TIME SERIES ANALYSIS IN AGROMETEOROLOGICAL PROBLEMS WITH EMPHASIS ON SPECTRUM ANALYSIS

This paper reviews the principles and use of spectrum analysis and discusses the effectiveness of combining multiple regression techniques with cross-spectrum analysis in time series studies. This method is useful in comparing the variations of an empirical index or the response of different instruments to environmental factors as a function of time. The variability of daily latent-evaporation estimates from a regression technique is examined and compared with evaporation readings from atmometers. A comparison is made of the variation in evaporation from two types of small atmometers and two types of tank evaporimeters. All four instruments appear to respond to meteorological parameters similarly at Swift Current and Ottawa.