Presentation is loading. Please wait.

Presentation is loading. Please wait.

(a) Measuring the memory in a time series with auto-correlation function (ACF)

Similar presentations


Presentation on theme: "(a) Measuring the memory in a time series with auto-correlation function (ACF)"— Presentation transcript:

1 (a) Measuring the memory in a time series with auto-correlation function (ACF)

2 (a) Shifting the time series by one time step gives pairs of observations We calculate the (auto-)correlation at lag 1 r = 0.45

3 If we have sufficient data we can shift the time series also by two or m time steps. The shifting is also called lag. r = 0.13

4 (a) Using an number of different shifts (‘lags’) we obtain a Auto-correlation function (ACF)

5  acf(x) Null hypothesis r-values that mark the significance level (at 5%) Example with a random sample from White Noise

6 . “White noise” “Red noise” Autocorrelation Function The mean, variance autocovariance (and thus the ACF) are not changing over time.

7 “Red noise” “White noise” The mean, variance autocovariance (and thus the ACF) are not changing over time.

8  We have studied already many times correlations between two time series (e.g. temperature records from Albany and New York Central Park)  This was done without a time lag.  But we can shift one time series by one time step, 2 or m time steps and then calculate the correlation  => Cross Correlation Function (ccf)

9 Article in Nature 2012.  Paleoclimate temperature reconstruction from temperature sensitve ‘proxies’

10  CO2 curve  ftp://aftp.cmdl.noaa.gov/products/trends/co2 /co2_mm_mlo.txt ftp://aftp.cmdl.noaa.gov/products/trends/co2 /co2_mm_mlo.txt


Download ppt "(a) Measuring the memory in a time series with auto-correlation function (ACF)"

Similar presentations


Ads by Google