NZ region ocean modelling with terrain-following and z-level models Mark Hadfield 1 & Graham Rickard, NIWA, NZ 1. Prognostic simulations Climatological-mean.

Slides:



Advertisements
Similar presentations
Numerical simulation of internal tides in the Sicily and Messina Straits Jihene Abdennadher and Moncef Boukthir Institut Preparatoire aux Etudes d’Ingenieur.
Advertisements

Freshwater cylinder test-case. Objectives To compare with Tartinville et al. 94 results of a similar test-case To compare with Tartinville et al. 94 results.
Essentials of Oceanography
Abstract Velocity profiles of Byrd glacier show a transition from a parabolic transverse profile upstream to a plug flow transverse velocity profile. A.
Wind-Driven Circulation in a Stratified Ocean Consider the ocean in several isopycnal layers that can be separated into two groups: Layers that outcrop.
Low-frequency variability in the mid-latitude atmosphere induced by an oceanic thermal front: Application to the North Atlantic Ocean Yizhak Feliks 1,2.
HYCOM and the need for overflow/entrainment parameterizations.
Modeling the M 2 and O 1 Barotropic and Baroclinic Tides in the Gulf of Mexico Using the HYbrid Coordinate Ocean Model (HYCOM) Flavien Gouillon 1 ; B.
Chesapeake Bay Lagrangian Floats Analysis. Motivation Lagrangian float has its advantage in describing waters from different origins. We follow definition.
For the Lesson: Eta Characteristics, Biases, and Usage December 1998 ETA-32 MODEL CHARACTERISTICS.
Interannual Variability in Summer Hydroclimate over North America in CAM2.0 and NSIPP AMIP Simulations By Alfredo Ruiz–Barradas 1, and Sumant Nigam University.
Atmospheric Motion ENVI 1400: Lecture 3.
Indirect Determination of Surface Heat Fluxes in the Northern Adriatic Sea via the Heat Budget R. P. Signell, A. Russo, J. W. Book, S. Carniel, J. Chiggiato,
Chapter 5: Other Major Current Systems
MODULATING FACTORS OF THE CLIMATOLOGICAL VARIABILITY OF THE MEXICAN PACIFIC; MODEL AND DATA. ABSTRACT. Sea Surface Temperature and wind from the Comprehensive.
Ocean Currents.
Lecture 7: The Oceans (1) EarthsClimate_Web_Chapter.pdfEarthsClimate_Web_Chapter.pdf, p
Wave communication of high latitude forcing perturbations over the North Atlantic Vassil Roussenov, Ric Williams & Chris Hughes How changes in the high.
Define Current decreases exponentially with depth. At the same time, its direction changes clockwise with depth (The Ekman spiral). we have,. and At the.
Japan/East Sea Hybrid Coordinate Ocean Model (HYCOM) Patrick J. Hogan and Harley E. Hurlburt Naval Research Laboratory, Code 7323, Stennis Space Center,
Modeling Study of Frontal Variability in Drake Passage Bin Zhang 1 and John M Klinck 2 Model Description Regional Ocean Modeling System (ROMS) Dimensions:
Surface wind stress Approaching sea surface, the geostrophic balance is broken, even for large scales. The major reason is the influences of the winds.
Effects of Ocean-Atmosphere Coupling in a Modeling Study of Coastal Upwelling in the Area of Orographically-Intensified Flow Natalie Perlin, Eric Skyllingstad,
Define Current decreases exponentially with depth and. At the same time, its direction changes clockwise with depth (The Ekman spiral). we have,. and At.
Two research cruises were successfully conducted in 2013 and Shipboard and moored observations show that: at first glance no significant decadal.
Progress on the Modification of Offshore Boundary Conditions for Full-Bay ROMS Simulations Have shown previously that ROMS simulations capture tides well,
RA-228 AND RA-226 FROFILES FROM THE NORTHERN SOUTH CHINA SEA Hsiu-Chuan Lin, Yu-Chia Chung and Chi-Ju Lin Institute of Marine Geology and Chemistry, National.
Rossby Wave Two-layer model with rigid lid η=0, p s ≠0 The pressures for the upper and lower layers are The perturbations are 
CCSM Simulations w/CORE Forcing Some preliminary results and a discussion of dataset issues Marika Holland With much input from Bill Large Steve Yeager.
