No revision after all

I have been repeating the same old argument that I might revise the Washingtonian Weather Index. After looking at how skewing things more towards temperature would work, it appears that the index would be less of an accurate representation of our weather with those kinds of changes.


There were two primary things that I have been looking at due to unexpected results. The first was the comparison of July 2013 to August 2012. I felt that the order was correct, but I also felt that the two months were too close. I struggled a lot more in August 2012 than in July 2013. If I go strictly by the temperature score, it's even closer. That's 53.94% compared to 53.74%.

As an interesting side note, August 2013 scores worse than either of those two months in regards to temperature. In fact, it was our third worst ever. The primary reason that it scored better overall is because we received some much needed rain.

10 Worst Months Ever (Temperatures)
Rank	Month	Year	Score
1	January	1950	38.34%
2	August	1967	45.74%
3	August	2013	48.45%
4	July	2009	49.83%
5	January	1949	49.95%
6	August	1977	52.65%
7	July	1958	52.65%
8	July	2004	52.82%
9	July	1985	53.08%
10	August	1986	53.09%

I already had some suspicions. 2012 seemed more unrelenting. Every day gave me problems. I think this made it feel worse than a lot of bad weather with at least a brief break in between. More specifically, I think July 2012 made the following month feel worse than it really was.

The other issue I have been looking at is the low score for October this year. We had a stretch a weather that was pretty much where I like to keep things. Despite this simple fact, it actually ranked as the worst October in terms of temperature.

October Temperatures Over the Years
Rank	Year	Score		Rank	Year	Score		Rank	Year	Score		Rank	Year	Score
1	1983	91.96%		18	1969	89.44%		35	1950	87.64%		51	1956	86.14%
2	1977	91.75%		19	1974	89.43%		36	1982	87.60%		52	1976	86.09%
3	1990	91.72%		20	2010	89.26%		37	1986	87.54%		53	1978	86.03%
4	1981	91.61%		21	2000	89.12%		38	2005	87.51%		54	2004	85.98%
5	1994	90.64%		22	2001	88.91%		39	1958	87.45%		55	1970	85.94%
6	1999	90.39%		23	2009	88.87%		40	2006	87.29%		56	2003	84.98%
7	1968	90.33%		24	1957	88.74%		41	1952	87.24%		57	1951	84.91%
8	1959	89.96%		25	1997	88.63%		42	1967	87.22%		58	1996	84.76%
9	1973	89.96%		26	1949	88.23%		43	2002	87.19%		59	1987	84.70%
10	2011	89.90%		27	1972	88.22%		44	1979	87.15%		60	1965	84.25%
11	2007	89.90%		28	1960	88.16%		45	1980	87.05%		61	1988	83.87%
12	1985	89.72%		29	1953	88.15%		46	1992	86.84%		62	1984	83.58%
13	1955	89.63%		30	1998	88.13%		47	1993	86.83%		63	1963	83.31%
14	2008	89.59%		31	1989	88.07%		48	2012	86.73%		64	1971	82.97%
15	1962	89.59%		32	1966	88.07%		49	1961	86.49%		65	1991	82.78%
16	1995	89.56%		33	1964	88.00%		50	1954	86.16%		66	2013	81.57%
17	1948	89.55%		34	1975	87.94%

I should also point out the 81.57% is not a bad score. While it's not quite April, October is a highly competitive month when it comes to temperatures. I have posted a table below showing the best score for each month, the worst score, and the difference between the two with rainfall excluded.

Month	Best	Worst	Difference
January	87.13%	38.33%	48.79%
February	89.37%	62.87%	26.49%
March	92.06%	70.5%	21.55%
April	91.33%	81.88%	9.45%
May	92.5%	74.44%	18.05%
June	88.72%	62.22%	26.5%
July	76.29%	49.83%	26.46%
August	74.76%	45.74%	29.02%
September	82.53%	60.69%	21.84%
October	91.96%	81.57%	10.38%
November	88.77%	61.26%	27.51%
December	83.95%	58.79%	25.15%

I don't want you to just look at October here. The best September is less than one point ahead of the worst October. Call it a hunch, but I don't think this will be the only time that October throws me off.

What I have really noticed in this evaluation is that the category that I was thinking of devaluing, rainfall, is actually helping to correct for problems with temperatures. It would be better to tweak how temperatures are scored than to adjust how categories are weighed. The problem with this is that the purpose of the weather index is to be as objective as possible (while maintaining the originally subjective purpose). Moving off of the averages and extremes would take away from the objectivity.

There will always be problems with attempts to objectively measure something as subjective as how nice the weather is. Among the problems with weather is that there are a lot of things that can change how you feel. Most do not have reliable historical data. Wind, humidity, and rainfall rate can influence how comfortable things seem. So can mood.

Most flaws encountered have explanations. Overall, the numbers seem to work. We had a below average start to the year and end of the year combined with a rotten summer. This resulted in our third worst year ever recorded. I think it's best to leave the Washingtonian Weather Index as is.