Evaluation of the accuracy of national land use datasets for the development of urban canopy parameterizations
Timothy N. McPherson, LANL, Los Alamos, NM; and M. J. Brown and S. J. Burian
Describing urban terrain and land use/land cover (LULC) characteristics accurately is an essential element in environmental modeling, assessment and management. Urban terrain and LULC information are especially important in meteorological modeling applications, such as simulations of atmospheric flow over cities, defining surface energy budgets, and fate and transport studies of urban contaminants. Urban terrain data is difficult to acquire and many researchers rely on LULC data to develop urban canopy parameterizations for mesoscale meteorological models. Although urban canopy parameterizations are an extremely useful tool in mesoscale meteorological models, they are often applied using out of date or poorly attributed land use data. The two major sources of national land use data are the U.S. Geological Survey (USGS) land use/land cover dataset and the National Land Cover Database (NLCD). The USGS database was developed based on aerial photographs from circa 1970 and may be significantly underestimating urban areas due to the extensive urban development that has occurred in the United States since the time of the imagery. The NLCD was constructed from more recent LANDSAT TM imagery (circa 1990), but it has limited attribution of urban classes. The age of the data and level of detail could be significantly impacting urban canopy parameterizations and their associated mesoscale models and the sufficiency of these two datasets for developing urban canopy parameterizations needs to be assessed.
The objective of this investigation is to compare qualitatively and quantitatively the USGS dataset and the NLCD with more detailed and recent datasets. We characterize the shortcomings of the USGS land use and NLCD, especially for environmental modeling applications. First, we analyze the similarities and differences between the urban footprint estimated from the USGS land use dataset, the NLCD, and more detailed land use datasets for select cities. Secondly, we evaluate the ability of the USGS dataset and NLCD to characterize correctly the urban land use classes from the detailed land use datasets. And finally, we present a new method for deriving urban footprints for the US.
Extended Abstract (548K)
Session 9, fine scale modeling with improved land surface, land cover databases (parallel with sessions J1, J2, J4, J5, 3, and 10)
Wednesday, 25 August 2004, 8:25 AM-2:45 PM
Previous paper Next paper
Browse or search entire meeting
AMS Home Page