Nested models of the Southland Current Mark Hadfield National Institute of Water & Atmospheric Research, Wellington, NZ.
1DMS/USM, 2SERF, 3NRL/SSC, 4COAPS/FSU
Imposed versus Dynamically Modeled Sea Ice: A ROMS study of the effects on polynyas and waters masses in the Ross Sea John M. Klinck, Y. Sinan Hüsrevoglu.
Air Pressure and Winds. Atmospheric Pressure  What causes air pressure to change in the horizontal?  Why does the air pressure change at the surface?
The Influence of the Indonesian Throughflow on the Eastern Pacific Biogeochimical Conditions Fig.1 The last year of the two runs is used to force offline.
The climate and climate variability of the wind power resource in the Great Lakes region of the United States Sharon Zhong 1 *, Xiuping Li 1, Xindi Bian.
2010 AMS Effect of changes in GCM resolution on the connection between summertime precipitation, moisture flux, and the position of the Bermuda High Laura.
First results from the isopycnic ocean carbon cycle model HAMOCC & MICOM/BCM Karen Assmann, Christoph Heinze, Mats Bentsen, Helge Drange Bjerknes Centre.
Monitoring Heat Transport Changes using Expendable Bathythermographs Molly Baringer and Silvia Garzoli NOAA, AOML What are time/space scales of climate.
The Southern Ocean geography, principal fronts, and oceanographic zones (see Table 13.1). The Subtropical Front (STF) is the oceanographic northern boundary.
Hadley Centre Evaluating modelled and observed trends and variability in ocean heat content Jonathan Gregory 1,2, Helene Banks 1, Peter Stott 1, Jason.
An example of vertical profiles of temperature, salinity and density.
Measuring the South Atlantic MOC – in the OCCAM ocean model Povl AbrahamsenJoel Hirschi Emily ShuckburghElaine McDonagh Mike MeredithBob Marsh British.
Ekman pumping Integrating the continuity equation through the layer:. Assume and let, we have is transport into or out of the bottom of the Ekman layer.
A Synthetic Drifter Analysis of Upper-Limb Meridional Overturning Circulation Interior Ocean Pathways in the Tropical/Subtropical Atlantic George Halliwell,
Conservation of Salt: Conservation of Heat: Equation of State: Conservation of Mass or Continuity: Equations that allow a quantitative look at the OCEAN.
Geopotential and isobaric surfaces
Mixing and Entrainment in the Orkney Passage Judy Twedt University of Washington Dept. of Physics NOAA, Geophysical Fluid Dynamics Lab Dr. Sonya Legg Dr.
Effect of the Gulf Stream on Winter Extratropical Cyclones Jill Nelson* and Ruoying He Marine, Earth, and Atmospheric Sciences, North Carolina State University,
NUMERICAL STUDY OF THE MEDITERRANEAN OUTFLOW WITH A SIMPLIFIED TOPOGRAPHY Sergio Ramírez-Garrido, Jordi Solé, Antonio García-Olivares, Josep L. Pelegrí.
Tropical Atlantic Biases in CCSM4 Semyon A. Grodsky 1, James A. Carton 1, Sumant Nigam 1, and Yuko M. Okumura 2 1 Department of Atmospheric and Oceanic.
Wind-SST Coupling in the Coastal Upwelling --- An Empirical Numerical Simulation X. Jin, C. Dong, and J. C. McWilliams (IGPP/UCLA) D. B. Chelton (COAS/OSU)
Coastal Oceanography Outline Global coastal ocean Dynamics Western boundary current systems Eastern boundary current systems Polar ocean boundaries Semi-enclosed.
Nansen Environmental and Remote Sensing Center Modifications of the MICOM version used in the Bergen Climate Model Mats Bentsen and Helge Drange Nansen.
MICHAEL A. ALEXANDER, ILEANA BLADE, MATTHEW NEWMAN, JOHN R. LANZANTE AND NGAR-CHEUNG LAU, JAMES D. SCOTT Mike Groenke (Atmospheric Sciences Major)
Matthew J. Hoffman CEAFM/Burgers Symposium May 8, 2009 Johns Hopkins University Courtesy NOAA/AVHRR Courtesy NASA Earth Observatory.
