The Ocean Archive System searches our original datasets as they were submitted to us, not individual points or profiles. If you want to search and retrieve ocean profiles in a common format, or objectively analyzed fields, your better option may be to use one of our project applications. See: Access Data
OAS accession Detail for 0220087, meta_version: 7. Current meta_version is: 7
| << previous | |revision: 7 |
| accessions_id: | 0220087 | archive |
|---|---|
| Title: | NCCOS spatial modeling of threatened Caribbean corals: process-based models for Acropora palmata (elkhorn coral) distributions in the U.S. Virgin Islands (NCEI Accession 0220087) |
| Abstract: | This dataset is a compilation of modeled current and future density distributions of threatened elkhorn corals Acropora palmata in the shallow water (bottom depth ≥ -20 m) off St. Thomas, St. John and St. Croix, U.S. Virgin Islands. The raster data sets contain predicted distributions of species density and the prediction uncertainty in 2013, 2014, 2015, 2035 and 2055 estimated using process-based random forest (RF) and dynamic range models (DRM). These predictions were generated to inform Caribbean A. palmata restoration plans in the U.S. Virgin Islands. |
| Date received: | 20200924 |
| Start date: | 20130101 |
| End date: | 20551231 |
| Seanames: | Caribbean Sea |
| West boundary: | -65.15859 |
| East boundary: | -64.42007 |
| North boundary: | 18.47301 |
| South boundary: | 17.62387 |
| Observation types: | GIS product, model output |
| Instrument types: | LIDAR |
| Datatypes: | CORAL |
| Submitter: | |
| Submitting institution: | US DOC; NOAA; NOS; National Centers for Coastal Ocean Science |
| Collecting institutions: | US DOC; NOAA; NOS; National Centers for Coastal Ocean Science |
| Contributing projects: | |
| Platforms: | |
| Number of observations: | |
| Supplementary information: | Methods: For more details, see Chen et al., 2020. To jointly model the spatio-temporal population dynamics and spatial habitat suitability of elkhorn corals Acropora palmata, we adopted the general framework of dynamic range models (DRMs; Pagel and Schurr, 2012) in which the demographic rates were related with a niche model that describes the response of population growth rates to environmental predictors, which were bathymetric depth, mean summer sea surface temperature, and maximum significant wave height in this study. Bathymetric depth data were derived from high resolution remote sensing products including multibeam sonar and light detection and ranging (LiDAR; NOAA). Mean summer sea surface temperature data in June-August were derived from NASA Multi-Scale Ultra-High Resolution Sea Surface Temperature (MUR SST; NASA) and were downscaled from an original 0.01° spatial resolution to 50 m through bilinear interpolation. Maximum significant wave height data were forecasted by CariCOOS Nearshore Wave model, which was based on a Simulating Waves Nearshore (SWAN) model and validated by field observations from CARICOOS wave buoys (Anselmi-Molina et al., 2012). Wave data were downscaled from an original 1 km spatial resolution to 50 m through bilinear interpolation. For the future climate scenario (years from 2016 to 2055), we assumed SST and significant wave height will increase at a rate of 0.19 °C decade−1, estimated from eleven global climate models (GCMs) downscaled for eastern Caribbean Sea based on Representative Concentration Pathways (RCPs) 6.0 scenario (Kibler et al., 2015), and 0.03 m year-1, based on a 30 years wave hindcast from 1979 to 2008 in the Caribbean Sea (Appendini et al., 2014), respectively. As with DRM framework, the latent population process was constructed with a spatially explicit, grid-based model, in which the local colonies in each grid cell is associated with neighborhood cells via dispersal on distance-dependent rates. Because the size of coral colonies is related to their contributions to population growth, a size-structured population model was used in this analysis instead of the Ricker model proposed in the original DRMs. The model was constructed with a hierarchical Bayesian approach and consists of three components: process model, data model, and parameter model. DRMs were fitted to the data using Differential Evolution Markov Chain (DE-MCZS; ter Braak and Vrugt, 2008) sampler, an efficient adaptive Markov Chain Monte Carlo sampling (MCMC) method, with two independent parallel chains for 600,000 iteration steps each in total, in which first 350,000 samples were discarded as ‘burn-in.’ Convergence of the chains was assessed using the Gelman-Rubin diagnostic test and visual inspection of the chain histories. The initial abundance in each grid cell, which is required for dynamic range models, was estimated from correlative species distribution modeling method random forest (RF; Breiman, 2001) and quadratic multinomial logit model. For each of estimated parameters, 500 sets of posterior samples were drawn from the posterior distributions and used in the spatio-temporal simulations of population dynamics from 2014 to 2055. The mean and standard deviation of the 500 simulated predicted density distributions of each year between 2014 and 2055 were calculated for the final density predictions and uncertainty. Also included in this accession are supplemental rda data sets containing Bayesian posterior distributions (unitless) of each parameter estimated from 2013-2015 observations with Differential Evolution Markov Chain sampler, an efficient adaptive Markov Chain Monte Carlo sampling (MCMC) method, with two independent parallel chains for 600000 iteration steps each in total. This dataset was calculated and can be read in R using package ‘BayesianTools’. Processing and analysis methods are described in the reference publications. Submission Package ID: 47HBUC |
| Availability date: | |
| Metadata version: | 7 |
| Keydate: | 2020-09-26 15:01:30+00 |
| Editdate: | 2020-10-30 17:43:18+00 |