A Hybrid Coordinate Ocean Model (HYCOM) for Data-Assimilative Ocean Modeling Year 2 Implementation Plan -------------------------- December 2000 This implementation plan summarizes the second year goals of the consortium for a) the release of HYCOM 2.0, b) the global and basin scale simulations, c) the data assimilation approach, and d) evaluation of the results. See the year 1 implementation plan for additional background information (available from http://panoramix.rsmas.miami.edu/hycom_nopp/ under internal) since this implementation plan only outlines the incremental tasks to be performed in year 2. 1. HYCOM 2.0 RELEASE 1.a Status of HYCOM 1.0 HYCOM 1.08 was released in August 2000: - MICOM-like coding style - More ``main-stream'' than MICOM - MKS units throughout - Grid orientation is West to East (x-axis), then South to North (y-axis) - User-tunable model parameters are read in at run time - Kraus-Turner or KPP mixed layer - Coordinate choices (i.e. isopycnal, pressure or sigma) - For shared memory machines only - Scalability via OpenMP (2-32 cpus) - Bit-for-bit multi-cpu reproducability - Simplified ``named pipe'' debugging capability HYCOM 1.10 will be released by the end of December 2000: - Minor bug fixes - Major capability upgrade - MICOM-like mode - Energy loan ice model - More control of hybrid layers - High frequency atmospheric forcing - Simplified OpenMP logic - Use ``j-loop'' subroutines when dealing with vertical processes - Minimizes the OpenMP changes when modifying the code - Improved climatology interpolator - Improved interface to plot package Limitations of HYCOM 1.0: - Fortran 77 (MICOM-like) coding style - Only for shared memory machines - Periodic (global) regions not supported - Nested-domain open boundaries not supported - PAKK I/O not efficient or accurate 1.b HYCOM 2.0 Remove the limitations of HYCOM 1.0: - Update to cleaner, Fortran 90 based, coding style - Add MPI/SHMEM option (P-HYCOM) - Either MPI or OpenMP or both selectable at compile time (single source code) - Add halos for MPI to automatically supports periodic boundaries - Pan-Am grid will requires a special halo exchange - Support nested-domain open boundaries - Based on (new) archive files - Interpolate to target domain off-line - Need to only deal with one domain - netCDF the is eventual target - Not scalable or thread-safe - Would limit code portability if required by HYCOM - How to represent Pan-Am grid? - HYCOM 2.0 reads/writes ``.a and .b'' files - ``.a'' is a raw IEEE REAL*4 array file (Fortran direct access) - ``.b'' is a plain-text header file (Fortran formatted) - This I/O is simple and portable - It can easily be parallelized - Have the N-th processor read/write every N-th 2-D array record - Require the record length to be a multiple of 64KB? - Convert to netCDF off-line Implementation: Wallcraft, O'Keefe, Bleck 2. CODE DEVELOPMENT MIXED LAYER MODELS: - Add the option to choose R. Bleck's Kraus-Turner bulk mixed layer model to the existing KPP and Kraus-Turner choices (implemented by G. Halliwell) - Evaluate Canuto's mixed layer model (Implementation: Romanou, Bleck) - Look into the possibility of adding Mellor-Yamada 2.5 (Halliwell?) MODE SPLITTING: Left over from year 1 => Higdon's mode splitting method will be implemented. Implementation: Bleck SEA-ICE/POLAR GRID: Ice effects are now incoporated in HYCOM 1.0 using a simple energy loan sea ice model. Inclusion of a full-fledged sea ice model will be considered once the polar grid in implemented. A "bipolar" grid generated by placing two poles on opposite sides of the 65N parallel ("Pan Am" grid) will be smoothly connected to the conventional Mercator projection grid. To be incorporated in HYCOM 2.0. Implementation: Bleck, Wallcraft EQUATIONAL WAVE GUIDE: Ocean dynamics near the equator are modulated by Rossby and Kelvin waves traveling along the equator, as well as by equatorial Ekman suction. Capturing the correct phase speed and latitudinal structure of higher modes of the equatorial waves is facilitated by reducing the meridional mesh size in the equatorial belt. Left over from year 1 => Under investigation (low priority) Implementation: Bleck OPEN BOUNDARY CONDITIONS: The released version of HYCOM now incorporates the traditional "buffer zone" approach in which open ocean boundaries are treated as closed, but are outfitted with buffer zones in which temperature T and salinity S are linearly relaxed toward climatological (or any other observed) values. These buffer zones restore the T and S fields to observations in order to approximately recover the vertical shear of the currents through geostrophic adjustment. There are no constraints on the barotropic circulation. A second boundary condition was implemented during year 1. These boundary conditions are "open" in the sense that relaxation to mass fluxes, interface