RESTORATION OF LONG-TERM TIME SERIES OF AIR TEMPERATURE IN EUROPEAN RUSSIA

摘要

In the European territory of Russia, 100 stations with long-term (about 80 years on average) time series of monthly air temperature were chosen. A technique was developed for a consecutive restoration of missing data and increase of records. The technique has three main stages: restoration on the basis of individual analogs, restoration on the basis of space models, and application of a seasonal function. To estimate the efficiency of restoration both on dependent and on independent data, a complex of parameters is offered. As a result of a consecutive restoration, it was possible to increase sizes of time series up to 110-130 years depending on month. More data were restored for coldest months of the year than for the warmest. The error of restoration did not usually exceed 20% with respect to a standard deviation of long-term time series.