS and TP in streams. This indicates that increasing the amount of corn rotated with non-leguminous crops within a subwatershed could increase degradation of water quality. Results showed that increase in corn-soybean rotation acreage within the watershed is associated with reduction in streams TSS and TP. Results also show that forest and water bodies were associated with reduction in TSS and TP. Based on our results we recommend adoption of the Low Impact Development (LID) approach in urban dominated subwatersheds. This approach attempts to replicate the pre-development hydrological regime by reducing the ratio of impervious area to natural cover wherever possible, as well as recycling or treating stormwater runoff using filter strips, ponds and wetlands. In agriculturally dominated subwatersheds, we recommend increasing corn-soybean rotation, keeping corn on areas with gentle slope and soils with lower erodibility.

Distribution of soil organic carbon in the conterminous United States

The U.S. Soil Survey Geographic (SSURGO) database provides detailed soil mapping for most of the conterminous United States (CONUS). These data have been used to formulate estimates of soil carbon stocks, and have been useful for environmental models, including plant productivity models, hydrologic models, and ecological models for studies of greenhouse gas exchange. The data were compiled by the U.S. Department of Agriculture Natural Resources Conservation Service (NRCS) from 1:24,000-scale or 1:12,000-scale maps. It was found that the total soil organic carbon stock in CONUS to 1 m depth is 57 Pg C and for the total profile is 73 Pg C, as estimated from SSURGO with data gaps filled from the 1:250,000-scale Digital General Soil Map. We explore the non-linear distribution of soil carbon on the landscape and with depth in the soil, and the implications for sampling strategies that result from the observed soil carbon variability.

Association between Natural Resources for Outdoor Activities and Physical Inactivity: Results from the Contiguous United States

Protected areas including national/state parks and recreational waters are excellent natural resources that promote physical activity and interaction with Nature, which can relieve stress and reduce disease risk. Despite their importance, however, their contribution to human health has not been properly quantified. This paper seeks to evaluate quantitatively how national/state parks and recreational waters are associated with human health and well-being, taking into account of the spatial dependence of environmental variables for the contiguous U.S., at the county level. First, we describe available natural resources for outdoor activities (ANROA), using national databases that include features from the Protected Areas Database, NAVSTREETS, and ATTAINSGEO 305(b) Waters. We then use spatial regression techniques to explore the association of ANROA and socioeconomic status factors on physical inactivity rates. Finally, we use variance analysis to analyze ANROA’s influence on income-related health inequality. We found a significantly negative association between ANROA and the rate of physical inactivity: ANROA and the spatial effect explained 69%, nationwide, of the variation in physical inactivity. Physical inactivity rate showed a strong spatial dependence—influenced not only by its own in-county ANROA, but also by that of its neighbors ANROA. Furthermore, community groups at the same income level and with the highest ANROA, always had the lowest physical inactivity rate. This finding may help to guide future land use planning and community development that will benefit human health and well-being.

Development of a spatially complete floodplain map of the conterminous United States using random forest

Floodplains perform several important ecosystem services, including storing water during precipitation events and reducing peak flows, thus reducing flooding of downstream communities. Understanding the relationship between flood inundation and floodplains is critical for ecosystem and community health and well-being, as well as targeting floodplain and riparian restoration. Many communities in the United States, particularly those in rural areas, lack inundation maps due to the high cost of flood modeling. Only 60% of the conterminous United States has Flood Insurance Rate Maps (FIRMs) through the U.S. Federal Emergency Management Agency (FEMA). We developed a 30-meter resolution flood inundation map of the conterminous United States (CONUS) using random forest classification to fill the gaps in the FIRM. Input datasets included digital elevation model (DEM)-derived variables, flood-related soil characteristics, and land cover. The existing FIRM 100-year floodplains, called the Special Flood Hazard Area (SHFA), were used to train and test the random forests for fluvial and coastal flooding. Models were developed for each hydrologic unit code level four (HUC-4) watershed and each 30-meter pixel in the CONUS was classified as floodplain or non-floodplain. The most important variables were DEM-derivatives and flood-based soil characteristics. Models captured 79% of the SFHA in the CONUS. The overall F1 score, which balances precision and recall, was 0.78. Performance varied geographically, exceeding the CONUS scores in temperate and coastal watersheds but were less robust in the arid southwest. The models also consistently identified headwater floodplains not present in the SFHA, lowering performance measures but providing critical information missing in many low-order stream systems. The performance of the random forest models demonstrates the methods ability to successfully fill in the remaining unmapped floodplains in the CONUS, while using only publicly available data and open source software.