Dataset Authors

Natalie Perlin


The dataset contains the results of the 2010 Deepwater Horizon (DWH) oil spill incident at Macondo well in the Gulf of Mexico, obtained from the ensemble of numerical simulations using an updated version of the oil application of the Connectivity Modeling System (CMS) or oil-CMS. The ensemble modeling is a well-established approach in a field of numerical prediction that provides a probabilistic framework for the analysis, aimed at improving the reliability of model estimates and at demonstrating the integrity of modeling studies. The dataset includes the main case scenario of a DWH blowout starting on 2010-04-20 and ten additional simulations that use some variations in the environmental conditions; such method enables the ensemble model spread in numerical estimates while remaining qualitatively consistent. Different conditions were enabled by starting model runs at 3-day intervals from 2010-04-23 – 2010-05-20. The scenario that starts on 2010-04-20 is identical to the “DWH_logDSD_untreated” simulation in the related dataset R6.x805.000:0085. The current dataset contains the post-processed model results yielding daily surface oil concentrations for all the experiments on a regular 0.02-degree horizontal grid, and the ungridded sedimented/beached oil mass. CMS has a Lagrangian, particle-tracking framework, computing particle evolution and transport in the ocean interior, the website for the main CMS code could be found: GitHub - beatrixparis/connectivity-modeling-system: The CMS is a multiscale stochastic Lagrangian framework developed by Paris' Lab at the Rosenstiel School of Marine, Atmospheric & Earth Science to study complex behaviors, giving probabilistic estimates of dispersion, connectivity, fate of pollutants, and other Lagrangian phenomena. This repository facilitates community contributions to CMS modules. The oil-CMS module is still under development. All the simulations lasted 100 days from the onset of the blowout; the oil well was open for 87 days in each case. Three thousand particles were released every two hours, for 87 days, equivalent to a total of 3,132,000 oil particles released during the simulation. Initial particle sizes were determined at random by the CMS in the range of 1-500 micron and were scaled during post-processing analysis to represent the log-normal initial droplet size distribution (DSD) for the untreated oil. The parameters for the log-normal DSD were mu=117 micron and sigma=0.72, chosen to represent oil not treated with chemical dispersants. Each particle contained three (3) pseudo-components accounting for the differential oil density as follows: 10% of light oil with a density of 800kg/m^3, 75% of the oil with 840 kg/m^3, and 15% of heavy oil with 950 kg/m^3 density. The half-life decay rates of oil fractions were 30 days, 40 days, and 180 days, respectively. The surface evaporation half-life was set to 250 hours; horizontal diffusion was set to 10 m^2/s in the present case; and model time step was 1200s. Ocean hydrodynamic forcing for the CMS model was drawn from the HYbrid Coordinate Ocean Model (HYCOM) for the Gulf of Mexico region on a 0.04-deg. horizontal grid and 40 vertical levels from the surface to 5500 m. It provided daily average 3-D momentum, temperature, and salinity forcing fields to the CMS model. The surface wind drift parameterization used surface winds and wind stresses from the 0.5-degree Navy Operational Global Atmospheric Prediction System (NOGAPS). The transport and evolution of the oil particles were tracked by the oil-CMS model during the 100 days of simulations, recording each particle’s horizontal position, depth, diameter, and density into the model raw output files every two hours. The model data needed to be post-processed to obtain oil concentrations and oil mass estimates; the post-processing algorithm takes into account the amount of oil spilled during the incident as estimated from the reports (730,000 tons) and the assumptions about the oil particle size distribution (DSD) at the time of the release as estimated in the prior studies.