15 Annual AOMIP Meeting. WHOI, 1- 4 November 2011 Numerical modeling of the Atlantic Water distribution in the upper Arctic Ocean: Sensitivity studies.
The effect of tides on the hydrophysical fields in the NEMO-shelf Arctic Ocean model. Maria Luneva National Oceanography Centre, Liverpool 2011 AOMIP meeting.
Numerical modeling of Atlantic and Pacific waters dynamics Elena Golubeva Institute of Computational Mathematics and Mathematical Geophysics Siberian Branch.
Seasonal Variations of MOC in the South Atlantic from Observations and Numerical Models Shenfu Dong CIMAS, University of Miami, and NOAA/AOML Coauthors:
Global Winds and Ocean Currents Ocean Currents are created by global winds. Ocean Currents are created by global winds. Global winds are created by the.
Coupling ROMS and CSIM in the Okhotsk Sea Rebecca Zanzig University of Washington November 7, 2006.
Climate and Global Dynamics Laboratory, NCAR
Section 1: Ocean Currents
OCEAN RESPONSE TO AIR-SEA FLUXES Oceanic and atmospheric mixed
Air Masses and fronts An air mass is a large body of air that has similar temperature and moisture properties throughout. A front is defined as the transition.
Shelf-basin exchange in the Western Arctic Ocean
Mark A. Bourassa and Qi Shi
Chapter 10 Wind: Global Systems.
Time mean MSLP bias (mbar) in CCSM its atmospheric component (CAM/AMIP). CCSM4 MSLP bias is weaker than CCSM3 bias in the northern subtropical maxima.
World Geography 3202 Unit 2 Climate Patterns.
Presentation transcript:

NZ region ocean modelling with terrain-following and z-level models Mark Hadfield 1 & Graham Rickard, NIWA, NZ 1. Prognostic simulations Climatological-mean simulations have been set up for a domain around New Zealand with the following models: MOMA: z-level, B-grid, rigid-lid ROMS: terrain-following, C-grid, free-surface They use the following datasets to provide initial and boundary conditions: Temperature and salinity from World Ocean Atlas 2001 Depth-integrated currents from HADCEM, a coupled ocean- atmosphere climate model run at the UK Met Office with an ocean resolution of 1/3° Surface stress from NCEP Reanalysis or HADCEM Each model was started from rest with climatological temperature and salinity. At the boundary the depth-integrated normal velocity was clamped at HADCEM values. Next to the boundary there was a buffer zone ~10 cells wide in which model fields were nudged toward climatology. The 3-D velocity climatology for the buffer zone was derived from diagnostic simulations (below). Horizontal resolution was 1/6°. ROMS had 20 layers vertically and MOMA had Prognostic simulations Climatological-mean simulations have been set up for a domain around New Zealand with the following models: MOMA: z-level, B-grid, rigid-lid ROMS: terrain-following, C-grid, free-surface They use the following datasets to provide initial and boundary conditions: Temperature and salinity from World Ocean Atlas 2001 Depth-integrated currents from HADCEM, a coupled ocean- atmosphere climate model run at the UK Met Office with an ocean resolution of 1/3° Surface stress from NCEP Reanalysis or HADCEM Each model was started from rest with climatological temperature and salinity. At the boundary the depth-integrated normal velocity was clamped at HADCEM values. Next to the boundary there was a buffer zone ~10 cells wide in which model fields were nudged toward climatology. The 3-D velocity climatology for the buffer zone was derived from diagnostic simulations (below). Horizontal resolution was 1/6°. ROMS had 20 layers vertically and MOMA had 36. Figure 1:Barotropic stream function from prognostic simulations 1 Mark Hadfield, NIWA, Private Bag , Kilbirnie, Wellington, NZ 2. A discrepancy Figure 1 (top centre) shows the barotropic stream function from the two models after 4–5 years. The two models differ substantially in their representation of the ACC. This is a strong eastward current (~ 170 Sv in the HADCEM climatology) that enters and leaves the domain south of 50° S. In the MOMA simulation its northern edge skirts the New Zealand continental shelf and reaches its furthest north at 178° W, 47 ° S. In the ROMS simulation it extends much further north, finally detaching from the coast at 37 ° S. From model temperature and salinity fields (not shown) it is clear that the path of the ACC in the ROMS simulation is wrong. The MOMA simulation agrees with observations much better. A similar difference between terrain-following and z-level models can be seen in the South Atlantic models of Penduff et al. (2001), where the terrain-following model traps a much larger portion of the ACC against the eastern coast of South America than the z-level model. The MOMA and ROMS simulations diverge quite soon after initialisation. The ROMS simulation progressively pushes cold water northward around the eastern edge of the New Zealand land mass. This process becomes apparent at ~50 