Upload
vivian-potter
View
218
Download
1
Tags:
Embed Size (px)
Citation preview
The Gravity Current Entrainment Climate Process Team
Sonya Legg
Princeton University, NOAA-GFDL
Representing overflows in climate models
•Ocean models need to represent overflow processes correctly to get dense water properties right.•Z-coordinate models have difficulties getting water downslope without excessive mixing at coarse climate resolution.•Isopycnal models need to parameterize diapycnal mixing.•Narrow straits are an issue for all coarse resolution models.
Shear instability, entrainment
detrainment
Geostrophic eddies
x
z
y
Downslope descent
Bottom friction
Physical processes in overflows
Hydraulic control
Dense water masses formed in marginal seas enter open ocean through overflows, e.g. Denmark Straits, Faroe Bank Channel, Antarctic slope overflows.
Mixing and transport in overflows determines properties of ocean bottom waters, e.g. NADW, AABW.
How do we improve ocean model overflow representation?
Observations Laboratory and numerical process studies
Regional simulations Simulations in idealized configurations
Global climate simulations
Improved parameterizationscomparecalibrate
evaluate
The Gravity Current Entrainment Climate Process Team: a multi-institutional collaboration between those studying processes in detail and
those building and running climate models. A US CLIVAR project funded by NSF and NOAA, 2003-2008.
CPT-GCE participants: Models, Numerical and Lab process studies, Observations
Core PIsStephen Griffies (GFDL), Robert Hallberg (GFDL),William Large (NCAR),
Gokhan Danabasoglu (NCAR), Peter Gent (NCAR), Jim Price (WHOI), Jiayan Yang (WHOI), Sonya Legg (WHOI/Princeton), Hartmut Peters
(Miami), Eric Chassignet (Miami), Tamay Ozgokmen (Miami), Tal Ezer
(Princeton), Arnold Gordon (Columbia), Paul Schopf (GMU)
Postdocs/ResearchersUlrike Riemenschneider (WHOI), Laura Jackson (GFDL), Yeon Chang
(Miami), Wanli Wu (NCAR)
For more details see http://www.cpt-gce.org
CPT Activities, years 1-3
Annual workshops•Put observationalists, process study modelers and GCM developers in the same room•Opportunity for team to provide input/feedback to results/plans of team members•Opportunity to share results/plans with invited members of wider community•Forum for discussion of joint activities/plans•Starting point for continuing one-on-one collaborationsTable of observations
Compiled by team observationalists, lead by Arnold Gordon.Quick reference for:•Parameters needed in GCM overflow representations•Intercomparison of observed overflow characteristics•Comparison with regional and climate simulations.
Publications and presentations•Numerous individual publications•2 group publications in USCLIVAR Variations•1 synthesis article in preparation for BAMS•Numerous team posters/presentations
Summary of principal achievements of first 3 years of CPT
• Parameterization of frictional bottom boundary mixing in overflows: implementation in Hallberg Isopycnal Model has made a 1st order difference to its credibility as a climate ocean model at 1 degree.
• The Marginal Sea Boundary Condition has been implemented in climate models, NCAR POP and HYCOM: Med Sea Outflow can now be represented credibly at coarse resolution.
• New and improved shear mixing parameterizations have been developed.
• Partially open barrier method: identified as a promising method for representing narrow straits.
• Reduction in spurious mixing in z-coordinate models: several promising solutions are under investigation.
All of these developments have involved input from observations, idealized and regional numerical simulations, in addition to GCMs.
Observed profiles from Red Sea plume from RedSOX (Hartmut Peters)
Interior Ri# + Drag Mixing
Interior Ri# Mixing
Only
Focus on Frictional bottom boundary mixingLegg et al., 2005
HIM10 km
HIM10 km
Resolved mixing (LES)
MITgcm500m x 30m
Well-mixed Bottom Boundary LayerMixing driven by bottom stresses
Actively mixingInterfacial LayerShear Ri# Param.appropriate here.
With thick plumes both interfacial shear mixing and drag-induced near bottom mixing are needed. (Legg, Hallberg and Girton,2006).
Impact of frictional bottom boundary mixing on GCM results
Mediterranean outflow salinity: comparison between 1 degree Hallberg Isopycnal Model and observed climatology.
Spurious bottom plumeNew parameterization eliminates spurious bottom plume
Surface
Heat and salt conservation
-z
ATLANTICARCTIC
QS = g’ (hS)2 / 2f
QS
2100m
945m
686m
BOTTOM TOPOGRAPHY
QE
QP
ρS
(Source density based on 12 grid points)
ρ0
(Ent. density based on 12 grid points)
g’ = g (ρS-ρO)/ρref
QP = QSFr2/3
QP = QS + QE
+
Focus on the Marginal Sea Boundary ConditionParameterizes both narrow straits and entrainment, suitable for both isopycnal and z-coord models. Developed by Yang and Price, implemented in NCAR POP for Med Sea and Faroe Bank Channel, and in HYCOM for Med Sea.
NCAR MSBC implementation for FBC
Team interactions involved in MSBC implementation
• Parameterization originally developed by Yang and Price, who continue to consult on implementation issues.
• Numerous parameters must be specified: values are taken from Table of observations, and/or from regional simulations (Ulrike Riemenschneider, FBC).
• Whole team provides input on scenarios for examining impact in GCMs.
