Near Term Effects

Ideally, the normal high and low temperatures, humidities, precipitation, winds and other normalized data are calculated based on years of climate data. WeatherCurrents.com does not yet have years of climate data for most of its sites (the major exception is Temecula, which has a decade of temperature data) to produce a completely accurate result for normal conditions. The root problem is that one hot or cold spell or other unusual weather conditions in a period when there aren't many years of data can skew the calculated "normal" temperature, wind speed, humidity, rainfall, etc. so that it appears hotter or colder than it really is. Several more years of data will reduce the impact of abnormal weather conditions and offer a truer reading of what "normal" is.

As a prime example of this effect, consider Temecula's data from October, 1999 and October, 2000. October, 1999 was unusually warm, featuring several 100 degree days, one as late as October 30th. On average, October, 1999 was a full seven degrees warmer than October, 2000. Now, October, 2000 was likely cooler than a normal October. But do they even each other out? Probably not. Since then, October, 2001 featured temperature data in between October, 1999 and October, 2000. The totals for late October in particular will likely be skewed a little higher than the true norm for a few years to come.

Calculating Normal Conditions

Currently, the normal conditions that you see on WeatherCurrents.com are an 11 day moving average of the particular weather condition around and including the target date, for all of the years currently on record. For example, the normal high and low temperatures for February 15th would be computed by averaging all of the high and low temperature samples for February 10th through the 20th for all of the years in the archive. Averaging gives a nice seasonal sampling from many days of data surrounding the target date and reduces aberrations for a single date. If you visit these pages often, you will notice the normal highs and lows sliding upwards as you might expect in the springtime and decaying in the autumn.

Likewise, normal temperatures, humidities, and wind conditions for time of day are also computed using the same 11 day moving average. The difference is that the samples used for averaging to compute the normal condition are derived from the nearest sample to that time of day for each day used in the computation. Thus, a normal temperature for 2:00pm on February 15th, for example, would currently use the following 33 samples:

Thus, the normal temp for 2:00 pm on February 15th is: (59.6 + 60.5 + 61.0 + 54.5 + 65.4 + 69.6 + 55.3 + 58.9 + 70.5 + 78.5 + 59.6 + 47.8 + 59.0 + 49.8 + 52.1 + 58.1 + 66.4 + 68.7 + 65.3 + 63.8 + 62.1 + 60.2 + 69.9 + 80.4 + 79.8 + 72.7 + 67.9 + 76.8 + 64.7 + 53.2 + 58.2 + 65.3 + 73.8) / 33 = 63.9°F.

For brand new WeatherCurrents sites, the normal temperature is only calculated from the last five days of data, and should be considered inaccurate.

Normal Rainfall

Normal rainfall is an entirely different matter. This is because the types of data that most people want to see are seasonal rainfall to date, and normal rainfall for a month. Wild differences in season to season rainfall mean that averages are difficult to calculate. For example, very low rainfall seasons were experienced in 2001-2002 and again in 2006-2007. A very high rainfall season occurred in 2004-2005. Some areas in California (Palm Springs) saw all-time low amounts of precipitation for the 2001-2002 season.

WeatherCurrents.com provides normal rainfall to date totals for each community that has more than three complete seasons of data, despite potential skewing from high and low seasons. Normal rainfall by month is provided for all communities, again despite potential skewing, on the Climate Pages (Temecula is given here as an example).

At this writing, with nine years of rainfall data, WeatherCurrents' Temecula site is approaching accurate values. The accuracy of the data for all the other sites is potentially suspect until more seasons of data are collected.

Until such time as WeatherCurrents can accumulate several years of data in our archive, please treat the normal conditions computed herein as best guesses, and know that they may be a little inaccurate. Time will correct any inaccuracies; every day that passes will make a difference. But we think even at this early stage that they will be close.