days, though it is not complete until ~1000 days. 2. A discrepancy Figure 1 (top centre) shows the barotropic stream function from the two models after 4–5 years. The two models differ substantially in their representation of the ACC. This is a strong eastward current (~ 170 Sv in the HADCEM climatology) that enters and leaves the domain south of 50° S. In the MOMA simulation its northern edge skirts the New Zealand continental shelf and reaches its furthest north at 178° W, 47 ° S. In the ROMS simulation it extends much further north, finally detaching from the coast at 37 ° S. From model temperature and salinity fields (not shown) it is clear that the path of the ACC in the ROMS simulation is wrong. The MOMA simulation agrees with observations much better. A similar difference between terrain-following and z-level models can be seen in the South Atlantic models of Penduff et al. (2001), where the terrain-following model traps a much larger portion of the ACC against the eastern coast of South America than the z-level model. The MOMA and ROMS simulations diverge quite soon after initialisation. The ROMS simulation progressively pushes cold water northward around the eastern edge of the New Zealand land mass. This process becomes apparent at ~50 days, though it is not complete until ~1000 days. 3. Diagnostic simulations To investigate what controls currents early in the simulations, before the density field has been substantially modified, we compared diagnostic simulations with the two models. In these simulations Temperature and density fields are held constant at their climatological values Velocity advection is disabled The latter change eliminates amplification of instabilities at the boundaries and it makes the system linear (and hence easier to analyse). Figure 2 shows the barotropic stream function from diagnostic simulations with the two models. Differences are as large as ± 40 Sv. East of New Zealand the difference is negative, so MOMA transports more water down the eastern side of the New Zealand land mass. Southeast of New Zealand the difference is positive, so MOMA transports more water around the southern tip of the land mass. 3. Diagnostic simulations To investigate what controls currents early in the simulations, before the density field has been substantially modified, we compared diagnostic simulations with the two models. In these simulations Temperature and density fields are held constant at their climatological values Velocity advection is disabled The latter change eliminates amplification of instabilities at the boundaries and it makes the system linear (and hence easier to analyse). Figure 2 shows the barotropic stream function from diagnostic simulations with the two models. Differences are as large as ± 40 Sv. East of New Zealand the difference is negative, so MOMA transports more water down the eastern side of the New Zealand land mass. Southeast of New Zealand the difference is positive, so MOMA transports more water around the southern tip of the land mass. Figure 2: Barotropic stream function from diagnostic simulations: a) comparison b) difference a) 4. Why are they different? We suspect that the difference between MOMA and ROMS prognostic simulations is related to the much stronger transport of water in MOMA down the eastern side of the land mass. We have taken advantage of the linearity of the diagnostic system to distinguish the effects of three forcings, each estimated in a different simulation: 1. Boundary transport: depth-integrated current is specified at the boundary; the fluid is horizontally homogenous and the surface stress is zero. 2. Baroclinicity: depth-integrated current at the boundary is zero; variations in density is retained and the surface stress is zero. 3. Surface stress: depth-integrated current at the boundary is zero; fluid is horizontally homogeneous and the surface stress is applied. This comparison (not shown) establishes that the difference between the barotropic stream function in MOMA and ROMS (Figure 2b) is largely accounted for by differences in simulations with baroclinicity only. 4. Why are they different? We suspect that the difference between MOMA and ROMS prognostic simulations is related to the much stronger transport of water in MOMA down the eastern side of the land mass. We have taken advantage of the linearity of the diagnostic system to distinguish the effects of three forcings, each estimated in a different simulation: 1. Boundary transport: depth-integrated current is specified at the boundary; the fluid is horizontally homogenous and the surface stress is zero. 2. Baroclinicity: depth-integrated current at the boundary is zero; variations in density is retained and the surface stress is zero. 3. Surface stress: depth-integrated current at the boundary is zero; fluid is horizontally homogeneous and the surface stress is applied. This comparison (not