{ "schema": "ausmt-station", "version": "0.1", "ausmt_id": "au.willyama-1982.13", "station": "13", "survey": "Willyama magnetotelluric traverse 1982", "survey_id": "willyama-1982", "country": "Australia", "organisation": "Geoscience Australia", "location": { "lat": -32.675479, "lon": 143.634269 }, "data": { "type": "BBMT", "n_periods": 23, "period_min_s": 0.05012, "period_max_s": 1258.004051 }, "diagnostics": { "median_relative_error": 0.527, "remote_reference": false, "tipper_available": false, "completeness_smoothness_diagnostic": { "value": 2.1, "basis": "e", "note": "not a quality or geological-value judgement" }, "classification": "indeterminate", "method": "phase-tensor (Caldwell 2004)", "note": "screening diagnostic, not an interpretation product" }, "processing": { "software": null, "algorithm": null, "remote_reference": false, "remote_site": null, "file_written_by": { "name": null, "version": null }, "note": "Magnetotelluric transfer functions minted from Cull, J.P. and Spence, A.G.\n(1985) 'Magnetotelluric soundings across the boundary of the Precambrian\nWillyama Complex and the Cainozoic Murray Basin, southeastern Australia',\nBureau of Mineral Resources, Geology and Geophysics, Report 250, Australian\nGovernment Publishing Service, Canberra. ISSN 0084-7100, ISBN 0 644 03986 8,\nBMR Microform MF208, Geoscience Australia eCat 15162. Apparent resistivity,\nphase, period and their two error columns come from TABLE 3 (PDF pages 26 to\n39, one page per site); positions and orientations from TABLE 1 (PDF page\n13); the recording bands from TABLE 2 (PDF page 14). Published\ninterpretation: Cull, J.P. (1985) 'Magnetotelluric soundings over a\nPrecambrian contact in Australia', Geophysical Journal of the Royal\nAstronomical Society 80(3): 661 to 675,\ndoi:10.1111/j.1365-246X.1985.tb05117.x, which prints no response table of\nits own.\n\nDIGITISATION. Every value here was read by code from 300 and 600 dpi page\nrenders by two OCR engines and the scan's own text layer, in\nDigitisation_MT-GDS_Data/surveys/willyama-bmr-1982/results/2026-09-18-run2/\nrun 2026-09-18-run2, amended in place on 18 September 2026. The two tables\nthat feed this mint are copied into GDS/GDS_AusMT/willyama-1982/ byte for\nbyte as responses.csv and sites.csv, so this generator reads nothing outside\nthe knowledge base. A cell no two readers agreed on is EMPTY in that\ntranscription and is EMPTY here: it is not written, not guessed and not\ninterpolated. The run's own adjudication file records every candidate and\nevery raw token for each such cell.\n\nWHAT TABLE 3 PRINTS, AND WHAT IT DOES NOT. Two listing blocks per page, one\nper orthogonal component, each with eight columns and nothing else:\nBAND | POINT | PERIOD | N | AP.RES. | SDLM | PHASE | SD\nThere is no impedance, no diagonal tensor element, no tipper, no coherence\nand no skew anywhere in the report, so ZXX and ZYY are written here as the\nEMPTY sentinel because they were never measured and NOT because they are\nzero. TABLE 1 prints five columns and no sixth: site number, 1:100 000 map\nsheet, latitude, longitude and an orientation. NO ELEVATION for any site is\nrecoverable from this report, so no ELEV is written. The vertical field was\nrecorded and used: page 11 cites Gamble and others (1982) for using the\nTipper to reduce the ninety degree ambiguity in the rotation angle. No\ntipper value is tabulated or plotted anywhere, so no tipper block and no HZ\nchannel is declared here.\n\nIDENTITY. One file per TABLE 1 site number, fourteen in all, and the\nstation id is that site number as the listing labels it, two digits. One\npair of sites is printed out of site-number order, 04 before 03, and each of\nthose two pages carries a field-location annotation naming the other one's\nnumber. On all fourteen pages the listing's AZ equals the TABLE 1\norientation of the SITE NUMBER printed on that page and not of the field\nlocation, so the site number is the identity and the field location is the\norder of occupation. Serving by field location would swap two sites.