CPC Outlooks for Major U.S. Cities

Translation of CPC Forecasts to Individual Cities

The CPC monthly and seasonal outlooks are for forecast divisions. Two files are available to translate temperature forecasts for the CPC forecast divisions into forecasts for temperature or degree days for individual cities within the U.S.

Data Files

Translation of CPC Forecasts to Individual Cities

The following file contains the information required to downscale the temperature outlook information to temperature outlooks for individual cities.

File Explanation

Filename: citytemptran.dat
Description: Gives the information required to translate the CPC temperature outlooks to individual cities.
Derivation: Regression relationships between downtown airport temperatures and the associated forecast division, enhanced by an analysis of recent trends. Derived from data from 1951-1997.
Contents:
Column 1: The city number. The city name and location are written in ASCII on the last (rightmost) columns of text, and also in a separate dictionary file.
Column 2: Forecast Division Number. This gives forecast division in which the city resides.
Column 3: Season Number, defined by the month number of the CENTER month of a three month season, 1=DJF, 2= JFM, ... 12=NDJ.
Column 4: Seasonal mean temperature of the downtown airport.
Column 5: Seasonal standard deviation of the downtown airport.
Column 6: Constant term, a, of the regression equation that translates the forecast division temperature (t(FD)) (See Col. 2 for division number) to the downtown airport temperature (t(City)). t(City) = a + b*t(FD). The “a” term is listed in Column 6.
Column 7: Multiplicative term, b, of the regression equation that translates the forecast division temperature (See Col. 2 for division number) to the downtown airport temperature. t(City) = a + b*t(FD). The “b” term is listed in Column 7.
Column 8: The skill of the equation: t(City) = a + b*t(CD). This value, r, is actually the correlation coefficient between t(FD) and t(City) for the given season and city. This value, can be used together with “b” in column 7 to find useful characteristics of the relationship between the city airport and its forecast division.
Column 9: The average difference between seasonal mean temperature at the downtown airport (Column 4) and the seasonal mean minimum temperature (MN-MIN) at that station. Note that the temperature range is twice the value presented in Col. 9, and the seasonal mean maximum temperature can be estimated by adding the result displayed in Col. 9 to the mean.
Column 10: Mean difference between the downtown airport mean temperature (AP) and that of the Forecast Division(FD) Column 10 gives the (AP-FD).
Column 11: Mean difference between the urban area average temperatures (U) and the forecast division average temperatures. (U-FD)
Column 12: Mean difference between the urban area average temperatures and the mean temperature at surrounding rural stations (R) within 100 km of the city center (U-R).

Data Set Usage
This data set is used to obtain forecast for individual cities from forecast division data.

A: Probability anomaly method.

Assume the probability anomaly for below, near, and above median values of the seasonal temperature and precipitation to be the same for the city as it is in the forecast division in which the city resides. The probability anomalies are available in digital form on the cpcllft.dat forecast file. Forecasts can also be obtained by manually interpolating values from the forecast maps to the exact city location and used and forecast distribution rules to determine the probability anomaly for other categories. The anomaly is then applied to the city's climatology. Information provided in the city temperature translation file can be used to obtain the class limits.
Class Limits for Below Normal Temperatures = Seasonal mean - .431*sd
Class Limits for Above Normal Temperatures = Seasonal mean + .431*sd

B: Forecast distribution method.

Locate the appropriate forecast division for the desired city in the city temperature translation file. The CPC outlook for that division is given in the probability of exceedence forecast file sets: (cpcllftd.YYYY.dat). Specific exceedence percentile values for the forecast probabilities can be converted to exceedence values for city temperatures by applying the the following formula:

variable definition:
t(city) - The mean seasonal temperature at a downtown airport.
t(FD) - The mean seasonal temperature for the forecast division appropriate for the city.
a, b - empirical constants found in in columns 6 and 7 of the city temperature translation file.
        
Formula:(FORTRAN syntax used for formula)
t(City) = a + b*t(FD)

EXAMPLE: The one-month lead CPC seasonal forecast issued in February, 1995 for MAM 1996 indicated that there was a 50% chance that the 3-month seasonal mean temperature for Forecast Division 4 would exceed 46.59 degrees F. Find the temperature with a similar exceedence percentage for New York’s LaGuardia Airport.

From the city temperature translation file:

T(LGA)= 5.46+1.01FD(4) = 5.46 + 1.01*(46.59) = 52.52 degrees F.

Because the distribution for seasonal mean temperatures are are very close to Gaussian for the majority of seasons and locations in the U.S., the entire forecast temperature distributions for given cities are easy to find. The forecast means and standard deviations for the forecast division temperatures can be found in columns 19 and 21, respectively, of the probability of exceedence forecast file. The information for the divisions can be translated into forecasts for cities with information in the city temperature translation file as follows:

 

variable definition:
t(city),t(FD) - forecast mean seasonal temperature at the downtown city airport and its forecast division, respectively.
sd(city), sd(FD) - forecast standard deviation of the city airport and forecast division mean temperatures.
a, b - constants found in the city temperature translation file in columns 6 and 7, respectively.
Formula: (FORTRAN syntax used for formula)
t(city) = a + b* t(FD)
sd(city) = b * sd(FD)