Abstract:

Model data used for a comparison of uncertainty quantification techniques, including polynomial chaos and gaussian process regression. This data set consists of oil plume ensemble simulations generated using Texas A&M oil calculator (TAMOC, version 0.1.5, https://github.com/socolofs). We quantify the uncertainty of the model outputs caused by the uncertainty of 5 model inputs parameters in this experiment, including entrainment coefficient, entrainment ratio, gas to oil ratio, 95 percentile of droplet size (d95) and droplet size distribution spreading ratio. We only select the plume trap height and peel height as our model outputs. This dataset supports the following publication: Iskandarani, M., Wang, S., Srinivasan, A., Carlisle Thacker, W., Winokur, J., & Knio, O. M. (2016). An overview of uncertainty quantification techniques with application to oceanic and oil-spill simulations. Journal of Geophysical Research: Oceans, 121(4), 2789–2808. doi:10.1002/2015jc011366

Suggested Citation:

Shitao Wang. 2017. Four ensembles of TAMOC integral oil-gas plume simulations for comparison of uncertainty quantification techniques. Distributed by: GRIIDC, Harte Research Institute, Texas A&M University–Corpus Christi. doi:10.7266/N7QN64SQ

Publications:

Iskandarani, M., Wang, S., Srinivasan, A., Carlisle Thacker, W., Winokur, J., & Knio, O. M. (2016). An overview of uncertainty quantification techniques with application to oceanic and oil-spill simulations. Journal of Geophysical Research: Oceans, 121(4), 2789–2808. doi:10.1002/2015jc011366

Purpose:

Model data used for a comparison of uncertainty quantification techniques, including polynomial chaos and gaussian process regression.

Data Parameters and Units:

This dataset consists different ensemble simulations of the plume trap height (m) and plume peel height (m) for different uncertainty quantification techniques. The trap height is defined as the level where the momentum of the downward-moving outer plume becomes zero and the peel height is defined as the level where the momentum of the upward-rising inner plume becomes zero. All the plume simulations use the CTD profile from R/V Brooks McCall at Station B54 (28°43.945'N, 88°22.607'W). The five uncertainty input parameters are entrainment coefficient, entrainment ratio, gas to oil ratio, 95 percentile of droplet size (d95) and droplet size distribution spreading ratio. This dataset consists the perturbations we applied on these parameters (normalized to [-1,1]). The order of the parameters in the perturbation is as follows: Column 1 = entrainment coefficient Column 2 = entrainment ratio Column 3 = gas to oil ratio Column 4 = 95 percentile of droplet size (d95) Column 5 = droplet size distribution spreading ratio The details of each data file and their purposes are as follows: 1) A 5 dimensional 50,000 member reference ensemble. (mc_50000.mat) 50,000 plume simulations via Latin Hypercube Sampling techniques are included in this file. lhs_sample represents the normalized perturbations applied on these five input uncertain parameters, peel_lhs is the associated plume peel height and trap_lhs denotes the associated plume trap height. 2) A 5 dimensional 100 member latin hypercube ensemble. (lhs_100.mat) 100 plume simulations via Latin Hypercube Sampling techniques are included in this file. The names of the variables in this file follow the same convention as in 1). 3) A 5 dimensional 903 member ensemble generated for polynomial chaos projection technique. (pc_projection.mat) 903 plume simulations via a smolyak sparse quadrature technique for projection (it is simply a technique to estimate the coefficients of a polynomial series). smolyak_sample represents the normalized perturbations applied on these five input uncertain parameters, peel_sparse is the associated plume peel height response and trap_sparse denotes the associated plume trap height response. 4) Five 1 dimension 201 member equal space ensembles (simulation_1d_para1.mat; simulation_1d_para2.mat; simulation_1d_para3.mat; simulation_1d_para4.mat; simulation_1d_para5.mat). In these files, the response of the plume trap height and peel height are given with respect to different input uncertainty parameters indicated by '_para1','_para2' and etc. The numbers from 1 to 5 are sequentially associated with 5 input uncertainty parameters.

Individual Files: