|
...beginning ending month an extended dry period using scheme described in Harnack and Small (2002). After the identification stage, dry period characteristics such as average length, longest length, and average time between prolonged dry periods were assessed. In each CD there are 60-63 dry periods in the sample record, with close to 50% of short duration (i.e., 6 months or less). The average dry period length is 9-10 months, as is the average time period between dry periods (i.e., average non-dry periods length). It is clear that the beginning (i.e., 1900s) and recent decades (1970s to 1990s) have been relatively wet, while the 1910s, 1930s, and 1960s have been relatively dry. In addition, available published data of antecedent circulation and sea surface temperature indices were used with New Jersey seasonal precipitation data to compute simple correlation statistics. The correlations, though no greater than about 0.50, were large enough to encourage prediction trials. The trials included the use of various indices as predictors in regression models for the purpose of determining the forecast skill level of one-season lead seasonal precipitation forecasts for New Jersey. Based on a small sample of independent test forecasts (i.e., 10 cases) for selected regression models, the number correct was not large enough, in general, to distinguish its performance from random chance and persistence. There may still be some utility in their use, but this was not pursued here.
KEY WORDS: New Jersey climate, New Jersey drought, New Jersey precipitation, seasonal precipitation forecasting
INTRODUCTION
Precipitation is necessary for sustaining ecosystem health and often has socioeconomic impacts on agriculture, energy use and human health resulting from having too much or too little precipitation. In particular, too little precipitation is manifested by drought. New Jersey has had several periods of drought in its history with significant socioeconomic consequences ranging from crop failure to severe water restriction (Ludlum, 1983).
In an earlier paper (Harnack and Small, 2002), a modified standardized precipitation index (SPI) was applied to the precipitation record of New Brunswick, New Jersey. This application resulted in the identification of important dry periods in New Jersey and the determination of the statistical characteristics of dry periods such as frequency, length, and timing. Some of the important results included:
1. Identification of 66 dry periods
2. An average dry period length of about 9 months
3. Eighteen dry periods of 12 months or longer
4. Four dry periods of 24 months or longer
5. A slight tendency for the fall months to be a turning point
6. Identified the 1930s and 1960s as the driest decades
7. Identified the 1900s and 1990s as the least dry decades, and
8. Identified the average time period between dry periods of about 9 months (19 months for moderate or greater dry periods).
The objectives of the current work are twofold:
1. To extend the Harnack and Small (2002) study from analysis at one station in New Jersey (New Brunswick) to the entire state using Climatic Division (CD) data.
2. To formulate and test various multiple linear regression models for the purpose of predicting seasonal precipitation for New Jersey using antecedent circulation and sea surface temperature indices.
There is an extensive literature describing long-range forecast experiments (also referred to as climate forecasting, or by some specific temporal terminology such as 'seasonal' forecasting). The methodology falls into two broad categories. The first is the numerical approach, which entails the use of mathematical models, usually composed of the primitive equations describing atmospheric fluid dynamics, to forecast the future state of the atmosphere. The second approach is empirical, which uses historic data of atmospheric and oceanic variables to determine statistical relationships that may be used to forecast surface air temperature or precipitation for periods of a month or longer. Several extensive reviews of these studies are available to the interested reader (Goddard et al., 2002; Stockdale, 2000). The forecast skill for surface variables, including precipitation, for time scales of a month to a season (three months) with lead times of a season or less is quite small. The estimated % correct for operational seasonal temperature forecasts verified for a three-class scheme (i.e., using below-, near-, and above-normal classes determined by ranking into terciles) are below 60% for even the best combination of season (winter) and U.S. region (southeast). The corresponding percent correct for seasonal precipitation is about 50%. The expected percent correct via random chance is 33%. Despite the many attempts at using statistical methods for this purpose, there is no indication that circulation indices, which are defined in the next section, have been employed for forecasting eastern U.S. seasonal precipitation.
MATERIALS AND METHODS
The methodology followed to meet the first objective is similar to that used in Harnack and Small (2002), therefore the description that follows is taken from the previous study, with appropriate modification.
Initially, dry periods are identified by using three-month precipitation totals. A critical value was then selected to objectively identify a "dry" three-month period. In order to identify only important dry periods, the chosen value had to be small enough so that the phenomena was somewhat rare in occurrence, yet would likely have socioeconomic consequences. The precipitation value (termed the critical value herein) corresponding to the driest 20% of the record for each three-month combination was computed. The precise procedure used is as follows:
1. For the purposes of objectivity, not for significance testing, it was assumed that the three-month precipitation data varied according to the normal distribution. Precipitation data appears to be "more normal" as the time period increases from daily to seasonal. While histograms show a slight degree of skewness in the three-month data, this is unimportant for the purpose described here.
2. Using the normal distribution, the number of standard deviations corresponding to the driest 20% of all cases was determined. This value is 0.83.
3. The critical three-month precipitation value used to identify dry periods for each three-month combination was calculated using:
CV = M - (0.83)S where CV is the critical three-month precipitation value, M is the sample mean three-month precipitation total, and S is the sample three-month precipitation standard deviation. The sample size is approximately 108, but varies slightly depending on the CD and season.
Every individual three-month period was subsequently compared to the appropriate critical value to identify dry three-month periods. Once...
NOTE: All illustrations and photos
have been removed from this article.
More articles from Bulletin of the New Jersey Academy of Science
Aggression and oxygen uptake in Siamese fighting fish., September 22, 2005
Medieval, copper-based green pigment., September 22, 2005
Looking for additional articles?
Search our database of over 3 million articles.
Looking for more in-depth information on this industry?
Search our complete database of Industry & Market reports by text, subject, publication
name or publication date.
About Goliath
Whether you're looking for sales prospects, competitive information, company
analysis or best practices in managing your organization,
Goliath can help you meet your business needs.
Our extensive business information databases empower business
professionals with both the breadth and depth of credible,
authoritative information they need to support their business
goals. Whether it be strategic planning, sales prospecting,
company research or defining management best practices -
Goliath is your leading source for accurate information.
|
