Spatiotemporal data for studying the dry forest-rainforest ecotone in Guanacaste, Costa Rica
Walter, Jonathan (2023), Spatiotemporal data for studying the dry forest-rainforest ecotone in Guanacaste, Costa Rica, Dryad, Dataset, https://doi.org/10.25338/B80642
This data package contains data supporting the manuscript "Climate and topography control variation in the tropical dry forest–rainforest ecotone." The contents include 1) compositeMODIS_ALL_20220713.tif: a GeoTIFF image file of biweekly composite MODIS NDVI from 2000 to 2021. This was acquired using GoogleEarthEngine and subsetted to the study area.
2) guanClimate.csv: a spreadsheet of climate variables averaged for the study area. These were derived from the CHELSA interpolated climate dataset and the Multivariate ENSO Index v2 (https://psl.noaa.gov/enso/mei/).
3) guan_DEM.tif: a GeoTIFF digital elevation model for the study area.
4) guan_TWI.tif: a GeoTIFF of the Topographic Wetness Index (TWI) for the study area.
5) guanacaste.shp, .shx, .dbf, .sbx, .sbn: a shapefile of the Guanacaste administrative boundary, which is also the study area boundary.
5) ImageMetadata_20220714.csv: a spreadsheet of information about the bands contained in compositeMODIS_ALL_20220613.tif.
These reflect derived datasets developed from publicly available sources.
Methods and data sources are described briefly below and in detail in the paper.
Processing and analysis code are archived in a linked Zenodo repository.
MODIS NDVI data (L3 product MOD13Q1) were obtained using GoogleEarthEngine and subset to the study area. Ecotone areas were delineated using spatial synchrony analyses. We measured synchrony between all pairs of pixels and used network modularity analyses to identify groups of pixels with high within-group and low between-group synchrony. Comparison with field plots and known vegetation distributions corroborated that major groupings distinguished dry forest from rainforest. Ecotone areas were then pixels that were weakly associated with either group. We analyzed how the probability of a pixel being ecotone depended on elevation and topographic wetness index (TWI). TWI was calculated using WhiteBox software in R. Applying the analysis to individual years of data, we used thresholds on group association strength to discretize the ecotone area and measured the area, median elevation, elevation range, and perimeter:area ratio of the ecotone. We then analyzed inter annual variation in ecotone characteristics for long-term trends and the influence of precipitation, potential evapotranspiration, and the Multivariate El Niño Index (MEI v2).
National Science Foundation, Award: 2042453