It is worthwhile thinking of a time series as the combination of components, a trend component (T), a seasonal component (S), and an error (or irregular) component (E).
The trend is the long term movement of a time series. For example, both the time series in the left column of Figure 11.1 have a steady trend. The trend of the temperature anomaly data, in contrast, appears to have a slight upward trend from 1880 to around 1960, at which point the trend appears to increase at a much faster rate. This looks as if it were a non-linear trend.
The seasonal component is a pattern in the series that always occurs at a fixed, unchanging period of time. Possible periods of seasonal patterns are over every week or year. For example, our school supplies series has a very strong seasonal component, with peaks every August (often a month before the start of a school year). The AirPassenger data set, too, has a very strong seasonal component with peaks every summer. The seasonal...