depths, T, S, and density is prescribed in a finite-width sponge zone, and that the barotropic pressure and velocities are advected into/out of the domain via characteristics. This boundary condition needs to be further tested and integrated in the next HYCOM release. Implementation: Townsend, Wallcraft, Smith, Chassignet ADVECTION SCHEME: Truncation errors in ADVEM lead to numerical cabelling when both T and S are advected separately. These are minimized when theta and S are advected, but to the detriment of heat conservation. The search/development of a more accurate scheme is under way. Implementation: Iskandarani, Chassignet, Bleck, Wallcraft 3. WEB PAGE A WEB page for the consortium is now in place at http://panoramix.rsmas.miami.edu/hycom_nopp/ Everybody is encouraged to provide feedbacks to Ed Ryan (eryan@rsmas.miami.edu). Each participant is required to provide a summary of their research in a format suitable to be displayed on the web and update it on a regular basis. Ed Ryan will provide pointers on how to do so. Implementation: Ryan, Chassignet, Mariano 4. MODEL CONFIGURATIONS 4.1 GLOBAL HYCOM HYCOM needs to be configured globally: HYCOM was first configured with the classic 2-degree quasi-global domain (closed boundary at 65N) forced with COADS data. A comparison to the MICOM solution was carried out for a 100-year climatological run (see web site). The second configuration will be fully global (see above comment on polar grid). Comparison to the first configuration and to data will be performed for a 50-year run. The third (fourth) climatological integration will be a repeat of the second one, but forced with ECMWF (NCEP) atmospheric climatology. After this spin-up, these experiments will be integrated with 6-hourly forcing from ECMWF (NCEP). In the first experiment, the E-P forcing consists of observed precipitation and computed evaporation. The drift and the necessity to add a small relaxation to observed surface salinity in the follow-up experiments will be evaluated. Implementation and evaluation: Bleck, Thacker, Chassignet, Wallcraft 2.2 NORTH ATLANTIC HYCOM This series of basin experiments will build on the CME, DYNAMO and MICOM experience (1, 1/3, and 1/12 degree grid spacing, respectively). The domain configuration will be from 70N to 28S for the 1/3 and 1/12 degree runs (65N to 15S for the 1 degree CME run) with relaxation to observed monthly values (Levitus) at the boundaries. The CME comparison is well underway. Several 20-year simulations were performed with sigma0, sigma2 and sigma2 + thermobaricity both with HYCOM and MICOM. A paper summarizing these results will be put together shortly (Chassignet, Smith, Halliwell). Two 20-year climatological simulations will be performed with forcing from a) ECMWF and b) NCEP in the 1/3 degree configuration. The ECMWF and NCEP runs will then be integrated for an additional 20 years with daily winds from the 1979-1999 period. Freshwater flux will consist of observed precipitation and computed evaporation + a small relaxation to monthly climatological surface salinity. The model output will be saved daily for the surface values (including mixed layer depth) and every three days for the interior quantities. The 1/12 degree simulation with ECMWF forcing will be initiated in year 2 (to be compared to the companion MICOM run). Analysis: Performance metrics (see year 1 implementation plan) Comparison to previous results Implementation and evaluation: Chassignet, Hurlburt, Hogan, Wallcraft 4.3 MODEL CONFIGURATION FOR DATA ASSIMILATION Reanalysis: 1.4 degree Atlantic and global (Thacker, Esenkov, Halliwell, Bleck) SSH assimilation: The basic configuration will be the 1/3 degree domain forced by the 1979-1999 ECMWF forcing. Expertise for the 1/12 degree will be acquired in a regional Intra-American Sea somain (Townsend). 5. DATA ASSIMILATION METHODOLOGY The methodology is well outlined in the year 1 implementation plan. Reanalysis: Assimilation of XBTs following Thacker and Esenkov (2001), North Atlantic followed by global domain (Thacker, Esenkov) SSH assimilation: Optimal interpolation to be ported to the 1/3 degree North Atlantic (Smedstad, Baraille) ROIF to be ported to the 1/3 degree North Atlantic (Chin, Mariano) ROAF to be ported to the 1/3 degree North Atlantic (Baraille, Chassignet) Data analysis prior to assimilation: Mean sea surface height (Cayula, Smedstad, Chassignet, Mariano) XBT (Rhodes, Mariano, Thacker) SSH (Jacobs, Mariano) Lagrangian "pseudo-Eulerian" velocities (Mariano) Skill assessment of each data assimilation method will use the following data sets: (a) the RSMAS two-day 18km blend of satellite and in-situ SST (b) velocities from near-surface drifters (c) SSH from tide gauges (d) WOCE hydrographic sections (e) mixed-layer depths from hydrography and profiling floats (f) forecast errors.