3. Relative Humidity Calibration
Procedure
These was done similar to the temperature calibrations.
I compared the relative humidity (with respect to liquid water)
readings from the file Ola created for
each of the calibration periods and levels. I plotted
RH1 - RH2,3,4,5 vs. RH1 and RH5 - RH4 vs. RH5
Here is an example of one of these
plots (Figure 7.). I could not discern any change in the temperature differences
as a function of relative humidity (what I will call a "slope" error). But
there were systematic average differences (I call "bias" error) that
seemed to be a function of time.
The median relative humidity difference for each level and each period
were calculated. Detailed notes for each period and level are
here.
The bias errors
were usually less than 1% absolute value. Period 1 was an exception.
Level 1 and Level 5 were about 2-3% lower than the other levels during this
period.
Unlike temperature there were no humidity flux measurements analyzed when
I did this study. Therefore I was unable to independently verify the
above biases nor check the nature of the temporal variability.
Results
I plotted a summary of the median calibration period biases and the recommended
correction for all levels(Figure 8).
This comparison tells us the relative biases between the different levels. But it
does not tell the absolute values of the relative humidities. As with
temperature, I assumed that the mean values of the biases from all levels
represented the "true" relative humidity. For periods when calibrations
were not available (such as period 2.5) I used a linear fit to the nearest
0.01% of the surrounding calibration period values.
The overall biases for each level (compared to level 1)
were:
| Period 1 |
Period 2 |
Period 2.5 |
Period 3 |
Period 4 |
Period 5 |
Period 6 |
| -0.72
| -0.29
| 0.02
| 0.22
| 0.54
| 0.39
| 0.27
|
I'm not claiming a RH accuracy of one thousandth of a percent, but I saw
no reason to eliminate digits.
Conclusions and Recommendations
All of the biases were quite small. I would estimate that
the relative accuracy (standard deviation) of our tower RH values, after the recommended correction
to be about 0.5%. I would use 1% as a 95% confidence interval value.
But as we discussed over the email earlier, I think that all of our
relative humidities were too high when I was there at the end of the summer.
Ola and Ed did not share this opinion. To me it just did not
seem reasonable to have a 98% humidity when the cloud deck was over 500
meters high. And when the SPO replaced their sensors they got lower
values than ours (tower and PAM). So this is an issue that needs more
work. But I do think that our relative (between levels) values are reliable to the above
accuracies, except perhaps near the end of the experiment when the calibration
with the lower level was not useful becasue the values were "pegged" at 100%.
Click here to view and/or download
the relative humidity correction MATLAB program Note that this program
does not correct mixing ratio, q, or relative humidity wrt ice, RHi. This
should also be done.
|
|
|
| Outline Page |
Temperature Calibrations |
Wind Calibrations |
Last update: 5/4/99
Please send all comments and suggestions to the author, Peter
Guest,