Search

The Conservation Biology Institute (CBI) used a combination of remote sensing and machine learning algorithms deployed on the innovative cloud-computing platform, Google Earth Engine, to map grassland characteristics. We used a rich suite of enviro-climatic data, multiple sources of satellite imagery, and Random Forest modeling techniques to predict land cover for study areas in Washington, Colorado, and Kansas, where CRP Grasslands holdings are most prevalent. We used machine learning to create predictive maps of vegetation type by leveraging an extensive set of satellite-derived variables, environmental layers, and federal survey data (from BLM’s AIM and USDA NRCS’s NRI programs). Our initial investigation utilized Landsat 8 satellite data to model vegetation cover across the Washington study area and then scaled up to the Colorado-Kansas study area. The Washington study site was selected for further model enhancements and an in-depth comparison of Landsat 8, Sentinel-2, and MODIS satellite imagery, to evaluate differences in model development and performance among sensor types. We generated vegetation cover predictions for the year 2019 using Random Forest classification models. Classified outputs for the five vegetation cover models - annual grass, perennial grass, annual forb, perennial forb and bare soil - were post-processed to exclude water and urban land cover and areas that were not relevant for mapping grasslands.

CBI initially developed predictive maps of tree height, tree density, biomass, basal area, and forest type using Random Forest machine learning models. Numerous satellite-derived indices from the European Space Agency���s (ESA) Sentinel-1 and Sentinel-2 sensors, in addition to soils and topography data, were used as predictor inputs. We then refined these predictive models, focusing primarily on biomass improvements, by implementing new methods for processing Sentinel-1 imagery on the cloud computing platform Google Earth Engine (GEE); significantly updating model code; and incorporating preliminary data products derived from NASA���s spaceborne LiDAR mission - the Global Ecosystem Dynamics Investigation (GEDI). We refined the GEDI LiDAR-derived data products and included them in our models, and overall accuracy for the four forest regression models ranged from 57% to 91%. The Biomass model saw the greatest improvement in accuracy with the R2 increasing by 8%, from 49% to 57%. The Basal Area and Tree Height models both had minor 1-2% increases in accuracy, while the Tree Density model had no improvement. The Forest Type classification model had a negligible improvement in overall accuracy, however, the Elm/Ash/Cottonwood class increased in accuracy by ~6%, from 64% to 70%.

Conservation Compliance ties eligibility for most Federal farm program benefits to soil and wetland conservation. To be eligible for farm program benefits, farmers must apply an approved soil conservation system on highly erodible cropland (Highly Erodible Land Conservation, or HELC) and refrain from draining wetlands (Wetland Conservation, or WC). Conservation Compliance is effective when the incentive—the farm program benefits that could be lost due to noncompliance—exceeds the cost of meeting soil and wetland conservation requirements. HELC significantly reduced soil erosion on highly erodible cropland and may have also encouraged erosion reduction on land not subject to HELC. Compliance incentives (farm program benefits) under the Agricultural Act of 2014 are found to (1) vary widely across farms with cropland in HEL (highly erodible land) fields, (2)approximate the overall level of incentive that would have been provided under an extension of the 2008 Farm Act (although incentives changed significantly on many farms), and (3) be significantly lower on many farms if crop insurance premium subsidies were not subject to Conservation Compliance. Compliance incentives for WC, although measured only in the Prairie Pothole region of the Northern Plains, are clearly larger than Compliance costs for an estimated 75 percent of wetlands that are already cropped or have characteristics (e.g., productivity, topography) that are favorable to crop production.

We specify a system of equations that fully reflects the supply and demand sides of the market for agricultural open space at equilibrium. Although simple, the system is exceedingly flexible and allows for household and parcel heterogeneity. We derive an empirical model directly from the structural equations and contrast this using a simulated landscape with the econometric specification most often found in the literature. We then show how the model can be used to project land-use change into the future and for policy simulation. Finally, we use the model to examine the impact of common land conservation policies in Europe., Discussion Paper / SFB 823;02/2012

Focusing on individual motorists in car-owning households in Germany, this analysis econometrically investigates the determinants of automobile travel for non-work service activities against the backdrop of two questions: 1) Does gender play a role in determining the probability of car use and the distance driven? 2) If so, how is this role mitigated or exacerbated by other socioeconomic attributes of the individual and the household in which they reside? Drawing on a panel of data collected between 1996 and 2003, we specify Heckman's sample selection model to control for biases that could otherwise arise from the existence of unobservable variables that determine both the discrete and continuous choices pertaining to car use.The results indicate that although women,on average, undertake more non-work travel than men, they undertake less of such travel by car, implying a greater reliance on other modes. Moreover, employment status, age, the number of children, automobile availability, and the proximity to public transit are all found to have significantly different effects on the probability of non-work car travel between men and women, but - with the exception of automobile availability - not on the distance driven.Taken together, these results suggest that policies targeted at reducing automobile dependency and associated negative externalities such as congestion are unlikely to have uniform effects across the sexes, findings having implications for both policy evaluation as well as travel demand forecasting.

This study examines the determinants of urbanized area across a 10,000-mile square swath in central North Carolina, an area undergoing extensive conversion of forest and agricultural land.We model the temporal and spatial dimensions of these landscape changes using a database that links five satellite images spanning 1976?2001 to a suite of socioeconomic, ecological and GIScreated explanatory variables. By specifying the complementary log-log derivation of the proportional hazards model, we employ a methodology for modeling a continuous time process – the conversion of land to impervious surface – using discrete-time satellite data. Spatial effects are captured by several variables derived from the imagery that measure the landscape configuration surrounding a pixel. Empirical results confirm the significance of several determinants of urbanization identified elsewhere in the literature, including proximity to roads and population density, but also suggest that the parameterization of these variables is biased when the influence of landscape configuration is unaccounted for. We conclude that the inclusion of spatial pattern metrics significantly improves both the explanatory and predictive power of the estimated model of urbanization.

This paper advances an empirical model assessing how, over both time and space, changes in land-use respond to changing economic and ecological conditions. Focusing on Central North Carolina, a region that has undergone extensive changes in forest cover and agricultural lands over the past two decades, landscape dynamics are modeled by exploiting a spatial database that links several satellite images spanning the years 1975-1999 to a suite of socioeconomic, institutional and GIS-created explanatory variables.

The Pollution Abatement Costs and Expenditures (PACE) Survey provides the only comprehensive source of data on polluation costs by manufacturing facilities in the U.S. Collected historically by the Bureau of Census until 1994, the U.S. Environmental Protection Agency’s (EPA) National Center for Environmental Economics initiated an agreement with the Bureau of Census to reinstitute the survey with the collection of 1999 data. This paper provides a brief history of the survey, including uses of the data by the EPA. We then discuss the issues we encountered with regard to scope and measurement concerns when redesigning the 1999 survey. The paper concludes with thoughts regarding the future direction of the PACE survey at the EPA.