Hi
I was hoping someone on this list may be able to help with a stats problem I've encountered trying to estimated error due to noise in a sumpled signal. If I'm directing this question to the wrong list then I apologise in advance.
I'm measuring time series data of flow velocities. Each velocity estimate is the average of a sample containing at least n*5 measurements, where n>=1. The 5 samples are not independent, but each group of 5 samples is independent. I can theoretically estimate the standard deviation resulting from measurement noise for each measurment (SDm). Can I assume that the standard error of the velocity estimate is equal to: SDm/(n*5)^0.5, even though the samples are not all independent? Similarly, can I assume that the standard error of the standard deviation will also be: SDm/(2*(n*5))^0.5 ?
A second and related problem is that I want to calculate the standard error of the mean for a complete time series due to measurement noise (SEnoise). Can I derive this from the standard deviation/error for a single velocity estimate given that I know the number of samples in my time series (M) i.e. SEnoise=SDm/(M*n*5)^0.5 Or can I only calculate the standard error of the time series based on the standard deviation of the all the velocity estimates and the number of velocity estimates. Again is it possible to calculate the standard error due to noise for the standard deviation of the complete time series.
Please email me offlist if you prefer.
Thanks
Stuart McLelland
************************************************************
Stuart McLelland email: [log in to unmask]
Department of Geography
Exeter University
Exeter
EX4 4RJ
Tel: +44 (0)1392 263912
Fax: +44 (0)1392 263342
************************************************************
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|