Using AST to Estimate Surface Level
by
Torstein Pedersen
—
last modified
Aug 30, 2007 01:14 PM
We all know that pressure sensors on the sea floor can not be used as an accurate estimate of sea surface elevation due to atmospheric variations (can be as much as 0.3-0.5 meters). It therefore it has been asked on more than one occasion if the Acoustic Surface Tracking for the AWAC may be used to measure surface elevation. We have generally cautioned against this as variations of the speed of sound in the water collumn can lead to similar errors as the pressure sensor.
Thanks to some work conducted by MetOcean Engineers and Tremarfon can we now whistle a different tune .... well maybe a portion of the melody has changed.
It turns out that if we can assume two characteristics then we get a pretty darn good estimate of absolute distance to the free surface. The first characteristic is if the water is well mixed then the speed of sound is uniform and thus constant. The second criteria is if we have a good estimate for the salinity (user defined) and temperature (measured) then we can fairly accurately estimate the speed of sound in the water.
However if there is substantial layering and the temperature is quite different from the top to the bottom (where temp is measured) then errors will be seen in this estimate.
Generally speaking, places with lots of current have good mixing and are not as sensitive. Also more shallower places tend to have less gradient with tempatue as well.
The AWAC with AST in Australia is online. This is at the entrance to a large port so the depth of roughly 15-20 meters and therefore the water is probably reasonably well mixed.
According to Tim Waldby, MetOcean Engineers
Side Note Wave measurements are less sensitive because we are more interested in the change of distance to the surface and not the absolute distance. So 1% error in 50 meters of water with 1 meter waves means that the error for waves is 1 cm whereas the error for distance is 50 cm.
Thanks to some work conducted by MetOcean Engineers and Tremarfon can we now whistle a different tune .... well maybe a portion of the melody has changed.
It turns out that if we can assume two characteristics then we get a pretty darn good estimate of absolute distance to the free surface. The first characteristic is if the water is well mixed then the speed of sound is uniform and thus constant. The second criteria is if we have a good estimate for the salinity (user defined) and temperature (measured) then we can fairly accurately estimate the speed of sound in the water.
However if there is substantial layering and the temperature is quite different from the top to the bottom (where temp is measured) then errors will be seen in this estimate.
Generally speaking, places with lots of current have good mixing and are not as sensitive. Also more shallower places tend to have less gradient with tempatue as well.
The AWAC with AST in Australia is online. This is at the entrance to a large port so the depth of roughly 15-20 meters and therefore the water is probably reasonably well mixed.
According to Tim Waldby, MetOcean Engineers
| Quote |
| ...attached is GPA AWC & Handar Recorded Tide.jpg - a plot showing the recorded tide levels measured by an AWAC AST, and a Handar shaft encoder. The Handar is located within the inner harbour and the AWAC is at the channel entrance ~3km from the Handar. Of course, both have been calibrated in to a datum. You will notice the Handar is much smoother because it is located within the harbour area, whereas the AWAC is in open sea. it is thought that the AWAC indicates a more real situation. |
Side Note Wave measurements are less sensitive because we are more interested in the change of distance to the surface and not the absolute distance. So 1% error in 50 meters of water with 1 meter waves means that the error for waves is 1 cm whereas the error for distance is 50 cm.
Current state:
Being created
Hi
I am comparing estimates of the depth over an AWAC instrument obtained by surface tracking distance measured by AST and by the AWAC's onboard pressure sensor.
I attach a plot from a month-long deployment. I have computed the mean AST distance for each wave burst (1024 seconds every hour) and plotted this in blue. I have also plotted the pressure record (in dbar) which has been corrected so that initial and final (in air) pressures are approximately zero (<0.05 dbar). I have not corrected for atmospheric pressure flutuations during the deployment, but these appear to be slight.
The deployment is in well mixed salt water with a reasonable salinity estimate (35 psu). Water temperature is stable at 13 degrees C. I have assumed that 1 dbar is equivalent to 1m of salt water, I think this is reasonable.
Do you have any idea why I am seeing an approximately 0.5m difference between the pressure and distance signals? If I had to believe one of them, which one would you believe?
Many thanks
Giles
I am comparing estimates of the depth over an AWAC instrument obtained by surface tracking distance measured by AST and by the AWAC's onboard pressure sensor.