Supplemental Information

Three-dimensional oil distribution (unprocessed results): time [days since initial blowout], latitude [decimal degrees], longitude [decimal degrees], vertical layer boundaries [m], depth (bottom, [m]), daily oil concentrations [ppb]. The output of the oil-CMS representing the trajectory, history, and attributes of individual oil droplets is post-processed to compute the spatio-temporal evolution of oil mass and volumetric concentrations. The oil concentrations are in daily average values in units of ppb, the oil mass is in kg.|The current dataset contains post-processed gridded and non-gridded analyses for the cases of untreated oil. The naming convention of the files are such that the blowout day (beginning of the run) is marked in the file name indicating day, month and two-digit year, i.e., *20Apr10*, 23Apr10*, …, 20May10*. The data for the surface oil concentrations are daily average values in ppb units. The horizontal 0.02-degree grid covers the entire Gulf of Mexico. Additional non-gridded files contain sedimented oil mass that reached the ocean floor/beach and no longer is transported by the oil-CMS model, in *.txt file format. The data represents the scaled oil mass (in kg) for a given droplet, in non-gridded format (minimally post-processed), and presented in a particle-tracking framework. The format of the *landed.txt files is as following: 1st column - time in seconds from the beginning of the run (i.e., onset of the blowout); 2nd column - depth where the particle landed/sedimented; 3rd column - latitude position of the landing site; 4th column - longitude position of the landing site; 5th column - oil mass in kg. In order obtain a gridded data, or to find a summary of oil at any particular location and/or time frame, the sum of all particles that reach a given location or given time cell needs to be done, across all the output data entries (lines). Post-processed files were created using Matlab versions R2016b and R2017a. The NetCDF data files were created using compression capability to keep file size small; maximum compression, or ‘DeflateLevel’ = 9 was used. Numerical simulations and post-processing were performed using a Pegasus supercomputer at the Center of Computational Science, University of Miami, in 2018-2019. A few figures are included to demonstrate the ensemble approach to estimating probabilistic quantities of oil concentration or sedimented oil mass exceeding a chosen threshold.||daily surface oil concentration (ppb) time (in seconds from the beginning of a the oil spill) of sedimented/beached oil (in kg), latitude (degrees), longitude (degrees), depth (m)||(for the CMS model): Paris C.B., J. Helgers, E. van Sebille, and A. Srinivasan, 2013: Connectivity Modeling System: A probabilistic modeling tool for the multi-scale tracking of biotic and abiotic variability in the ocean. Environmental Modelling & Software, 42, 47-54. (for the previous version of oil-CMS model): Paris., C.B., M. Le Hénaff, Z.M. Aman, A. Subramaniam, J. Helgers, D.-P. Wang, V.H. Kourafalou, and A. Srinivasan, 2012: Evolution of the Macondo Well Blowout: Simulating the Effects of the Circulation and Synthetic Dispersants on the Subsea Oil Transport. Environ. Sci. Technol., 46, 13293−13302. (for the current version of oil-CMS model): Perlin N., C.B. Paris, I.Berenshtein, A.C. Vaz, R. Faillettaz, Z.M. Aman, P.T. Schwing, I.C. Romero, M. Schlüter, A. Liese, N. Noirungsee, and S. Hackbusch, 2020: Far-Field Modeling of a Deep- Sea Blowout: Sensitivity Studies of Initial Conditions, Biodegradation, Sedimentation, and Subsurface Dispersant Injection on Surface Slicks and Oil Plume Concentrations. In: Murawski, S. et al. (eds.) Deep Oil Spills. Facts, Fate, and Effects. Springer, Cham,


This data was computed to provide a probabilistic ensemble framework for the estimates of the surface oil concentrations and sedimented/beached oil in the Deepwater Horizon oil spill. It provides state-of-the-art modeling data of the far-field hindcast simulations, with the ensemble approach aimed at improving the reliability of model estimates and at demonstrating the integrity of modeling studies. The data may be used for comparative analysis with the observations and other validation techniques, as well as with other modeling efforts in sensitivity studies. It may also be used as spatio-temporal input for ecosystem modeling applications.


Oceans, geoscientific information, deep-water oil spill, numerical modeling, oil transport modeling, environment, Connectivity Modeling System, CMS, surface oil distribution, beached oil distribution, oil-CMS, oil concentrations




May 2023

Point of Contact


Claire B. Paris-Limouzy


University of Miami / Rosenstiel School of Marine and Atmospheric Science


Natalie Perlin


University of Miami / Center for Computational Science

Funding Source




Rights Information

Creative Commons License
This work is licensed under a Creative Commons Public Domain Dedication 1.0 License.