Artificial neural networks in forecasting minimum temperature (weather)

Forecasting the minimum temperature (T/sub min/) is one of the most important operational practices carried out by Meteorological Services worldwide. In tackling the problem of the minimum temperature forecasting using artificial neural networks, the authors use raw data extracted from the synoptic meteorological observations recorded at three-hour intervals at Larnaca Airport, Cyprus (34 degrees 52'52"N, 33 degrees 37'32"E). From each one of the eight sets of observations available daily, the following elements have been extracted: total cloud cover, eastward wind component, northward wind component, visibility, present weather condition, atmospheric pressure, dry-bulb temperature, wet-bulb temperature, and low cloud cover. Besides these records, astronomical day length and observed minimum temperature of the previous night were used for forming the input data vector to the neural network. A total of 141 days of the winter and spring of 1984 were used for training and evaluating the forecasting models. The learning algorithm is discussed.