Question about IRIS pressure gauge data inconsistencies

Dear OOI:

I am Li Wang, a Ph.D. candidate working on a geophysics project associated with the Axial Seamount. Thanks for running the well-maintained database. The data I downloaded from IRIS is the vertical component of the broadband seismometer (BHZ) and differential pressure gauge (HDH) of station AXCC1 deployed from 2015.03.04 to 2020.12.31. The displacement-to-pressure ratio calculated from these data is far from synthetic predictions.

For benchmark purposes, I downloaded three instruments of the pressure gauge — HDH (200 sps), LDH (1 sps) from the IRIS, and Nano Bottom Pressure Recorder (20 sps) from the OOI Data Access & Visualization. I checked the differences between the three pressure data. After the same data preprocessing (remove the trend, the mean, taper, and a bandpass filter of 0.1-0.3Hz), we found the amplitude of three instruments are very different: The amplitudes of HDH and LDH are almost identical, both are much larger than Nano-BPR.

I suspect that I might mess up with the gain of these instruments, but neither the instrument responses of HDH and LDH downloaded from IRIS has an effective response for removal.

I am just wondering do you have any idea about how to deal with the instrument response of these pressure gauges or point me to someone who might know about it. I am grateful for your help and looking forward to your response.

Sincerely,

Li Wang

2021.09.02

WANG Li, Ph. D. candidate
School of Earth Science and Engineering
Nanjing University
No.163 Xianlin Rd, Qixia District, Nanjing, Jiangsu 210023
China
Tel: +86-158-9288-0403
dz1929007@smail.nju.edu.cn

1 Like

@AlexWang

Hi,

Thank you for the inquiry, and for all of the info on your pressure data investigation! The amplitude difference is concerning, but we’ll have to look into it more to say for sure what is causing the gap you’re seeing. Can you confirm for us which data you downloaded (instrument names and locations, and the start and end time of each data record)?

I’ll consult with some of the instrument experts and we’ll get back to you soon.

Thanks,
Mike Vardaro & UW data team

Hello Michael:

The data I used consist of three types of pressure gauges. I list their details below

Type 1: Broadband Differential Pressure Gauge Location: Regional Cabled Array at Axial Seamount; Station: Central Caldera(AXCC1); Instrument: HT1-90-U/Diff Hydrophone-DM-24 Mk3 Fixed Gain, g; Channel: HDH

Type 2: Short-period Differential Pressure Gauge Location: Regional Cabled Array at Axial Seamount; Station: Central Caldera(AXCC1); Instrument: HT1-90-U/Diff Hydrophone-DM-24 Mk3 Fixed Gain, g; Channel: LDH

Type3: Nano Bottom Pressure Recorder Location: Regional Cabled Array at Axial Seamount; Station: Central Caldera(AXCC1); Instrument: Bottom Pressure and Tilt; Node: Medium-Power JBox (MJ03F); Stream Content: NANO Sensor Data Products

The date range is different. I downloaded the HDH-channel data from 2015.03.04 to 2020.12.31 and all of the data obtained the amplification problem. Then I downloaded the other two kinds of data during a week for benchmark and found the amplitude order difference between the two DPGs and the Nano-BPR. In addition, Adrian K. Doran had confirmed the gain of the Nano-BPR is accurate in his dissertation (Doran 2019). Therefore, I doubt that it is the wrong gain of DPGs in which the misfit occurs.

Thanks for your help and looking forward to your reply.

Hi again,

Sorry your reply got stuck in the system! Thanks for the info - that’s useful. I was actually going to suggest talking to Adrian Doran, so that’s good to hear that you’re already in touch.

I sent a note to William Wilcock (UW) and Bill Chadwick (OSU) to see if they had any insights into the issue. Here’s some of the follow-up discussion:

First, what are the units on the x-axis of the plot you sent? I am guessing they are observations or data points, but what is the scale?

Dr. Wilcock also recommended checking out the Doran and Crawford (2020) Geology paper (link) since they looked at compliance (frequency dependent ratio of displacement to pressure) using the AXCC1 vertical seismometer and the absolute pressure gauge on the bottom pressure recorder (BOTPT) and they got sensible results.

He also said:

As for AXCC1 HDH it is not what I would call a differential pressure gauge (which is a Cox-Webb differential pressure gauge that will go down to frequencies of ~ 1 mHz)

HDH on AXCC1 is described at IRIS as a
HT1-90-U/Diff Hydrophone-DM-24 Mk3 Fixed Gain, g

The HT1-90-U is a hydrophone with a response from 2 Hz on up
HTI-90-U Hydrophone

I am not sure what differential mode means in terms of response. IRIS shows it as having a flat response down to 0.00001 Hz
IRIS: MDA : OO : AXCC1 : -- : HDH
but I have never heard of that for a hydrophone

I am going to look for the spec sheets from the manufacturer for the sensor in differential mode that shows its response. Dr. Wilcock thought that if it really did respond at such low frequencies, Adrian Doran and Wayne Crawford would have used it in their paper.

I’ll reply again if I find anything useful in the spec sheets. Let me know if you have follow-up questions.

Thanks,
Mike Vardaro
University of Washington

Hello Mike,

Thank you for your detailed reply and thanks for the help of Dr. Wilcock and Dr. Chadwick.

According to Dr. Wilcock’s answer, I found that only the APG data is available for my method by testing several sensors set at station AXCC1. Then I would replace the HT1-90-U Hydrophone data with APG data in my subject. However, I derived two kinds of APG data in my test.

