Jump to content
Julia Austin

AWEA Windpower 2009 Presentation "Investigation of the NRG #40 Anemometer Slowdown"

Recommended Posts

Can you elaborate on the specifics for the quantitative statistics used to identify sensor error? 

For clarity:

mean bias = difference between two means of the (10 minute) averages for a 7 day period

ratio = ratio of the means of the (10 minute) averages for a 7 day period

correlation coefficient = corr coeff between the 2 sets of (10 minute) average wind speeds over the 7 day period

st dev wind speed ratio = std dev of the ratios of the (10 minute) average wind speeds over the 7 day period

 

Thanks in advance for the help!

Share this post


Link to post
Share on other sites

You will need to analyze several months of data ? at least 4 ? 8 months, ideally ? in order to detect a potential problem. 

 

You?ll need to compare wind speed data from pairs of sensors at the same height on the tower, excluding any wind speed data potentially affected by tower shadowing or shadowing due to nearby structures and/or terrain features.  Your best bet may be to contact which ever consulting company you normally work with for data analysis as they have the expertise in scrutinizing the data at this level of detail.  They are likely already familiar with the problem and how to find it in the data, and they may also be able to provide some recommendations about how to treat any potentially affected data.

 

When we have been looking at our research data, we were sorting data from redundant, same height, un-shadowed sensors as follows:

Mean Bias:

0.2

Ratio:

1.02

Ratio SD:

> 0.02

R Value:

 

However, this is not based on some established standard, nor are we advocating these as appropriate criteria for any and all data sets.  These criteria, for example, do not take into consideration the effects of complex terrain.  Unfortunately the nature of this problem and the vast variation among sites makes it very difficult to recommend a standard way of dealing with the data.

 

I hope this helps.

 

Share this post


Link to post
Share on other sites

Yes, thank you but the response does not answer my original question.

 I'm looking for more description of exactly how the four quantiative stats above are calculated; especially for ratio SD... do you calculate the ratio of each 10 minute average over the 7 day period and then take the standard deviation of that or some other method?  Please if you can provide clarity on this that would be great.

 Thanks!

Share this post


Link to post
Share on other sites

Hi  - Here is some further clarification.

 

Mean Bias

Mean bias is defined as the average of the 10 minute wind speed differences between two anemometers over each week.  The second sensor in the pair is subtracted from the first for each 10 minute interval.  Those results are then averaged over the week.  That result is the mean bias for the week.

Ratio

 

Ratio is defined as the average of the 10 minute ratios between two anemometers over each week.  The first sensor in the pair is divided by the second sensor for each 10 minute interval.  Those results are then averaged over the week.  That result is the ratio for the week. 

 

 

Correlation R

 

 

R is the statistical correlation of the 10 minute averages for each week, where Ch1 is the X parameter and Ch2 is the Y parameter. 

 

 

Ratio SD

 

Ratio SD is the statistical standard deviation of the 10 minute ratios for each week.  

 

 

 

Share this post


Link to post
Share on other sites

Join the conversation

You are posting as a guest. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

×
×
  • Create New...