Oceanography 569 Oceanographic Data Analysis Laboratory Kathie Kelly

Oceanography 569 Oceanographic Data Analysis Laboratory Kathie Kelly Applied Physics Laboratory 515 Ben Hall IR Bldg class web site: faculty. washington. edu/kellyapl/classes/ocean 569_ 2014/

Applying Analysis Tools practice exercise for project

Sea Level Variability in the Eastern Mediterranean Sea Data: % sshm - weekly sea surface height anomaly (SSH) % Qnet - daily net surface heat flux (OAFlux) % alf _rho_cp - alpha/(rho*cp) % where alpha is the coefficient of thermal expansion % cp is the specific heat of sea water, rho is density of sea water % txqs, tyqs - daily Quik. SCAT wind stress components % coastline - lon, lat for graphics Model: % thermal expansion from seasonal heating (vertical integral of temperature, converted to SSH using the coefficient of thermal expansion) % % d(SSH)/dt = alf*Qnet/(rho*c_p) % Other candidate (for statistical comparison): % wind stress

Sea Level Variability in the Eastern Mediterranean Sea Look at time-longitude to determine obvious signals Seasonal cycle apparent No evidence of RW propagation

Sea Level Variability in the Eastern Mediterranean Sea Run the heating model and compare with observed SSH

Sea Level Variability in the Eastern Mediterranean Sea Run the heating model and compare with observed SSH One outcome: Assess skill

Sea Level Variability in the Eastern Mediterranean Sea Compare winds with (nonseaonal) SSH Match integral time scales of ssh and winds (approximately)

Sea Level Variability in the Eastern Mediterranean Sea Compare winds with (nonseaonal) SSH What analysis tools do we have for comparing data without a model?

Sea Level Variability in the Eastern Mediterranean Sea Compare winds with (nonseaonal) SSH 1 and stress 1 correlated with zero lag: Northward winds high sea level 2) SSH 2 and stress 2 marginally correlated 3) Stress 1 leads stress 2: Northward winds lead cyclone (eastward propagation of storms) Overall: heating accounts for much of seasonal cycle Rest is wind-driven, but mechanism unclear

Applying Analysis Tools practice exercise for project

Projects Chose either project 1 or 2 • Use analysis tools from class (including models for the project) • Write up results in a slide presentation (approximately 30 min) • Do NOT include your Matlab code – think of this as a seminar! Include in presentation: • What is the science question? • What is your analysis plan? Why? • What special considerations are there? • What tools did you use? models, statistical procedures, etc • Show metrics: skill, significance tests, etc • What did you conclude? Why? • Show data examples, analysis outcomes, etc

Project 1: What Causes SST Variability in the Gulf Stream? Data: SST, climatological MLD, Gulf Stream path, winds One dimensional mixed layer model: d. T/dt = (Q-qrad)/(rho*cp*h) where qrad is the radiative flux leakage for the given climatological MLD h 1) Climatological analysis: • Do you see climatological errors? (errors that recur each year) • Is there a pattern to the errors? • How would you correct for an error in climatological Q/h?

Project 1: What Causes SST Variability in the Gulf Stream? 2) Analyze residual: d. T/dt(model) - d. T/dt(obs) What are the errors in the heating model? Other candidates: • Gulf Stream path changes • Winds How do each of these affect SST?

Project 1: What Causes SST Variability in the Gulf Stream? What are the missing terms in this mixed layer model? Without writing a new model can you relate these candidate variables to the residual? (statistical relationships) • What are some ways that wind can effect d. T/dt? (Note: changes in wind speed are accounted for in Q) • How does the change in Gulf Stream path affect d. T/dt?

Project 2: What causes SSH anomalies in tropical Atlantic? Candidates: surface heating (OAFlux/ISCCP net surface flux) wind-forced Rossby waves (Quik. SCAT wind stress curl)

Project 2: What causes SSH anomalies in tropical Atlantic? response to surface heating: dη/dt = Qnet*α/(rho*c_p) wind-forced Rossby waves: dη/dt = c*dη/dx –r*η -C*curl where c is the phase speed of the Rossby waves and r is a damping factor (eddy diffusion) RW model needs an eastern boundary condition (because waves travel westward), so use SSH interpolated to the model t grid

Project 2: What causes SSH anomalies in tropical Atlantic? More information: RW model does not include thermosteric sea level, so remove heating response from SSH first RW model has some fungible parameters, g’, r (damping), and phase speed, c. Estimate phase speed, but adjust to reduce errors. Reduced gravity (g’) can be adjusted also. The damping factor (alf/r) has been optimized so no need to change it.