\n\nTHE FRAME. The heading of every listing page prints AZ, ROTATION and\nSTRIKE, and the report never says how the three relate. Measured on all\nfourteen pages of this run's own transcription, and re-measured here on the\nrows being written:\nSTRIKE = (AZ + ROTATION) mod 360 on 14 of 14 pages\nAZ = the TABLE 1 orientation of that site number on 14 of 14 pages\nTABLE 1's footnote 2 defines its orientation, as printed, as 'the magnetic\nazimuth of the x-coordinate of the original survey axes'. The rotation is\ntherefore applied to that x axis and the rotated x axis lies along the\nazimuth the listing calls STRIKE. This file is served in that rotated frame:\nthe ZROT block carries the printed STRIKE at every period, uniform for the\nstation, and every impedance block's ROT= header carries the same angle.\nNOTHING IS ROTATED BY THIS GENERATOR: the angle is a declaration of the\nframe the printed values are already in. The HMEAS and EMEAS azimuths carry\nthe printed AZ and AZ+90, which is the acquisition frame the rotation was\napplied to; they are acquisition metadata and the ZROT declaration is the\nstored-tensor frame. The report prints no electrode separation anywhere, so\nthe electric channels declare an azimuth and no dipole length.\n\nOWNER RULING, 2026-09-18, THE STRIKE DATUM: geomagnetic. The listing prints\nSTRIKE with the suffix M and TABLE 1's orientations are magnetic azimuths,\nso the served frame is declared geomagnetic and the rotation angles are\nserved as printed. Page 11 says 'All rotation angles were then calculated\nrelative to true north', and that sentence is recorded here as the text's\nstatement and is NOT applied: no declination is computed, quoted or applied\nanywhere in this package. The consequence for a consumer is the whole 1982\ndeclination at this traverse, which this package does not state.\n\nWHICH COMPONENT LIES ALONG STRIKE. Report 250 does not say. Its own figure\ncaptions read 'Resistivity pseudosection component 1 (ExHy)' and 'component\n2 (EyHx)' and stop there. The published interpretation settles it: Cull\n(1985), printed page 666, prints the same two pseudosections as one figure\nwhose panels carry those two component labels, panel (a) the component 1\n(ExHy) panel and panel (b) the component 2 (EyHx) panel, under the caption,\nas printed:\nFigure 3. Pseudo-section of apparent resistivity for (a) E along strike, and\n(b) E across strike.\nSo component 1, ExHy, is the mode whose electric field lies ALONG the\nprinted strike, and it is written here as Zxy with x along that strike;\ncomponent 2, EyHx, is Zyx. Report 250's own words agree without stating it,\npage 7: 'One component shows the resistivity in the direction of easiest\ncurrent flow, while a second, orthogonal (or perpendicular) component is\nused for data in the direction of greatest resistance', and 'The magnitude\nof the necessary rotation is then used to define the regional strike'.\n\nTHE PHASE CONVENTION, AND WHY IT IS AN INFERENCE. TABLE 3 is titled\n'APPARENT RESISTIVITIES AND PHASE DELAY EXPRESSED IN BAND AVERAGES FOR\nORTHOGONAL COMPONENTS AT EACH SITE IN TABLE 1.' (PDF page 26, printed page\n22) and the contents page calls the same quantity phase delay (PDF page 3).\nEvery printed phase in both components is negative. The report states no\ntime dependence, no tensor definition and no statement of which field is\nreferenced to which, anywhere, and neither does Cull (1985). A delay\nprinted as a negative number is the exp(-i.omega.t) sign of the impedance\nphase, which is the opposite of the EDI convention, and a printed table that\nprints both modes as the same kind of delay has dropped the 180 degrees that\nseparates arg(Zyx) from arg(Zxy). The mapping applied here restores both:\nphi_xy = -phi_printed phi_yx = -phi_printed - 180\nwhich is the conjugate of the printed impedance with the yx branch put back\nin its own quadrant. THIS IS AN INFERENCE, not a reading, and it is a\nruling owed. The identity mapping, the only alternative, would publish both\noff-diagonal elements in the fourth quadrant, which neither of them can\noccupy under the EDI convention. Measured on the cells this survey actually\nwrites: 263 of 263 written xy phases lie strictly inside the first quadrant\nand 274 of 274 written yx phases strictly inside the third. The acceptance\noracle is the AusMT engine's own sign-convention gate, run on these written\nfiles; its verdict is recorded in the package's provenance and is pinned by\nthe package test, not carried here as a literal.