The data set consists of the bathymetry of the Red Sea and an ensemble (50 members) of time-dependent 3D flow and scalar fields on a regular grid (500x500x50, 60 time steps) covering one month of simulation time. The size of the ensemble data in uncompressed NetCDF format is 1.5 TB (64 GB compressed).
The different ensemble members were generated with an ensemble data assimilation system based on the MIT ocean general circulation model (MITgcm) and the Data Research Testbed (DART). The Ensemble Adjustment Kalman filter (EAKF) was used for assimilation in this experiment; it samples the ensemble members deterministically from the estimated posterior, assumed Gaussian - Kalman based, and conditioned on the available observations (here satellite Sea Surface Temperature, Sea Level anomalies and in situ Salinity and Temperature data were assimilated).
In more detail, the different ensemble members are the forecasts from 50 different MITgcm setups, prepared by perturbing initial conditions and model physics and driving each of the MITgcm with different atmospheric forcing extracted from the 50-member atmospheric ensemble forcing of the TIGGE project. Each of the MITgcm is configured for the domain 30°E-50°E and 10°N-30°N covering the whole Red Sea, including the Gulf of Suez, the Gulf of Aqaba, and part of the Gulf of Aden where an open boundary connects it to the Arabian Sea. They are implemented on Cartesian coordinates at an eddy-resolving horizontal resolution of 0.04° x 0.04° (4km) and 50 vertical layers, with 4m spacing at the surface and 300m near the bottom. The bathymetry, which is derived from the General Bathymetric Chart of the Ocean (GEBCO, available at http://www.gebco.net/data_and_products/gridded_bathymetry_data), is the same across all the 50 different MITgcm setups. More details about the configuration can be found in Sivareddy et al. (2020).
If you use this data set in your publications, please acknowledge with the following:
Red Sea data courtesy of Red Sea Modeling and Prediction Group (PI Prof. Ibrahim Hoteit), KAUST available at https://kaust-vislab.github.io/SciVis2020/data.html.and cite the paper
H. Toye, P. Zhan, G. Gopalakrishnan, A. Kartadikaria, H. Huang, O. Knio, and I. Hoteit: Ensemble data assimilation in the Red Sea: sensitivity to ensemble selection and atmospheric forcing. Ocean Dynamics, 67, 915-933, 2017.
Download the data from here:
Data Repository
Password: SciVisContest2020
You can either download all ensembles plus the bathymetry in a zipped archive (~64GB) by pressing the "Download" button on the upper right, or as individual files. Each ensemble member extracts to one netcdf file (~32GB), the whole data set extracts to approximately 1.5 TB.
After you have downloaded some or all ensemble members as .tgz files you can check against the provided md5 checksum to see if the download succeeded.
As an example, we demonstrate how to load the data in ParaView. ParaView can load the netcdf file, but will not assemble all the data correctly. This requires a few steps which are shown in this Python script (resampling the staggered grid, requires ParaView 5.7 or later). Run ParaView from the directory where the data was extracted (i.e., containing the Folder SciVisContest2020), or update the file names, and load the script as a state.
For information on the staggered grid, see http://mitgcm.org/sealion/online_documents/manual.pdf (Chapter 2.10.5, FIgure 2.9).