Using Web-enabled Landsat Data time series to analyze the impacts of urban areas on remotely sensed vegetation dynamics

Earth is currently experiencing rapid urban growth with >50% of global population living in urban areas. Urbanization occurs as cities expand to meet the demands of increasing populations and socioeconomic growth. Consequently, there is a need for remote sensing research to detect, quantify, and monitor urbanization and subsequent impacts on the environment. Here we used Normalized Difference Vegetation Index (NDVI) data products derived from the Web-enabled Landsat Data (WELD) project to (1) characterize the response of vegetation to urban land cover change and (2) analyze the impacts of urban areas on land surface phenology across rural to urban gradients for two cities located on the United States Great Plains. Here we fit the decade (2003-2012) of NDVI observations as a quadratic function of thermal time to calculate land surface phenology (LSP) metrics and characterize vegetation dynamics on an urban-rural gradient. We found croplands to exhibit greater variation in NDVI at half thermal time to peak compared to forest and developed land cover types. We found a linear relationship between modeled peak height NDVI and NDVI at half thermal time to peak in forest and developed pixels, as well as pixels that experienced a land cover change from cropland to developed. In general, duration of season decreased with distance from the city center in deciduous forest pixels for both cities. Developed pixels had lower modeled peak height NDVI, longer duration of season and greater variation compared to forest pixels.