Generation of synthetic sequences of half-hourly temperatures

We present tools to generate synthetic sequences of half-hourly temperatures with similar statistical characteristics to observed historical data. Temperatures are generated using a combination of daily and half-hourly temperature models which account for intra-day and intra-year seasonality, as well as short- and long-term serial correlations. Details of the model estimation are given as well as a description of the synthetic generation. Keywords: temperature data, time series, Fourier series, ARMA models, seasonal block-bootstrap, synthetic …