search

smousavi05 / STEAD

STanford EArthquake Dataset (STEAD):A Global Data Set of Seismic Signals for AI
Creative Commons Attribution 4.0 International
278 stars 67 forks source link

wrong label data of back_azimuth_deg #8

Closed Seismic-wave-Han closed 2 years ago

Seismic-wave-Han commented 2 years ago

Hello. I'm Sangwoo Han in Seismology, Geophysics and Tectonophysics Lab in Seoul National Univ. in South Korea. First of all, thank you for your effort to make a very cool global benchmark dataset.

I found some problems in the label data of 'back_azimuth_deg'. There are some 180-degree differences between 'back_azimuth_deg' in STEAD and back-azimuth obtained using locations of source and receiver. I download STEAD data from mega cloud link. I attached an image of back-azimuth differences. figure

Also, I gave you one example of this problem.

I used function gps2dist_azimuth() from obspy.geodetics.base to get back-azimuth. I also want to know the formula that you used to calculate 'back_azimuth_deg' Please check this problem and could you reply to this?

smousavi05 commented 2 years ago

Hi @Seismic-wave-Han , you are right. The reported value for 'AKGG.AV_20060814223729_EV' is azimuth not the back azimuth.I used the same function from Obspy. And this is how it was calculated during the compiling of the dataset:

            distanceKm, azimuth, back_azimuth = obspy.geodetics.base.gps2dist_azimuth(
                                                                        float(event_lat), 
                                                                        float(event_lon),
                                                                        float(station_lat), 
                                                                        float(station_lon), 
                                                                        a=6378137.0, 
                                                                        f=0.0033528106647474805)
            distanceKm = (distanceKm/1000);
            distance = round((distanceKm / 111), 4);
            back_azimuth = round(back_azimuth, 4)

However, I haven't figured out what went wrong yet. I checked the station lat and long and they are correct. So no flipping of the event-station coordinates. Could you tell me how many of these you found in the dataset?

Seismic-wave-Han commented 2 years ago

Hi @smousavi05. I prepared some figures related to the error of back-azimuth (also, distance) baz_error baz_error_hist These figures are the back-azimuth error along with whole samples and their histogram. Samples with errors in the 170-190 degree range are 8% of the total.

I also found the distance discrepancy with obspy. dist_error The error of distance is nearly zero, but it shows quite a large error for some samples. I hope this information will help you find systematic errors in STEAD.

Rohit-Tomar commented 2 years ago

I have downloaded all the chunks of the dataset and trying to run the code but not getting the image folder in the dataset. Can you please help

smousavi05 commented 2 years ago

I am not sure what are you referring to? Which image folder?

On Thu, May 5, 2022, 11:45 AM Rohit-Tomar @.***> wrote:

I have downloaded all the chunks of the dataset and trying to run the code but not getting the image folder in the dataset. Can you please help

— Reply to this email directly, view it on GitHub https://github.com/smousavi05/STEAD/issues/8#issuecomment-1118930169, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADHWAPAFCAEGJU35ZUIGXJTVIQJMDANCNFSM5I4D4R3Q . You are receiving this because you modified the open/close state.Message ID: @.***>