Abstract:

The data consists of an ensemble of HYbrid Coordinate Ocean Model (HYCOM) forecasts that sample the uncertainty in the initial conditions using a Latin Hypercube approach. The samples were generated in order to investigate the construction of surrogates for uncertain time series of a given quantity of interest. The perturbations consist of the two leading Empirical Orthogonal Function (EOF) modes identified in Iskandarani et al., (2016). The HYCOM configuration is similar to the experiment 20.1 used operationally by the US Navy for ocean prediction during the period 2003–2010 (see https://hycom.org/data/goml0pt04/expt-20pt1 for more details). The model has a 1/25° (~4 km) horizontal resolution and 20 vertical levels. The computational domain is open along portions of its southern, eastern and northern boundaries, where values are provided by a lower-resolution (1/12◦ vs. 1/25◦) simulation of HYCOM configured for the Atlantic Ocean. The model is forced at the surface by three hourly outputs from the Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMP), which has a 27-km resolution. For the present study, model simulations are run for 2 months, from May 1 to June 29, 2010. Each model output is the daily averaged, centered at 12 noon, of the date corresponding to the file name (e.g. archm.2010_121_12.cdf is the file corresponding to the 121st day of the year 2010).

Suggested Citation:

Mohamed Iskandarani. 2020. Uncertainty Quantification (UQ) of Gulf of Mexico forecasts stemming from initial conditions uncertainties, surrogate construction for time-dependent uncertain time series. Distributed by: GRIIDC, Harte Research Institute, Texas A&M University–Corpus Christi. doi:10.7266/n7-d8ga-6c22

Purpose:

Investigate whether temporal information can be used effectively to build more accurate Polynomial Chaos (PC) surrogates. A variety of approaches were tried for the surrogate construction (PC or GPR) and different levels of complexities were tried for the time-correlations. The input uncertainties were in the initial conditions that were perturbed according to the EOF in Iskandarani et al., (2016).

Data Parameters and Units:

X, integer array referencing the index of the i-th zonal grid point in the hycom grid (unitless); Y integer array referencing the index of the j-th meridional grid point in the hycom grid. (unitless). layer_sigma, list of target densities in the HYCOM model (kg/m^3) dpmixl, depth of the mixed-layer, (Pa) th3d, potential density with dimensions, (kg/m^3) time, Time in julian days,(day) surflx, surface net thermal energy flux into ocean, (W/m^2) salflx, surface salinity salt flux into ocean (psu m/s kg/m^3) into ocean plongitude' longitude (degrees) for p points 'platitude' latitude (degrees) for p points 'ulongitude' longitude (degrees) for u points; 'ulatitude' latitude (degrees) for u points; 'vlongitude' longitude (degrees) for v points; 'vlatitude' latitude (degrees) for v points; 'montg1' montgomery potential non-dimensional; 'ssh' sea surface height (m); 'dpbl' turbulent boundary layer depth (m); 'tmix' temperature of mixed layer (Celsius); 'smix' Salt concentration in mixed layer (psu); 'umix' zonal velocity of mixed layer (m/s); 'vmix' meridional velocity of mixed layer (m/s); 'ubaro' barotropic zonal velocity (m/s); 'vbaro' barotropic meridional velocity (m/s); 'u' zonal velocity in each layer (m/s); 'v' meridional velocity in each layer (m/s); 'dp' potential density (sigma); 'temp' temperature (degree Celsius); 'saln' salt (psu);

Methods:

HYCOM simulations.

Provenance and Historical References:

Iskandarani, M., Le Hénaff, M., Thacker, W. C., Srinivasan, A., & Knio, O. M. (2016). Quantifying uncertainty in Gulf of Mexico forecasts stemming from uncertain initial conditions. Journal of Geophysical Research: Oceans, 121(7), 4819–4832. doi:10.1002/2015jc011573

Individual Files: