A MARKOV CHAIN MODEL FOR MEAN TEMPERATURE OCCURRENCES AT NEW YORK CITY, NEW YORK.
NEW YORK UNIV N Y
Pagination or Media Count:
A simple Markov chain probability model is found to fit most of the mean daily temperature data for New York City. Daily mean temperature data are tested for a twenty-four year period 1936-1959, and monthly mean temperature data are tested for a ninety year period 1871-1960. Both sets of data are tested for each month of the year and for the following periods January and February, April and May, July and August, October and November, and January through December. Chi-square tests show that the ninety year monthly data are not Markovian except for the month of July. During the periods that the model fits the mean daily temperature data, the probability that the mean temperature for any day will be above or below the long term mean twenty-four years for the respective month depends only on whether the mean of the preceeding day is above or below the monthly mean. This probability is independent of the events of earlier days. The probabilities of several temperature occurrence patterns can be determined by using the mathematically derived properties of the Markov chain model. Author