• Yang and Price continue to improve/update MSBC, with new developments to be ported to GCMs when ready.
Results: Impact of MSBC in GCM simulations
Salinity at 1100m depth: comparison between climatology and NCAR 3 degree simulations for Med outflow. MSBC leads to credible Med salt tongue.
CPT plans for years 4 and 5 include:
• Complete implementation of MSBC in NCAR POP and HYCOM, for all climatologically important overflows.
• Complete new shear-driven mixing parameterizations, implement in HIM, MOM4 and other models.
• Complete implementation of partially open barriers in global models and compare with other methods of representing narrow straits.
• Focus on impact of parameterizations on Nordic overflows and Antarctic overflows.
• Examine impact of parameterizations on global climate simulations, including climate change scenarios, Greenland ice-melt scenarios.
Extra slides…..
Results:Table of observations
Faroe Bank Denmark St Ross Sea Weddell Sea Red Sea Med SeaJ .Price J . Girton A. Gordon A.Gordon H. Peters J .Price
Transport, SvSource 1.8 2.9 0.6 1.0 0.3 0.8Product 3.3 5.2 2 5 0.55 2.3
Thermobaric Parameter very small small significant significant very small very small
Froude number 1 0.3 - 1.2 0.9 to 1.1 1 0.3-0.9 1Descent radians of gravity current 0.15 0.2-0.6 0.6 0.6 1/2 PI 0.2Entrainment Rate [We, mm/ sec] 0.8 2 to 4 0.1Entrainment Rate [We/ u] 5 x 10-4 1 x 10-3 6 x 10-3 2 x 10-4 5 to 20 x 10-4
Deformation Radius [Rd] Km 30 1.7 - 8 7 7 40 100
Bottom slope near source 13 x 10-3 0.01 - 0.03 0.75 12 x 10-3
transit time, source to product 3-5 days 4-7 days 1 day 1 day 3.5 3-5 daystidal current [characteristic amp] 0.2 0.1 0.3 0.2 0.8 / a few cm/s 0.1
Purpose: Quick reference for:•Parameters needed in GCM overflow representations•Intercomparison of observed overflow characteristics•Comparison with regional and climate simulations. Effort led by Arnold Gordon.
Abbreviated version: for full version see http://www.cpt-gce.org/Table_of_observations.htm
Results: Improved parameterizations of shear-driven mixing.
cRi
RiEE 10
with E0= 0.2 and Ric=0.25 (Xu et al, 2006)
Downstream evolution of entrainment in HYCOM simulations with different entrainment parameterizations compared to nonhydrostatic (Nek) benchmark calculations.
Calibration of entrainment parameterization by comparison with nonhydrostatic benchmark calculations (Miami CPT members)
Calibrated parameterization is validated by comparison between regional simulations and observations. Example: Mediterranean outflow simulated in 0.08 degree HYCOM regional implementation (Xu et al, 2006).
Results: Improved parameterizations of shear-
driven mixing II
)(222
2
RiSFLz B
where S is the vertical shear of the resolved horizontal velocity NQLB /2/1is the buoyancy length scale (the scale of the overturns), N is the buoyancy frequency, and Q is the turbulent kinetic energy, found from an energy budget.
F(Ri) is a function of shear Richardson number Ri such as )/1(
)/1()( 0
c
c
RiRi
RiRiFRiF
cRiRi
Jackson and Hallberg (GFDL) are developing a new parameterization of shear-driven mixing for both isopycnal and z-coord models, with a parameterized diffusivity of the form:
Initial comparison with DNS of shear layers and jets looks promising. Validation with LES is continuing.
DNS data
ET parameterisation
New parameterisation
F0 = 0.14, cN = 0.41, cS = 0.10, = 0.6
F0 = 0.11, cN = 0.20, cS = 0.10, = 0.7
F0 = 0.12, cN = 1.87, cS = 0.10, = 0.9
Results: Impact of shear-driven mixing parameterization on climate
In coupled simulations using Hallberg Isopycnal model, with entrainment in Nordic overflows SSTs are warmer near entrainment site, and cooler to south, due to change in location of Gulf Stream induced by DWBC transport changes.
Results: Treatment of narrow straits
Representation of channels below grid scale by thin walls and partially open barriers is under development by Adcroft and Hallberg (GFDL).
This technique improves Mediterranean outflow simulations in Hallberg Isopycnal model at 1 degree (110km) resolution by reducing Gibraltar width to 12km.
Salinity in HIM global simulations and observations
Regional simulations of Red Sea (Chang et al, 2006), Mediterranean (Xu et al, 2006) and Faroe Bank Channel (Riemenschneider and Legg, 2006) show results are more sensitive to resolution of topography than to mixing parameterization.
Results: reducing spurious mixing in z-coordinate models
Idealized simulations (Legg et al, 2006) and regional simulations (Riemenschneider and Legg, 2006) show that z-coordinate models produce too much numerical mixing, principally due to advection schemes.
Efforts to reduce spurious mixing in MOM4 (Griffies, GFDL) include:•New, less diffusive advection schemes•Improved sigma-diffusion schemes (Beckmann and Doescher, 1997).
Difference between CM2.1 simulations with and without sigma-diffusion at 100m in Med outflow region.
Promising recent development includes non-local horizontal communication in sigma-diffusion.