The first kind of data was downloaded from OOI Data Access & Visualization (Type3 in my last response). These data were downloaded into CSV format and contained a 20Hz sampling rate and the whole 86400 seconds data length in each one-day file. As for the second type, they were downloaded from IRIS, which is documented as channel BDO. This type of data is SAC format and has the same sampling rate as the CSV format. But, every one-day file only has the first four hours in one day at most (showed as below).
image

I checked the two kinds of data to make sure that the two data types present the same data (as below, bandpass between 0.1-0.3Hz). The plot illustrates that two data types describe the same data of absolute pressure. But the data lengths are not identical when they are from the different data sources.
image

Furthermore, I tested two kinds of data with the same code. Results expressed distinct characteristics. In this test, I employed the same calculation code with an identical data-processing procedure. Two kinds of data have the same data length and amplitude (showed in the 2nd figure). Nevertheless, I cannot acquire the uniform output and I have no idea about this.
image

I wonder that what caused the different data lengths of the two kinds of data, and why would the same APG data lead to the diverse result.

Wish to see your answer soon.
Thank you so much.

Li Wang

Dear Mike,

I am sorry to bother you but I have sent my reply back to your last response on the HelpDesk 4 days ago. I got no answer from you till now. I wonder that does something goes wrong?

Furthermore, I have been downloading the APG data during which I need. I wrote some codes for the automatic download of the data due to a large number of files. However, I cannot access the data from the dataset website https://downloads-west.oceanobservatories.org/ through Python’s request module or the Pydap module. I tried to download by wget command and it did not work, either. All these methods would end by time out or connection failed. The only way I could download the data from the dataset website is a directly manual download. I want to know what caused this problem and is there a way to solve it.

Wish to see your reply soon.

Li Wang
2021.09.16

Hi,

We did receive the previous message, and we are actively investigating the issue. It’s not clear yet whether the time gaps and diverse results you’re seeing are originating on the OOI side or the IRIS side (which is handled by a different team), and some of the interpretation requires input from subject-matter experts. We will send a reply as soon we have more information.

Also, can you send a little more information about the data download errors you’re encountering? Which data were you attempting to download, how did you get to that “downloads-west” URL (I’m not familiar with that one), and what errors did you receive? There are definitely better and faster ways to access data than manual download, but the data center also just moved from the east coast to the west coast so there could be some bugs to work out, or inaccurate information on the website that needs to be updated. If you provide a few more details we can forward that question to the cyberinfrastructure team to investigate.

Thanks,
Mike Vardaro (and UW data team)

Hello Mike,

At first, I am grateful for your generous help to me.

I have access several times of the APG data. Each time I would receive a message from OOI Service for the dataset links (as below).

I used the Pydap module in Python to access these data in the first link “Thredds Data Server”. But the connection always failed (as below).

Then I tried to download by entering the second link “Direct Download”. These data is stored at the website Index of /async_results/dz1929007@smail.nju.edu.cn, with a huge number of files in a single directory. Thus, I would like to automatically access these data by traversing each file’s URL in this directory. I choose the Python module requests, but it did not work, either. (as below)

Finally, I tried the wget command to download these data. But this attempt failed again (as below).

Up to now, the only way I download these data is manual downloading one by one. But the large number of files brought a lot of inconveniences. I hope there could be some means to solve it.

Thank you so much

Li Wang
2021.09.17

下载视频

Hi Li,
.csv files are not aggregated by the system, so they will be produced in large numbers. We have provided an application that will assist you in automating the download process. The system will aggregate and try to control the file size for NetCDF files.

With regard to the IRIS data:

We looked into one of the issues you pointed out with the IRIS data, and we were able to find the full 24 hours of data you were referring to (BDO at AXCC1 on 2021-12-14). The data are in the system, but it seems to be stored in ~4 hour blocks (see below):
OO AXCC1 11 BDO M 2015-12-14T00:00:00.050000Z 2015-12-14T04:00:00.050000Z 2018-01-09T07:29:52Z
OO AXCC1 11 BDO M 2015-12-14T04:00:00.150000Z 2015-12-14T06:25:06.500000Z 2018-01-09T07:29:52Z
OO AXCC1 11 BDO M 2015-12-14T06:25:09.100000Z 2015-12-14T08:00:00.150000Z 2018-01-09T07:29:52Z
OO AXCC1 11 BDO M 2015-12-14T08:00:00.250000Z 2015-12-14T12:00:00.250000Z 2018-01-09T07:29:52Z
OO AXCC1 11 BDO M 2015-12-14T12:00:00.350000Z 2015-12-14T16:00:00.350000Z 2018-01-09T07:29:52Z
OO AXCC1 11 BDO M 2015-12-14T16:00:00.450000Z 2015-12-14T20:00:00.450000Z 2018-01-09T07:29:52Z
OO AXCC1 11 BDO M 2015-12-14T20:00:00.550000Z 2015-12-14T23:59:59.950000Z 2018-01-09T07:29:52Z

There are also some real data gaps due to interruptions in the time-series, e.g., (ds.iris.edu/cgi-bin/seismiquery/goat/goatframes.pl?TIMEQC=%2Fwww_dynamic%2Ftmp%2FSYNC%2Fsyncfile.12861.dmc3.1005&YEAR1=2015&JDAY1=348&hhmmss1=000000&YEAR2=2015&JDAY2=348&hhmmss2=235959&GAP=on&OVERLAP=on&CONTINUOUS=on&QUALITY=All&VIEWGRAPH=View+Timeseries+(GOAT))

IRIS has changed how some data access tools work with their databases, so it’s possible that you didn’t receive the full time range you requested due to a request error.

Thanks,
Jeff