Prices on natural gas spot markets exhibit a complex seasonality that is dependent on exogenous factors. The analysis of historical prices and temperatures shows an important relationship between price and temperature, especially in longer periods of low temperatures. We propose a model for natural gas spot prices under consideration of influencing variables derived from heating degree days. The stochastic dynamics of price and temperature are modeled by appropriate autoregressive moving average processes. Daily title-transfer-facility (TTF) dayahead and weekend prices are used for calibration. We present two applications of this model: valuation of gas storage by Monte Carlo least-squares methods and pricing of full supply contracts for customers whose consumption strongly depends on temperature.