|
http://www.atmos.washington.edu/academics/classes/2013Q1/380/HW10.html Due Friday Mar 20 |
|
You will analyze Sea Surface Temperature (SST) in a long "control" run with the Community Climate System Model (CCSM4). This output was made for the 2013 IPCC. We call it a control because it has only period forcing from the sun, greenhouse gases and aerosols. It is run for a long time so we can understand intrinsic (or natural) climate variability. The greenhouse gases and aerosols are intended to be appropriate for preindustrial (1850s) times, before humans had started altering them substantially. The model has an active ocean, land, atmosphere, and sea ice and has approximately 1 deg resolution in every component. I could have provided the figures for you, but then you would not have the joy of waiting for the data to load while anticipating the result. These scripts are loading 100 to 500 years of data. It is slow, but worth it! Copy a set of scripts to whatever directory you choose. cp /home/disk/p/atms380/scripts/ex10* . matlab & To Turn In: Begin your write-up with a one-paragraph introduction to the exercise. A. Have a look at this figure on the web that shows a ~16 year period of SST in the tropical Pacific in a Hovmöller diagram. (Note that I spelled Hovmöller wrong on the script name for exercise 6. This is the correct spelling.) Time increases downward in these figures, unlike ours where time increases upward. This is a matter of taste. The data are observations from the TOGA TOA array of buoy in the Pacific. They are binned into 5-day means. The panel on the left has a distinct annual cycle in the eastern Pacific with a lot of interannual variability as well. Hence, in the eastern Pacific in most years there is a cold period and a warm period. Notice the intense warming in the eastern Pacific that occured in 1997. It is easy to see in either panel. There is virtually no cold season in 1997 in the eastern Pacific. This was a strong El Nino year. 1988 was a strong La Nina year. The SST anomaly in the right panel is computed by taking the SST at each point in time and removing the average across the 16 years for that 5-day chunk. It should eliminate the annual cycle nearly completely, so there is no cold-warm oscillation on a yearly period in the Eastern Pacific. Run the first MATLAB script ex10_a.m to produce a pair of Hovmöller diagrams from the CCSM4 data. You can reduce the number of years on the y-axis for better comparing with the observations by typing, for example, "figure(1); subplot(1,2,1); ylim([0 20])" in the MATLAB command window. Look at the full 100 years and a few 20-yr excerpts. To Turn In: First compare the two figures for CCSM4. How does the eastern Pacific and the Atlantic compare in terms of the relative amount of annually periodic and interannual variability. For example, estimate by choosing a longitude in the eastern Pacific and estimate the range of temperature in the left panel (with annual cycle and interannual variability) and on the right panel (with only interannual variability). Do the same for the Atlantic at a longitude in the middle. Now compare the Pacific in CCSM4 with TOGA TAO observations. How does the character of the variability compare in terms of magnitude and frequency of large el nino/la nina events? Which one is more regular/irregular? B. Run ex10_b.m to compute the standard deviation of the SST at all grid cells. The anomaly is computed for each month by taking each January and subtracting the average of all Januaries, etc. The standard deviation is computed for the monthly anomalies. Hence the standard deviation does not include variability for the annually periodic cycle. The first thing to notice is the very weird looking grid. This is the POP ocean model grid at 1 deg. It has the grid's north pole yanked into Greenland. Greenland is stretched across the top edge. This is desirable because the real north pole does not have converging meridians. I rather like seeing the Arctic Ocean in one basin too. Too bad the real Earth is so distorted. To Turn In: Discuss where the standard deviation is high (above ~3 deg C). Do not dwell on the very high values (above ~5 deg). Try to relate the high variability regions to where you think the winds are fastest or other mechanisms. Where is the variability largest on the equator? Why is there so little variability in SST at the highest latitudes (in the real world, not on the screwy grid)? C. Run ex10_c.m to compute the standard deviation again but with the SST anomalies averaged first into chunks of N years before computing the standard deviations. N should be equal to 1 when you begin. Edit the script and run again for N=2, 3, 5, 10, 25. To Turn In: Notice how pronounced the variability is in the tropical Pacific for N<5 now with the month to month variability eliminated in this series. How does this compare with the estimate of the return period of large amplitude el nino and la nina events that you found in part A? Where are the regions with the lowest frequency variability (high standard deviation at N=25 year averaging)? Is the map at N=25 much different than N=5? In class I said that places with a lot of low-frequency variability (at N>5) also tend to have high variability in general (say at N=1). Is this a good rule of thumb? I'm happy to discuss any of these results further in class or via email. Good luck studying for your finals!
|
| Return to Homepage |
