A k-nearest-neighhor simulator for daily precipitation and other weather variables

A multivariate, nonparametric time series simulation method is provided to generate random sequences of daily weather variables that "honor" the stati stical properties of the historical data of the same weather variables at t he site. A vector of weather variables (solar radiation, maximum temperatur e, minimum temperature, average dew point temperature, average wind speed, and precipitation) on a day of interest is resampled from the historical da ta by conditioning on the vector of the same variables (feature vector) on the preceding day. The resampling is done from the k nearest neighbors in s tate space of the feature vector using a weight function. This approach is equivalent to a nonparametric approximation of a multivariate, lag 1 Markov process. It does not require prior assumptions as to the form of the joint probability density function of the variables. An application of the resam pling scheme with 30 years of daily weather data at Salt Lake City, Utah, i s provided. Results are compared with those from the application of a multi variate autoregressive model similar to that of Richardson [1981].