| T O P I C R E V I E W |
| TonyNt |
Posted - 10/31/2012 : 02:37:30 AM
Origin Ver. and Service Release (Select Help-->About Origin): 8
Operating System: XP
Hi everyone. I need some help: I am interested in correlating the data between two columns: one has data from the irradiance of the sun and the other one has data of the irradiance on the ground. It seems that there is not a linear relatioship between them (both data sets should have a periodicity of 11 years) and I want to see how the correlation is between them. I believe that there should be a time lag between the two data sets. So how can I use the lag correlation (or cross correlation) and how do I intepret the results?
Thanks so much in advance.
Take care
Tony |
| 3 L A T E S T R E P L I E S (Newest First) |
| couturier |
Posted - 11/01/2012 : 3:51:40 PM
Maybe you can try this tool, that computes the lag betwwen two or more datasets. |
| TonyNt |
Posted - 11/01/2012 : 2:30:42 PM
quote:
Originally posted by greg
To do a Cross Correlation, select the two columns of data and choose Analysis : Signal Processing : Correlation.
Assuming that your X data units is days (which it is if you are using Date values) and that you have several years of data, the cross correlation should produce a (very) roughly sinusoidal form decreasing in amplitude as you move away from zero. The period of the sinusoid should approximate the ~365.24 days in a year. The offset of the minimum or maximum from zero will be the lag in days between the two signals.
If you do an FFT on each of the original signals, you will likely see at least one peak at around 0.002738 Hz which equates to ~ 1/365.25.
First of all truly thank you for your answer! I really appreciate the help!
Just one thing: my data are in months covering the period 1960 - 2100 so totally 1692 in number. Is it possible to adjust your answer but in months (not days)? Also, I have to do a FFT in order to see the 11-year solar signal (there should be a total 12.8 solar cycles in these 1692 months (or 141 years), since it's a phenomenon that lasts approximately 11 years (or 132 months). What should I expect in the FFT diagramm. Because it's in Hz... (Does it mean number of repetitions of the phenomenon in the whole 1692 months?).
Thanks once again and truly looking forward to your reply.
Tony
|
| greg |
Posted - 11/01/2012 : 2:02:48 PM
To do a Cross Correlation, select the two columns of data and choose Analysis : Signal Processing : Correlation.
Assuming that your X data units is days (which it is if you are using Date values) and that you have several years of data, the cross correlation should produce a (very) roughly sinusoidal form decreasing in amplitude as you move away from zero. The period of the sinusoid should approximate the ~365.24 days in a year. The offset of the minimum or maximum from zero will be the lag in days between the two signals.
If you do an FFT on each of the original signals, you will likely see at least one peak at around 0.002738 Hz which equates to ~ 1/365.25.
|
