Precipitation and Southerly Wind

To show the relationship between precipitation and southerly winds over our domain of study, I present the following figure:

I first noticed that the occurrence of southerly winds on days with precipitation (>=0.1mm/day) is amplified around topographic features. This makes physical sense, however, because these features will work to channel winds more northward as a result of up-slope orographic lifting.

Also important to note is the seemingly lack of a positive relationship between these variables along the western portions of our domain. This could be a sign of the main, north-to-south gulf surge corridor being further east, or simply an artifact of my data processing and calculations. Both are possibilities, but based on my knowledge thus far of the characteristics associated with surges, I believe some confidence can be given to the first assumption.

-jamie

Your EOF is Showing...

Before I get to the EOF-talk, here is a plot of climatology vs the 21-day moving mean for a pixel, chosen at random. I think this looks pretty good, or at least close to what I expected to see, i.e., close to climo but with some deviations.

-------------------

Now on to the EOF fun

EOF Procedures, same for IWVF and IWV, thus far:

(1) 28yr mean for each day for each pixel
(2) Calculated 21-day moving mean for each day from 1 July to 31 August, a.k.a. "Tha Meat"
(3) Perform EOF calculations on Step (2) data

Here are the plots of pattern variance for IWVF (top) and IWV (bottom).

These plots show that most of the variance is associated with the first pattern. This is a good sign, I believe, since it shows that we have a dominate pattern in both fields. Next, will be the contour plot of this first pattern for IWVF (top) and IWV (bottom).


These plots are hot off the presses, so I haven't come to any profound understanding of their deeper meaning as of yet. Any thoughts?

-jamie

Quick Aside with 10m Winds

After reading over Simona Bordoni again, I felt compelled to take a quick look at 10m winds over our domain.

I first plotted a Hovmoller of only V component winds averaged over -116 to -107W (Lat = 24 to 26N) for the whole of 1983. After looking at this data, I noticed a lot of noise in the northern portions of the graph. As a means to loose that noise and help pick-out patterns associated with surges I restricted the domain to 24 to roughly 33N (longitude held same). An example of this domain is shown below.
Looking at the first Hovmoller plot for this new domain, I noticed that the prevailing southerly winds were not allowing any clear patterns to show up. To compensate for the abundance of southerly, or positive V winds, I calculated the mean-wind for the year to provide a dividing point in the data. On the next Hovmoller, warm colors, i.e., white to red, will signify above average winds and cool colors, i.e., white to blue, will signify below average winds. A black contour interval is added to provide reference to the 0 m/s2 value.

It can be seen by comparing the winds and precipitation, Hovmoller over the same year and domain as V winds, that the two variables show quite close temporal agreement with one another. Though these results should be expected, it is of importance to note this agreement now so that later analysis can use winds as a parameter with confidence.

-jamie

Event Classification of IWV and IWVF

Welcome to the wild world of classifying northward surges in the North America monsoon!

Currently, I am working with surge characteristics of IWV and IWVF. My domain of study is, for now, 29 to 33N and -116 to -107W. Here is an example of the domain (plotted is precipitation on 19 Aug 1983).

For both IWV and IWVF a daily, domain-aggregated value was calculated from the daily-mean value of each pixel. This provided a daily value (1 June - 30 Sept) for each year of study (1980-2007). From this data the 95th, 90th, 80th, 70th and 50th percentile values were calculated for each day. This data provided the threshold from which events could be characterized for IWV and IWVF.

An event was first classified by one of five ways: The daily-aggregated value for each day was greater than or equal to the (1) 95th, (2) 90th, (3) 80th, (4) 70th or (5) 50th percentile for the specific given day. Events were further restricted to (A) two consecutive days or (B) three consecutive days of 1 through 5. Here are examples of events for IWVF and IWV, 3 consecutive days of equal to or greater than 90th percentile (2B), and a Hovmoller of precipitation for 1983:



Though the criteria for IWVF and IWV seem to be picking-up on some of the events displayed in the precipitation Hovmoller, it is clear that this classification is not yet perfected.

As for tomorrow, further work will be done to refine this classification scheme. I will also begin work using EOF analysis to better understand IWVF and IWV patterns for July and August.

-jamie