Separating Different Scales of Motion in Time Series of Meteorological Variables

Abstract

The removal of synoptic and seasonal signals from time series of meteorological variables leaves datasets amenable to the study of trends, climate change, and the reasons for such trends and changes. In this paper, four techniques for separating different scales of motion are examined and their effectiveness compared. These techniques are PEST, anomalies, wavelet transform, and the Kolmogorov-Zurbenko (KZ) filter. It is shown that PEST and anomalies do not cleanly separate the synoptic and seasonal signals from the data as well as the other two methods. The KZ filter method is shown to have the same level of accuracy as the wavelet transform method. However, the KZ filter method can be applied to datasets with missing observations and is much easier to use than the wavelet transform method.

Multivariate quantile mapping bias correction: an N-dimensional probability density function transform for climate model simulations of multiple variables

“…For example, Mehrotra and Sharma (2016) corrected lag 1 autocorrelation statistics at daily, monthly, quarterly, and annual time scales. With MBCn, one could decompose a given time series into different time scales, for example using a multiresolution wavelet analysis or Kolmogorov-Zurbenko filtering (Eskridge et al 1997), simultaneously bias correct the partitioned time series, and then reconstruct the original series. The basic approach can easily be extended to both space and time dimensions of multiple variables.…”

“…While biases in spatial dependence are strongest in summer, a similar pattern of results is evident in the other seasons. Table 1 shows seasonal values of the Kolmogorov-Smirnov statistic, the maximum difference between Spatial and temporal scales for a meteorological field like precipitation are intrinsically linked (Eskridge et al 1997). Does correcting spatial coherence lead to an attendant improvement in temporal dependence?…”

A Southern Hemisphere booster of super El Niño

“…Monthly anomalies were obtained as departures from monthly mean climatology for the period of 1958-2001; then, the KolmogorovZurbenko (KZ) filter [Eskridge et al, 1997] was applied to remove high-frequency (less than six months) and longer-than-ENSO scale (8 years) variations. The KZ filter gives iterative moving average that removes highfrequency variation relative to the window size; the method cleanly separates various time scales of meteorological variables and has the same accuracy as the wavelet method.…”

A new methodology for the extension of the impact of data assimilation on ocean wave prediction

“…Focusing now on the statistical methodologies employed, a short description of KZ filters is given (for details, see Eskridge et al 1997;Rao et al 1997). They consist of a series of iterative moving averages aiming at the removal of high-frequency variations from the initial data and, therefore, modifying the observations and the corresponding model forecasts into a compatible form.…”