I attach a plot from a month-long deployment. I have computed the mean AST distance for each wave burst (1024 seconds every hour) and plotted this in blue. I have also plotted the pressure record (in dbar) which has been corrected so that initial and final (in air) pressures are approximately zero (<0.05 dbar). I have not corrected for atmospheric pressure flutuations during the deployment, but these appear to be slight.
The deployment is in well mixed salt water with a reasonable salinity estimate (35 psu). Water temperature is stable at 13 degrees C. I have assumed that 1 dbar is equivalent to 1m of salt water, I think this is reasonable.
Do you have any idea why I am seeing an approximately 0.5m difference between the pressure and distance signals? If I had to believe one of them, which one would you believe?
Many thanks
Giles
Current state:
Being created
hello Giles,
Sorry for not getting back to you sooner. I was initially stumped then somebody here pointed out to me that the conversion factor that you are using from pressure to depth is off a little. 1 dBar = 1 meter for fresh water but in salt water the pressure is a little more for 1 meter. A good number is 1026 kg/m^3 for salt water of 35 ppt salinity.
This would lead to 2.6% error in the estimate of depth for the pressure. For 18.5 meters of water depth this is pretty close to 0.5 meters, which is what we see.
Hope that helps,
Torstein
Sorry for not getting back to you sooner. I was initially stumped then somebody here pointed out to me that the conversion factor that you are using from pressure to depth is off a little. 1 dBar = 1 meter for fresh water but in salt water the pressure is a little more for 1 meter. A good number is 1026 kg/m^3 for salt water of 35 ppt salinity.
This would lead to 2.6% error in the estimate of depth for the pressure. For 18.5 meters of water depth this is pretty close to 0.5 meters, which is what we see.
Hope that helps,
Torstein
Current state:
Being created
Hi Torstein
Many thanks for your reply.
However, isn't density just part of the equation?
I figured that for seawater pressure increased as rho*g
1026 kg/m^3 * 9.81m/s^2 = 10065 Pa/m = 1.0065 dbar/m.
So I think the factor of 1 isn't very far out for seawater...
Any other thoughts?
Giles
Many thanks for your reply.
However, isn't density just part of the equation?
I figured that for seawater pressure increased as rho*g
1026 kg/m^3 * 9.81m/s^2 = 10065 Pa/m = 1.0065 dbar/m.
So I think the factor of 1 isn't very far out for seawater...
Any other thoughts?
Giles
Current state:
Being created
Hi again
Just a little extra info.
Attached is an identical plot for another AWAC instrument that was deployed nearby at the same time.
The results look almost identical - i.e. the depth over instrument from the pressure record is about 0.4 m greater than the distance given by the AST. So I think we can rule out any instrument specific faults of mis-calibrations.
As I see it, either:
a) I am doing something wrong in my mathematics (however doing a simple mean of the AST readings for each wave burst isn't TOO hard), or
b) the salinity estimate is poor - however I don't think this can be the answer because if the actual salinity was lower than estimated (it couldn't be higher) then the actual depth would be GREATER than than caluculated from pressure and would be LESS than that calculated by the AST - which just increases the discrepancy, or
c) there is consistent (between instruments) error in the calibration of the pressure sensors.
d) there is some consistent error in the distance calculation, or
None of those sound like very pleasant alternatives.
Does the AST measure distance from effectively the same point on the instrument as the pressure sensor?
Is the approach of taking a mean AST measurement over a wave burst reasonable?
Any other ideas?
Many thanks for your help.
Giles
Just a little extra info.
Attached is an identical plot for another AWAC instrument that was deployed nearby at the same time.
The results look almost identical - i.e. the depth over instrument from the pressure record is about 0.4 m greater than the distance given by the AST. So I think we can rule out any instrument specific faults of mis-calibrations.
As I see it, either:
a) I am doing something wrong in my mathematics (however doing a simple mean of the AST readings for each wave burst isn't TOO hard), or
b) the salinity estimate is poor - however I don't think this can be the answer because if the actual salinity was lower than estimated (it couldn't be higher) then the actual depth would be GREATER than than caluculated from pressure and would be LESS than that calculated by the AST - which just increases the discrepancy, or
c) there is consistent (between instruments) error in the calibration of the pressure sensors.
d) there is some consistent error in the distance calculation, or
None of those sound like very pleasant alternatives.