\n\nUNITS AND THE IMPEDANCE ARITHMETIC. Per printed row, with |Z| in mV/km per\nnT:\nT = printed PERIOD seconds\nrho_a = printed AP.RES. ohm metres, linear and NOT log10\n|Z| = sqrt(rho_a / (0.2 T)) mV/km per nT\nZ = |Z| (cos phi + i sin phi) phi from the mapping above, degrees\nThe period is in seconds: the sounding curves on PDF pages 19 and 22 label\ntheir abscissa PERIOD (SECONDS). The apparent resistivity is linear ohm\nmetres: it is the pseudosection CONTOURS, not the table, that the captions\nof figures 5 and 6 describe as Log10 R at 0.2 increments. FREQ is 1/T\nwritten to five significant figures.\n\nPERIODS ARE SERVED AS RECOVERED, by owner ruling of 2026-09-18. Page 7 says\nthe rotated data were 'averaged to 10 points per decade in non-overlapping\nbands'. The printed ladder is five: log10(printed PERIOD) against the\nprinted BAND index has slope 0.199998 and intercept -2.100007 over the run's\n613 band and period points, all 613 within 0.02 in log10. The ruling\nrecords the prose as the text's statement and serves the printed PERIOD. NO\nperiod in this file is reconstructed from a band index.\n\nUNCERTAINTIES, AND THE SECOND INFERENCE. The report never expands SDLM. It\noccurs only as a column heading over the twenty-eight listing blocks and\nnowhere else in the document, and it is zero on every one of the fourteen\nrows where N is 1, so it behaves as a spread statistic and not as an\ninstrument constant. It is READ HERE as one standard deviation of log10\napparent resistivity, which is what the Bowen Basin package reads its own\nprinted sigma as. THAT READING IS AN INFERENCE and it is a ruling owed: if\nSDLM is instead a two standard deviation interval, every variance in this\nfile is four times too large. Propagation, identical to the Bowen Basin\npackage's:\nsigma_ln|Z| = ln(10) * SDLM / 2\nsigma_|Z| = |Z| * sigma_ln|Z|\nZXY.VAR = sigma_|Z| squared likewise ZYX.VAR\nWhat is written is a MAGNITUDE variance mapped onto a complex element. It\nis not a full covariance and it does not constrain the phase: a consumer\nweighting phase by it will over-weight that phase. Where an impedance is\nwritten and its SDLM cell is empty the variance is written as the literal\ntoken nan, which says the uncertainty is unknown; an EMPTY there would read\nback as an exact zero error on a present value, which is an infinite weight.\n41 variance cells in this survey are written nan.\n\nA PRINTED SDLM OF ZERO IS ALSO WRITTEN nan, and that is a ruling. 13 of the\nwritten impedance cells in this survey carry a printed SDLM of 0.000, and\nevery one of them prints N = 1: the spread of a single estimate is not an\nuncertainty of zero, it is no estimate of spread at all, and the mint\nrefuses to publish it as a variance of zero because that is an infinite\nweight. The condition is enforced and not assumed: a printed zero SDLM\nbeside any other N stops this mint. The printed 0.000 survives in the\ndigitisation tables and in the package's provenance.\n\nTWO PRINTED COLUMNS ARE NOT IN THIS FILE. The listing's SD column is the\nphase's own standard deviation in degrees, and the EDI blocks this generator\nwrites carry no phase error, so no SD value is written here; it is carried\nin the package's provenance instead, cell by cell, and is not lost. The N\ncolumn is the number of estimates the band average was taken over; it is\nprovenance and not a transfer function value, and it is carried the same\nway.