data | home

The Houston SPI consists of 66 raster images, where pixels represent predicted impervious fractional cover values (integer values between 0-100 correspond to continuous values between 0.00-1.00, or 0-100% fractional cover), at a 30m spatial resolution, triannual temporal resolution, and 22-year temporal extent. Impervious cover percentages are derived from 30m Landsat 5 Thematic Mapper (TM), Landsat 7 Enhanced Thematic Mapper (ETM+), and Landsat 8 Operational Land Imager (OLI) imagery. Training data derived from National Land Cover Database (NLCD) Percent Developed Imperviousness product from 2001, 2006 and 2011. Automated continuous fields (percent impervious) prediction based on Automated Adaptive Signature Generalization for Regression (AASGr) (Hakkenberg et al. 2019) and random forests regression. All classifications validated with out of bag NLCD comparisons as well as independent accuracy assessments (Hakkenberg et al. 2019). Citation: Hakkenberg, C. R. (2019). Houston Subannual Percent Impervious (SPI) Land Cover Dataset: 1997-2018. [Data set]. Rice University-Kinder Institute: UDP. doi.org/10.25612/837.d8nxbzwj01ad

Land cover data for the 13-county Houston-Galveston Area Council (HGAC) at 30m spatial resolution, annual temporal resolution, 21-year temporal extent, and nine class thematic resolution. Classifications are derived from 30m Landsat 5 Thematic Mapper (TM), Landsat 7 Enhanced Thematic Mapper (ETM+), and Landsat 8 Operational Land Imager (OLI) imagery. Training classes are based on NLCD land cover data from 2001, 2006, and 2011. Classification based on Automated Adaptive Signature Generalization (AASG), random forests classification, and spatio-temporal filtering algorithms. All classifications validated with independent accuracy assessments and inter-classification comparisons. Citation: Hakkenberg, C. R. (2018). Greater Houston Land Cover Change Dataset: 1997-2017 (Version 2) [Data set]. Rice University-Kinder Institute: UDP. https://doi.org/10.25612/837.zbn96g5x658z

The automatic adaptive signature generalization (AASG) algorithm overcomes many of the limitations associated with classification of multitemporal imagery. By locating stable sites between two images and using them to adapt class spectral signatures from a high-quality reference classification to a new image, AASG mitigates the impacts of radiometric and phenological differences between images and ensures that class definitions remain consistent between the two classifications. Here, I provide source code (in the R programming environment), as well as a comprehensive user guide, for the AASG algorithm. See Dannenberg, Hakkenberg and Song (2016) for details of the algorithm.

The 2.8 km2 study site in the Duke Forest Blackwood Division, NC, USA consists primarily of Piedmont secondary old-field successional pine and mature hardwood forests following selective cutting, agriculture, and grazing in the 19th and early 20th century. Specific field plot locations were based on a stratified random sampling design, randomly pre-determined within the constraints of stratified bands along an east-west and a north-south topographic gradient. Thus conceived, field plots span the spectrum of compositional variability and physiognomic types of the relatively taxonomically diverse and structurally heterogeneous study area to include upland, riparian, and bottomland forests. In all field plots, species presence was recorded following Carolina Vegetation Survey protocols for all vascular plant species in 0.01m2, 0.1m2, 1m2, 10m2, 100m2 spatially-nested subplots, and at 400m2 and 900m2 full plots. Includes two county records: Fraxinus caroliniana and Acer rubrum var. trilobum (Orange, NC).