Does the AST measure distance from effectively the same point on the instrument as the pressure sensor?
Is the approach of taking a mean AST measurement over a wave burst reasonable?
Any other ideas?
Many thanks for your help.
Giles
Current state:
Being created
Hello Giles,
Sorry my mistake, I thought that 1 dbar was 1 meter in fresh water.
I am looking into this and hope to get back to you soon. But to answer your questions.
Yes, the Pressure reading and the AST distance has been calibrated so that they both have the same distance.
Yes, this should be fine for understanding the bias error. If you are just using the Pressure value from the sensor file then it is the value just before the burst, while the mean AST is the average over the burst. Therefore if there is a tidal signature, then the AST may appear slightly attenuated. But again this should not create any bias for such a long time record. Just out of curiousity, what is the source of the Pressure estimate? Sensor data, wave data?
Regards,
Torstein
Sorry my mistake, I thought that 1 dbar was 1 meter in fresh water.
I am looking into this and hope to get back to you soon. But to answer your questions.
| Quote |
Does the AST measure distance from effectively the same point on the instrument as the pressure sensor? |
Yes, the Pressure reading and the AST distance has been calibrated so that they both have the same distance.
| Quote |
Is the approach of taking a mean AST measurement over a wave burst reasonable? |
Yes, this should be fine for understanding the bias error. If you are just using the Pressure value from the sensor file then it is the value just before the burst, while the mean AST is the average over the burst. Therefore if there is a tidal signature, then the AST may appear slightly attenuated. But again this should not create any bias for such a long time record. Just out of curiousity, what is the source of the Pressure estimate? Sensor data, wave data?
Regards,
Torstein
Current state:
Being created
Hi Giles
This is nice work -- we are all looking forward to hearing more about the detailed results once the offset issue is resolved. As you know, the "error" in the tide=surface elevation (as derived from pressure) will come from variations in the density profile and from fluctuations in the atmospheric pressure. The "error" in the tide (as derived from AST) will come from variations in the speed of sound. With these two independent sources of information, our hope is that it will be possible to eliminate atmospheric variations in the pressure signal and even refine the combined tidal signal to give more accurate information. This may require a long time series but it looks like you may already have what you need to do this type of analyses.
As for the pressure offset, it could simply comes from the production system we have here. Because the pressure sensor cannot output negative values and because it is confusing when you see a zero output from the pressure sensor, we give the pressure sensor a positive offset of about 0.3 m during production. This can be adjusted later on by using the pressure offset function in the software.
In addition, the pressure sensor does not have linear response so in a 100-m sensor you can have absolute errors as large as 0.5 m (0.5% of full scale). The good news is that this absolute error will not affect (much) your ability to correctly estimate variations in the pressure.
Please let us know how it works out
" />
'>
Best regards,
Atle Lohrmann
This is nice work -- we are all looking forward to hearing more about the detailed results once the offset issue is resolved. As you know, the "error" in the tide=surface elevation (as derived from pressure) will come from variations in the density profile and from fluctuations in the atmospheric pressure. The "error" in the tide (as derived from AST) will come from variations in the speed of sound. With these two independent sources of information, our hope is that it will be possible to eliminate atmospheric variations in the pressure signal and even refine the combined tidal signal to give more accurate information. This may require a long time series but it looks like you may already have what you need to do this type of analyses.
As for the pressure offset, it could simply comes from the production system we have here. Because the pressure sensor cannot output negative values and because it is confusing when you see a zero output from the pressure sensor, we give the pressure sensor a positive offset of about 0.3 m during production. This can be adjusted later on by using the pressure offset function in the software.
In addition, the pressure sensor does not have linear response so in a 100-m sensor you can have absolute errors as large as 0.5 m (0.5% of full scale). The good news is that this absolute error will not affect (much) your ability to correctly estimate variations in the pressure.
Please let us know how it works out
" />'> Best regards,
Atle Lohrmann
Current state:
Being created
Hi Torstein
Thanks for your answers. The pressure data is from the SENSOR.
I have emailed you the location of the files on ftp, they are too large to post here.
Hi Artle
Thanks for your input.
Once I understand this depth offset I hope to have more interesting discussions with you about the wave processing - that is what I am really interested in. See you soon in the Wave forum I hope...