\n\nCOORDINATES. TABLE 1 prints latitude and longitude to three decimals of a\ndegree and names NO datum. OWNER RULING, 2026-09-18: the datum is taken as\nAGD84, the newer of the two candidates for 1982. The printed values are\nconverted here to WGS84, which is what this corpus serves and what a\nconsumer reading a served LAT/LONG assumes, by\naxis order change (2D) + AGD84 to WGS 84 (7) + axis order change (2D)\nas selected by pyproj from EPSG:4203 to EPSG:4326, stated accuracy 3 metres.\nA better operation exists and is NOT used: AGD84 to WGS 84 (9), stated\naccuracy 2.9 metres, needs a transformation grid this machine does not hold,\nand nothing is fetched at mint time. The printed values are kept beside the\nconverted ones in this file's own station note below, in the kit's\nstations.csv and in the package's provenance. The shift is about 207 metres\nat this traverse, and three decimals of a degree is about 111 metres, so the\nconversion moves a station by about twice the quantum its printed position\ncarries: it is not a rounding and it matters. The ruling is a ruling and\nnot a reading; if the datum is ever ruled AGD66 instead, every position in\nthis package moves again.\n\nEMPTY, AND HOW IT READS. The EMPTY sentinel is 1.0E+32. mt_metadata\nconverts it to an exact complex zero on read rather than to a NaN, so an\nexact complex zero in an impedance element here is the sentinel and NOT a\nmeasurement; AusMT's own reader classifies an exact complex zero as absent\nfor that reason. No printed phase in this survey lies on 0, plus or minus\n90 or plus or minus 180 degrees and every printed apparent resistivity is\nfinite and positive, so no real value here can collide with the sentinel. A\nmode is written only where its apparent resistivity AND its phase were both\nrecovered. 76 printed apparent resistivities in this survey stand beside a\nphase cell no two readers agreed on, out of 613 printed apparent\nresistivities in all, and those cells are EMPTY here rather than published\nwith a guessed phase. Across the survey 36 rows carry the xy mode alone and\n47 the yx mode alone and 227 carry both, out of 310 printed rows whose\nperiod was recovered; the further 11 rows serve neither mode and are not\nwritten.\n\nRIGHTS. The report prints '(c) Commonwealth of Australia 1985' and states\nno licence of any kind: no Creative Commons notice, no permission to\nreproduce, no reuse clause. OWNER RULING, 2026-09-18: the package carries CC\nBY 4.0 under Geoscience Australia's publications policy, which releases the\nBureau of Mineral Resources' legacy publications under Creative Commons\nAttribution 4.0, and cites Report 250, eCat 15162, ISSN 0084-7100 and ISBN 0\n644 03986 8 as the licensed work. Geoscience Australia is the successor of\nthe Bureau of Mineral Resources and is the custodian of record, ROR\nhttps://ror.org/04ge02x20.\n\nCAMPAIGN. The fieldwork was carried out over a five-week period in July to\nAugust 1982 (page 7). No per-site date is printed anywhere, and a five-week\nwindow is not a date, which is why ACQDATE is absent rather than filled with\na boilerplate value. The traverse ran about 320 km from Broken Hill\neastward across the Murray Basin margin, with an average site separation of\nabout 15 km from Broken Hill to Menindee. TABLE 2 prints the seven BMR\nrecording bands, 0.001 to 40 Hz, sampled from 5000 ms down to 4 ms at 1024\nor 2048 samples a band.\n\nSTATION: TABLE 1 site 13, listing page 38 of the PDF. TABLE 1 prints its\nposition as 32.677 degrees south, 143.633 degrees east, which on the ruled\nAGD84 datum converts to -32.675479, 143.634269 on WGS84, a move of 207\nmetres. The 1:100 000 map sheet is NOT recovered: the readers disagreed on\nthe printed name, so this file states no sheet rather than a guess. Heading\nof the listing page, as printed: AZ8, ROTATION : 10, STRIKE : 18 M, and 8 +\n10 = 18 on the page's own arithmetic. TABLE 1's orientation for site 13 is\n8, the same as the page's AZ.\n\nThis file carries the 23 printed periods that hold a value in at least one\nmode, from 0.05012 s to 1258 s. Of those 13 carry the xy mode and 22 the yx\nmode, so 1 carry xy alone and 10 carry yx alone; the absent mode is written\nas the EMPTY sentinel and is absent, never zero. 1 variance cell here is\nwritten nan, the uncertainty being unknown rather than zero. 7 further\nprinted apparent resistivities on this page are NOT published: their phase\ncell was not recovered, and an impedance cannot be formed from a magnitude\nalone.