Estimating Infrastructure Conditions Based on Inferential Geospatial Modeling: Fid-Bau Portal

Estimating Infrastructure Conditions Based on Inferential Geospatial Modeling

Zhang, Su / Bogus, Susan M. / Lippitt, Christopher D.

All infrastructure management agencies face a big challenge that every year they need to inspect the conditions for numerous assets. With routinely acquired and publicly available geospatial data and inferential geospatial modeling (IGM), there is a potential for a new approach to substantially reduce the number of survey sites required to characterize overall infrastructure condition. Using roadway pavement assets in the State of New Mexico as an example, this research investigated if overall conditions could be estimated based on inferential geospatial modeling. Three types of geospatial data were collected and used, including traffic factors, environmental factors, and topographic factors. A total of 17 variables were extracted from those data and used as explanatory. Results demonstrate that overall pavement surface condition can be effectively estimated based on traffic volumes, environmental conditions, and topography. For New Mexico, the minimum number of survey sites required to characterize overall pavement surface conditions within the range of errors expected due to human interpretation error is 4,000, which is substantially less than the ~12,000 annual samples currently collected. These results indicate great promise in the use of IGM to estimate overall infrastructure condition and an opportunity for substantial time and cost savings.