SENSITIVITY OF ESTIMATED TRENDS OF TROPOSPHERIC AND LOWER STRATOSPHERIC TEMPERATURE FROM RADIOSONDE DATA. PART 1. CHOICE OF DATA SET, LENGTH OF SERIES, AND ANALYSIS METHODS

A. M. Sterin

Different options of estimating upper-air temperature trends from radiosonde data are considered. The sensitivity of the upper-air long-term temperature trend estimates obtained for the longest possible period of radiosonde observations to a number of factors is addressed, such as the choice of the data set, a trend calculation period, a trend estimation technique, and the detection and elimination of inhomogeneities in temperature series. The sensitivity of the estimates to the choice of radiosonde data processing, generalization, and analysis technique is also considered. The robust regression analysis methods used for estimating the upper-air temperature trends along with the traditional methods are taken into account. The statistical parameters of the models of the troposphere and lower atmosphere temperature anomaly time series of monthly resolution are given. Along with the trends, the models incorporate autoregression components.