shown) establishes that the difference between the barotropic stream function in MOMA and ROMS (Figure 2b) is largely accounted for by differences in simulations with baroclinicity only. b) 5. Still wondering why Significant differences have been found between the 2 models in a relatively simple system, involving currents generated by baroclinic pressure gradients interacting with bathymetry. How do these differences arise? Each model has well-known weaknesses that might be responsible for the differences: Terrain-following models have pressure-gradient errors due to their sloping coordinate surfaces. Because of these errors it is necessary to run these models with smoothed bathymetry. Z-level models represent the bottom as a series of horizontal and vertical segments. In a B-grid, z-level model each of the vertical segments imposes a no-slip condition on the horizontal velocities. We set up a domain covering the southern portion of the NZ region (same east, west & south boundaries as above, northern boundary at 48° S). We ran various simulations with ROMS and MOMA, looking at the strength of the barotropic transport in the Subantarctic Front. To summarise: In the control simulations, total transport along the Subantarctic Slope is 96 Sv with MOMA and 45 Sv with ROMS. If the horizontal grid spacing in ROMS is reduced, without changing the steepness of the bathymetry, the transport is unchanged. This suggests pressure-gradient errors are not an issue. If we make the bathymetry in ROMS steeper (which may require smaller grid spacing) the transport is reduced somewhat. The highest resolution simulation with the steepest bathymetry gives a transport of 35 Sv. In ROMS, results are not sensitive to horizontal viscosity or bottom- drag coefficient (but if we increase the latter much the model crashes). If MOMA grid spacing is reduced, the transport drops sharply. The highest-resolution simulation with MOMA gives a transport of 60 Sv. Increasing the number of levels in MOMA has no effect. This suggests to us that the culprit is the poor representation of bathymetry due to MOMA’s B-grid z-level layout. 5. Still wondering why Significant differences have been found between the 2 models in a relatively simple system, involving currents generated by baroclinic pressure gradients interacting with bathymetry. How do these differences arise? Each model has well-known weaknesses that might be responsible for the differences: Terrain-following models have pressure-gradient errors due to their sloping coordinate surfaces. Because of these errors it is necessary to run these models with smoothed bathymetry. Z-level models represent the bottom as a series of horizontal and vertical segments. In a B-grid, z-level model each of the vertical segments imposes a no-slip condition on the horizontal velocities. We set up a domain covering the southern portion of the NZ region (same east, west & south boundaries as above, northern boundary at 48° S). We ran various simulations with ROMS and MOMA, looking at the strength of the barotropic transport in the Subantarctic Front. To summarise: In the control simulations, total transport along the Subantarctic Slope is 96 Sv with MOMA and 45 Sv with ROMS. If the horizontal grid spacing in ROMS is reduced, without changing the steepness of the bathymetry, the transport is unchanged. This suggests pressure-gradient errors are not an issue. If we make the bathymetry in ROMS steeper (which may require smaller grid spacing) the transport is reduced somewhat. The highest resolution simulation with the steepest bathymetry gives a transport of 35 Sv. In ROMS, results are not sensitive to horizontal viscosity or bottom- drag coefficient (but if we increase the latter much the model crashes). If MOMA grid spacing is reduced, the transport drops sharply. The highest-resolution simulation with MOMA gives a transport of 60 Sv. Increasing the number of levels in MOMA has no effect. This suggests to us that the culprit is the poor representation of bathymetry due to MOMA’s B-grid z-level layout. 6. Provisional Conclusions The MOMA prognostic simulations are clearly better. The differences in barotropic transport between ROMS and MOMA diagnostic simulations might be caused by MOMA having a poorer representation of bathymetry. If so, this error in MOMA is correcting for errors elsewhere in the system, possibly a too– vigorous ACC. We will analyse the diagnostic simulations further to elucidate the differences. 6. Provisional Conclusions The MOMA prognostic simulations are clearly better. The differences in barotropic transport between ROMS and MOMA diagnostic simulations might be caused by MOMA having a poorer representation of bathymetry. If so, this error in MOMA is correcting for errors elsewhere in the system, possibly a too– vigorous ACC. We will analyse the diagnostic simulations further to elucidate the differences.