Land cover maps of Sierra de San Javier, Tucumán, Argentina, 1986-2018

Abstract

Land cover maps of Sierra de San Javier, Tucumán, Argentina, 1986-2018 This series of Land Cover Maps (LCM) was built using atmospherically corrected surface reflectance (TIER 1) Landsat 5 ETM and Landsat 8 OLI images, with less than 10% of cloud cover. A collection of Landsat 5 (1986, 1990, 2010) and Landsat 8 (2018) images from June to December, dividing images of each year in two periods were selected: dry (June to September) and wet seasons (September to December), to take advantage of seasonal phenological differences between native and exotic (L. lucidum) forests. Image collections were then combined by calculating the median of all values at each pixel across the stack of all matching bands per year, obtaining two images per year (dry/wet season). Land cover maps were constructed by applying the machine learning algorithm Random Forest (RF), based on an ensemble of classification trees, fully run on the Google Earth Engine platform. The maps include 8 land-use categories: Subtropical montane forest (SMF, category 40) Dry forest (DF, category 41) Montane grasslands (MGr, category 30) Anthropogenic grasslands and shrubland used for livestock and temporary agriculture, a mixed class including also herbaceous agriculture and low-density urban areas (AGr, category 31 ) Sugar cane (SC, category 51) Citrus plantations, mostly lemon (CP, category 50) High/medium-density urban areas (URB, category 11). Ligustrum lucidum (L, category 42). The extent of the maps are between -65,500 to -65.160 latitude to -26.600 to -26.970 longitude, GCS_WGS1984 and 30 x 30 m of resolution.