The pressure sensors on both of these instruments appear to read approximately 0.65 dbar when in the air. I subtracted this value from all pressure data. I did not attempt to correct for fluctuations in atmospheric pressure during the deployment.
I am beginning to suspect that the non-linearity of the pressure sensor is the culprit. I understand that this should have very little impact on the fluctuations in depth (which are a necessary part of the PUV wave processing) but does this mean that if I want to know the absolute depth of water over the instrument I would be better to carefully check the speed of sound calculation and then trust the AST data over the pressure data?
Many thanks to both of you for your help
Giles
Thanks for your answers. The pressure data is from the SENSOR.
I have emailed you the location of the files on ftp, they are too large to post here.
Hi Artle
Thanks for your input.
Once I understand this depth offset I hope to have more interesting discussions with you about the wave processing - that is what I am really interested in. See you soon in the Wave forum I hope...
The pressure sensors on both of these instruments appear to read approximately 0.65 dbar when in the air. I subtracted this value from all pressure data. I did not attempt to correct for fluctuations in atmospheric pressure during the deployment.
I am beginning to suspect that the non-linearity of the pressure sensor is the culprit. I understand that this should have very little impact on the fluctuations in depth (which are a necessary part of the PUV wave processing) but does this mean that if I want to know the absolute depth of water over the instrument I would be better to carefully check the speed of sound calculation and then trust the AST data over the pressure data?
Many thanks to both of you for your help
Giles
Current state:
Being created
Average AST-Average Pressure values
Statement: The average AST value has an added value!!
For a West African offshore location (12 m water depth) we have processed wave current and tidal data as measured with the AWAC. Knowing form previous projects we were aware of the daily airpressure fluctuatons.
The tidal analysis we have perfomed consisted of harmonic analysis of the average pressure data as measured during the current measurements (interval 15 minutes) as well as the average pressure and average AST data measured during the wave measurements (interval 60 minutes).
I did process the ast time series file to obtain the mean ast value. Note to my opinion it should be a standard output value
The reason for using the average ast value as well as the average pressure is to estimate the influence of airpressure fluctuations on the tide. The ast value ,although in general less accurate than the pressure value, is not influenced by the airpressure. The comparison of the tidal analysis results performed on both time series clearly shows this impact. At the moment also a weather station including airpressure is on site so we could verify if our previous analysis is correct. And it seems to be so (see figure below) As expected the level variation is in anti phase with the pressure fluctuation, further most of the differences between the pressure and AST tidal measurements can be attributed to air pressure fluctuations.
So other advantages of having the average AST value:
Comparing tidal amplitudes as deduced form the AST time with those from the average pressure time series enables you to verify that your salinity, temperature and density have been properly chosen. In the above described case the M2 amplitudes from both time series were approximately 50 cm and the differences between both time series were less than 2 mm.
>
Statement: The average AST value has an added value!!
For a West African offshore location (12 m water depth) we have processed wave current and tidal data as measured with the AWAC. Knowing form previous projects we were aware of the daily airpressure fluctuatons.
The tidal analysis we have perfomed consisted of harmonic analysis of the average pressure data as measured during the current measurements (interval 15 minutes) as well as the average pressure and average AST data measured during the wave measurements (interval 60 minutes).
I did process the ast time series file to obtain the mean ast value. Note to my opinion it should be a standard output value
The reason for using the average ast value as well as the average pressure is to estimate the influence of airpressure fluctuations on the tide. The ast value ,although in general less accurate than the pressure value, is not influenced by the airpressure. The comparison of the tidal analysis results performed on both time series clearly shows this impact. At the moment also a weather station including airpressure is on site so we could verify if our previous analysis is correct. And it seems to be so (see figure below) As expected the level variation is in anti phase with the pressure fluctuation, further most of the differences between the pressure and AST tidal measurements can be attributed to air pressure fluctuations.
So other advantages of having the average AST value:
Comparing tidal amplitudes as deduced form the AST time with those from the average pressure time series enables you to verify that your salinity, temperature and density have been properly chosen. In the above described case the M2 amplitudes from both time series were approximately 50 cm and the differences between both time series were less than 2 mm.
>
Current state:
Being created
Powered by
Ploneboard
Forgot your password?
New user?