\n\nPRINTED ROWS NOT CARRIED. 3 listing rows on this page print a response\nwhose PERIOD cell was not recovered: component 1, listing row 1 (BAND not\nrecovered, POINT 1), component 1, listing row 2 (BAND not recovered, POINT\n2) and component 1, listing row 23 (BAND 26, POINT 23). A row with no\nperiod has no place on the frequency axis, so it is absent from this file\nrather than placed at a period reconstructed from its band index. Their\nprinted values survive in the digitisation tables named above.\n\nThis file is published as station au.willyama-1982.13. Dates in this file\nare ISO 8601." }, "distribution": { "edi_available": true, "license": "CC-BY-4.0", "edi_path": "edi/willyama-1982/13.edi" }, "provenance": { "pipeline": "ausmt/extract.build_portal", "pipeline_version": "0.2.1", "extractor": "mt_metadata (community canonical)", "software": { "python": "3.12.14", "mt_metadata": "1.0.10", "mth5": "0.6.8" }, "git_commit": "e208b15fb70740d8c44de51c532e6debd59c4e24", "parameters": { "dimensionality": { "beta_per_period_deg": 3.0, "skew_3d_deg": 5.0, "pct_periods_3d_threshold": 40, "ellip_2d_deg": 0.1, "min_rez_row_sine": 0.01, "beta_physical_cap_deg": 15.0, "min_usable_period_frac": 0.5, "skew_aggregation": "median" }, "diagnostic": "completeness/smoothness (median rel error + coverage + smoothness)" }, "generated": "2026-09-18T14:15:57.818603+00:00", "input_file": "13.edi", "input_sha256": "4f111a9c946bdbca85d8993b0fc76fca5e71032d7e6f7b06f5da0d07d7e287fa" }, "coordinate_qc": null, "canonical_conditioning": [ "survey.id->willyama-1982", "survey.project->Willyama_magnetotelluric_traverse_1982", "survey.geographic_name sanitized", "station.geographic_name sanitized", "sign_convention: '+' is a library default — EDI carries no machine-readable sign convention; NOT asserted by source", "declination.epoch/model: library defaults (epoch=None, model=IGRF) — an EDI states only the declination value; epoch/model NOT asserted by source", "channel orientations: degenerate/defaulted (ex and ey azimuths both 0 deg — parallel dipoles are non-physical); orientations NOT asserted by source EMEAS geometry", "citation.authors<-survey_meta:J.P. Cull; A.G. Spence", "citation.title<-survey_meta", "copyright.release_status=Data Citation Required; conditions_of_use set from licence CC-BY-4.0 (replaced mt_metadata default boilerplate)" ], "frame": { "convention_check": { "detail": null, "n_periods_used": 12, "phs_xy_median_deg": 29.81, "phs_yx_median_deg": -150.93, "verdict": "ok" }, "declared_azimuth_deg": 18.0, "derotated": false, "evidence": { "branch": "mt", "emeas_azimuths": { "ex": 8.0, "ey": 98.0 }, "hmeas_azimuths": { "hx": 8.0, "hy": 98.0 }, "rotspec": null, "trot": null, "zrot": "18" }, "frame_served": "declared-azimuth", "impedance_rotation_deg_source": null, "tipper_rotation_deg_source": null, "survey_frame_note": "frame: mixed declared frames across stations: 8°…84° — each station is served in its own declared acquisition frame; this survey does not share one frame (nothing de-rotated)" }, "resources": [ { "id": "edi", "kind": "transfer_function", "format": "edi", "provenance_role": "source", "representation_role": "original", "path": "edi/willyama-1982/13.edi" }, { "id": "emtfxml", "kind": "transfer_function", "format": "emtfxml", "provenance_role": "derived", "representation_role": "alternate", "path": "xml/willyama-1982/13.xml" }, { "id": "mth5", "kind": "transfer_function", "format": "mth5", "provenance_role": "derived", "representation_role": "alternate", "path": "h5/willyama-1982/13.h5" }, { "id": "edi-zip", "kind": "archive", "format": "zip", "path": "bundles/willyama-1982-edi.zip" }, { "id": "xml-zip", "kind": "archive", "format": "zip", "path": "bundles/willyama-1982-xml.zip" }, { "id": "survey-mth5", "kind": "archive", "format": "mth5", "path": "bundles/willyama-1982-tf.h5" } ] }