01661285

Component

Remote sensing techniques provide a practical solution for obtaining information for large-scale areas, which can benefit high level planning projects, such as statewide highway construction. To better control the impacts of human development on the natural environment, ecological applications are critical to guide construction activities. For the specific purpose of wetland protection to maintain important environmental functions, wetland prediction models have been developed to map the spatial distribution of different wetland types across the study area. Thus, land-use development can be planned accordingly to minimize impacts on wetland systems. However, for models based on remote sensing data, the selection of an appropriate scale (for data acquisition, processing, and presentation) is critical to model performance. The selection of an appropriate scale is challenging in that the appropriate abstract level changes under various situations. In this paper, the authors conduct a scale analysis for selecting the appropriate scale that leads to better model performance in terms of wetland classification accuracy. The methodology and analysis process can provide insights for wetland modeling applications for specific study areas in the future.

This paper was sponsored by TRB committee ADC30 Standing Committee on Ecology and Transportation.

18-01812

English

16p

2018

Transportation Research Board 97th Annual Meeting

Digital/other

Figures; References; Tables

Classification; Environmental impacts; Geographic information systems; Laser radar; Machine learning; Remote sensing; Wetlands

Data and Information Technology; Environment; Highways

Transportation Research Board Annual Meeting 2018 Paper #18-01812

TRIS, TRB, ATRI

Jan 8 2018 10:27AM