Аннотация:The article is dedicated to the problem of recovering gaps in data series of experimental long-term continuous high-frequency observations of carbon dioxide concentration and air temperature. The study was conducted using the observation results from an automatic eco-climatic station located in a tropical monsoon forest in southern Vietnam (Dong Nai biosphere reserve). Gaps in observation series are, as a rule, random and caused by technical malfunctions of the instrumentation. Accurately recovered observation series allow for the assessment of the temporal variability of observed parameters on different time scales. In the scope of this study, options for recovering the continuity of time series based on mathematical statistics methods — autoregression (ARIMA) and the linear prediction method — have been considered. A comparative analysis of the accuracy of gap recovering using different methods is provided.