An example of principal component analysis application on climate change assessment

Abstract

Climate change assessment is usually based upon air temperature and precipitation changes on an annual and seasonal basis, but there are more levels to their significance as presented by parameters derived from these two basic parameters. In order to define their relevance for climate changes, the principal component analysis (PCA) was performed. In this case, ten meteorological parameters and climate change indicators were defined for two meteorological stations located in geographically completely opposite parts of the country; station Osijek is in continental region of Croatia, and Dubrovnik station is located in the Mediterranean region. Analyses were done for the period 1985–2016 on an annual and seasonal basis. All defined indicators present basic climate change characteristics on annual and seasonal basis as follows: precipitation sum, mean air temperature, air temperature sum, standard deviation of daily air temperature, maximum daily air temperature, maximum daily precipitation, number of days with precipitation > 30 mm, number of days with no precipitation, 1-month standardized precipitation index, and aridity index. In the first step, it was applied on the set of linear regression coefficients defined for 10 climate change indicators. During the second step, PCA was applied on the computed Mann–Kendall test statistic, ZMK.in order to determine the existence of significant temporal tendencies in the indicator values. The provided research proves PCA is a very useful tool for implementing this approach, particularly in the Mediterranean region which shows high sensitivity to many variables important for climate characterization.