Statistical Simulation of Daily Air Temperature Patterns Eastern North America to Forecast Seasonal Events in Insect Pest Management

Two simple stochastic models were developed to generate daily minimum and maximum air temperahlre time series from input monthly mean and extreme minimum and maximum normals. Factors taken into account in the temperature generation are latitudinal and vertical lapse rates, as well as the latitudinally and seasonally-variable correlation between daily minima and maxima. The first model assumes that daily values are independently and normally distributed random variables, whereas the second takes into account serial autocorrelations in the temperatures of successive days. Parameter values for both models were obtained using weather data from stations distributed over the eastern seaboard of the United States. The temperature traces generated by each model were compared with actual data. Four different insect phenology models were also used to compare the two weather simulation methods with actual weather data in terms of the phenology forecasts generated. It was found that the degree of autocorrelation between minima or maxima on successive days, as well as the correlation between daily minima and maxima, vary systematically with time of year and latitude. Altitudinal and latitudinal lapse rates also vary with time of year. Both models produce weather traces that are statistically indistinguishable from actual weather data. Phenology forecasts using deterministic air temperature regimes were considerably late compared with forecasts obtained from either stochastic regimes or actual data. The variability of phenology forecasts produced using synthetic weather traces as input was smaller than that obtained using real weather data as input, particularly with the serially independent temperature regimes generated by model 1.