Impervious Surfaces from High Resolution Aerial Imagery: Cities in Fresno County

Impervious surfaces are landscapes covered by material with little or no water permeability, such as pavement, sidewalks, parking lots, and other hard surfaces. Impervious surfaces prevent water from being absorbed into the ground, which can lead to a number of environmentally negative impacts, such as flooding, erosion and pollution of streams, and waterways. An impervious surface study involves mapping and analyzing the extent and distribution of impervious surfaces within the study areas such as cities, counties, census tracts, etc. The study typically uses remote sensing data - satellites/aerial imagery, to identify and classify impervious surfaces. Fresno County was classified and assessed using pixel-based, object-based, and deep-learning classification. The deep learning classification outperforms in classification accuracy assessment, showing 85-92% overall accuracy. The analysis of percent impervious surfaces in cities in Fresno County is estimated to be 45% which is equivalent to that of medium-density residential areas. For the Fresno/Clovis city area, the percent impervious surface increased from 53% in 2010 (EnviroAtlas) to 63% in 2020. This 10% increase in 10 years is aligned well with the population increase.

Principal Investigator: 
Yushin Ahn
PI Contact Information: 

yahn@mail.fresnostate.edu

California State University, Fresno 

Dates: 
January 2021 to August 2023
Implementation of Research Outcomes: 

Findings:
●    The average impervious cover across all of the cities of Fresno County using the deep learning method is estimated to be 45%.
●    For the Fresno/Clovis city area, the percent impervious surface increased from 53% in 2010 (EnviroAtlas) to 63% in 2020. This 10% increase in 10 years is aligned well with the population increase.
●    The percentage of impervious cover is not strictly related to city size. For example, the highest percentage of impervious cover in Fresno County is found in the city of Kerman (pop. 16,000).
●    Deep learning classification provided more accurate results than pixel-based and object-based classification methods.

Impacts/Benefits of Implementation: 

Future classification studies should consider the use of deep learning when practicable because it provides more accurate results than traditional pixel- or object-based methods in an off-the-shelf software package. City codes and land use planning should take into account the negative impacts of excessive impervious cover on the urban fabric. While impervious cover may be perceived traditionally as a “big city” problem, small city planners also need to reckon with the negative impacts of excessive concrete and asphalt.

Project Number: 
2257

-

CSUTC
MCEEST
MCTM
NTFC
NTSC

Contact Us

SJSU Research Foundation   210 N. 4th Street, 4th Floor, San Jose, CA 95112    Phone: 408-924-7560   Email: mineta-institute@sjsu.edu