Linking Remote Sensing Estimates of Land-Cover to Census Statistics on Land-Use for the Coterminous United States


There are two primary sources of information for studies of large-scale land-cover and land-use, space-borne remote sensing and ground-based census statistics, each with advantages and disadvantages. Regional remote sensing products typically provide high spatial resolution ($\sim$1 km$^{2}$) maps of {\it land-cover} (e.g., deciduous {\it vs.} evergreen forest). Census statistics provide more information on {\it land-use} (e.g., grazed {\it vs.} non-grazed forested land), but at lower spatial resolution; data are typically aggregated by political units like counties ($\sim$10$^{2}$ -10$^{4}$ km$^{2}$), states ($\sim$10$^{4}$ -10$^{5}$ km$^{2}$), or nations ($\sim$10$^{4}$ -10$^{7}$ km$^{2}$). Neither of these sources of information alone is sufficient for applications, such as terrestrial ecosystem modeling, that require spatially-explicit information on land-use. However, a synthesis of these two data sources may provide the necessary information. Here we develop a correlation between (1) the IGBP DISCover 1-km remote sensing land-cover map of the coterminous U.S. and (2) state-level statistics from the USDA Major Land Uses dataset. We evaluate the consistency between these two views of the U.S. landscape. We also generate a hybrid land-use product: an 0.5$\deg$x0.5$\deg$ land-use map for the coterminous U.S.

Publication Date


Journal Title

EOS, Transactions American Geophysical Union, Fall Meeting, Supplement


American Geophysical Union Publications

Document Type

Conference Proceeding