That said, maybe the tuning is a good way to transform from depth to time, but then to
do time series analysis on the result and claim anything about the frequencies associated with the tuning target seems misguided to say the least.
Last I heard, Tamino claims to have an unrelated day job
doing time series analysis, whatever that means.
Not exact matches
Lacking a mechanistic explanation for this
time -
series analysis - derived cycle, we don't know — in fact, all the PDO is (so far) is a statistical phenomenon.
An archival
time -
series analysis is simple enough that readers
do not need a background in statistics to understand the underlying logic.
A
time -
series analysis just might
do the trick!
Is the
time series analysis consistent, i.e.
does it take account of academy conversions?
We
do not want our
analysis to rest heavily on synthesised data, but we consider one very long
time series: that of the Dow Jones Industrial Average (DJIA).
Thompson et al.
do not provide a
time series estimate on the effects of the bias on the global temperature record, but Steve McIntyre, who is building an impressive track record of
analyses outside the peer - review system, discussed this topic on his weblog
I think that
doing this on the basis of
time series analysis is somewhere between hard and impossible.
This is a grossly imperfect way to
do this, of course, but a reasonable approximation to the real
time -
series analysis.
If it matters at all, I never found a use for FFT in my previous line of work (population ecology), although I
did have occasion to use PCA on a
time series analysis.
Discounting Tamino's
analysis because you don't like it or don't consider him objective doesn't detract from the fact the Tamino is a professional
time -
series analyst whose work in climate science not only stands the test of
time but is widely considered a de facto standard in climate science.
@DS: This is why you
do not smooth a
time series then use the smoothed data as input into a subsequent
analysis.
Second why not hold out 50 % of your temp
time series (random selection of each data point perhaps),
do your
analysis on on one half and check the fit to the other.
Of course, the statistical
analysis I've presented here is very crude, and there are lots of better things you can
do with more data and fancier
time -
series techniques.
It is worth noting that even under Dr K's
analysis of 1 / f noise, that 1 / f noise
time series does indeed have a population mean, and it is quite relevant to compute forcings of things like CO2 doubling relative to the population mean.
The misstatement that they are not is evidence that somebody
did not pay attention in the
time series analysis cours they should have taken prior to being called an expert in climate science
This doesn't affect your results, but I think you have your terminology switched around: The term «eof» refers to the loading (which is usually a spatial pattern in these
analyses), and «pc» can mean either the
time -
series or the combination of
time -
series + eof.
Personally, I would not
do straight
time -
series analysis of temperature data in the current era unless it contained a larger dynamic range.
for lack of warming since 1998» refers to a model that
does address serial correlation (being based on Kaufman, A., H. Kauppi, and J. H. Stock, 2006: «Emissions, concentrations and temperature: a
time series analysis.»
Nicola Scafetta says «It seems to me that you are arguing that because a clear physical mechanism is still missing and the calculations were
done on a short
time series than the result of our
analysis should be «rejected».
It seems to me that you are arguing that because a clear physical mechanism is still missing and the calculations were
done on a short
time series than the result of our
analysis should be «rejected».
Analyses of the annual mean streamflow
time series for the 14 streamflow clusters indicated periods of extended wet and dry periods, but
did not indicate any strong monotonic trends.
Improved
time series analysis methods
do not support the statistical significance and likelihood levels of the IPCC's conclusion that sea level rise has accelerated in the 20th century relative to the 19th century.
For example, the proxies in Gergis were screened against correlations with other grid cells within 500 Km (a rationale for why 500, and not say 477, or 567 km was appropriate... we aren't told how many of the the
time series correlated with adjacent grid cells, and how often the included or excluded
times series DID NT correlate with adjacent grid cells... This kind of stuff is reported and considered when conducting an exploratory
analysis.
However,
time -
series analyses did not reveal any relationship between the proportion of male stillbirths and mean annual ambient temperature.
In short, the global temperature
time series clearly
does not follow the model adopted in Schwartz's
analysis.
Fraedrich & Blender find persistence up to decades, Kiraly et al. find persistence lasting several years, so even if their
analysis applied to temperature
time series (which it doesn't) rather than fluctuations (which it
does), those
time scales aren't long enough to explain the trend on a century
time scale in observed temperature
time series.
Rather than looking at the dataset available in 1998 and using standard
time series analysis techniques (available in 1998) to establish its underlying structure and hence the null hypothesis, the choice in null hypothesis was limited to a couple of options (trend vs no - trend) that were particularly naive (probably more naive than human instinct) and a selection process was established that appears to use the skill criteria (exactly what was
done is unclear from the text) and inappropriate multiple use of the dataset.
Nonstationarity of error terms is a serious problem in
time -
series analysis, but I don't have a good sense of how well this issue has been treated in climate
analysis.
The posts
did point out that «Some of the other pieces in this
series are fine» but
do not reflect the large amount of
analysis in the investigation of the way the emails have been misused by those with a political agenda and the extensive context we included to indicate the pressure scientists writing those emails were under from
time - consuming requests for data.
Also, for
time series analyse4s, where it would be useful, say, to update a published
analysis with the latest few years of data, this is already normally
done via a database on on the authors website...
I have a feeling a lot of PhDs could get minted from extending your
analysis here alone ----
does it hold for longer
time series, using Vostok Ice Cores (not tree rings please!!)
b) then start on the CO2 forcing / CO2 concentration
time series analysis which would similar process as just
done of the
analysis for GISS
time series
What I don't understand is why there is so much angst about what is after all only simple empirical observations about the nature of a
time series (even if aspects of the
analysis maybe open to theoretical debate), and so little curiosity about what this all means for statistical inference more generally in climate science.
If I understood VS correctly (don't be too harsh on me if not) there's a way to detect such behaviour in data by applying some
time series analysis procedures.