MORSE Bugs

Bugs recorded for MORSE versions 11NOV04 onwards

Bug#20: Reported: 22 Aug 05 by Anu Dudhia; Symptom: Updated sweep altitudes ignored in constituent retrievals; Cause: When subroutine drvl2p reads in modified tangent altitudes in the pT .atm file in the *L2P section, it updates the profile levels but doesn't modify the L1C tangent altitudes.; Fixed: MORSE v3 (new version of drvl2p.for available as a patch for MORSE v2 if required).
Bug#19: Reported: 08 Aug 05 by Chiara Piccolo; Symptom: Divide by zero in or NaN in ChiSq statistics; Cause: Subroutine micbad checks for no useable measurements in a particular sweep but doesn't set the flag to skip the sweep if no measurements are found, causing a divide by (number of used measurements) to fail.; Fixed: MORSE v3 (new version of micbad.for available as a patch for MORSE v2 if required).
Bug#18: Reported: 27 Jul 05 by Stefano Corradini; Symptom: Stops unexpectedly with random garbled text printed to screen; Cause: Subroutine swpfwd calls fwdptb assuming that the latter has arguments FAIL (LOGICAL) and ERRMSG (CHARACTER STRING) (as in the RFM version rfmptb), but it doesn't. Consequently, FAIL and ERRMSG are undefined so if FAIL arbitrarily becomes TRUE the test for a fatal error is passed and the program stops, printing the (undefined) ERRMSG text to the screen. Clear? Anyway, MORSE just ends up stopping for no good reason.; Fixed: MORSE v3 (new version of swpfwd.for available as a patch for MORSE v2 if required).
Bug#17: Reported: 03 May 05 by Alastair Burgess (= RFM Bug#56 ); Symptom: Fails with "Segmentation Fault" when using heavy molecule cross-section (.xsc) files; Cause: Variable NTRI in subroutine TRIANG not SAVEd (this is the triangulation of the cross-section data in (p,T) coordinates).; Affects: All versions of MORSE; Fixed: v24MAY05
Bug#16: Reported: 01 Mar 05 by Alison Waterfall; Symptom: Stops with message
F-FWDXSC: Logical Error #1
Cause: Array IGSMOL which stores the absorber index for each molecule was not reinitialised for each microwindow in subroutine micgas.for, causing two different cross-section molecules to be stored in the same absorber index.; Affects: All versions of MORSE; Fixed: v17MAR05
Bug#15: Reported: 01 Mar 05 by Alison Waterfall; Symptom: No cloud detection if bad .xsc data files used; Cause: If an error is encountered reading in contents of a .xsc file the routine drvchk.for immediately exits (bypassing the remainder of the routine which would have performed the cloud detection), as intended, except that there was no FAIL flag set to convey this fatal error condition to the calling routine so the program continued as normal.; Affects: All versions of MORSE; Fixed: v17MAR05
Bug#14: Reported: 08 Mar 05 by Alastair Burgess; Symptom: As in Bug#13; Cause: As in Bug#13, but negative absorption coefficients can also arise if pT retrieval causes a pressure inversion. To avoid this in future, if the HYD Flag is selected, there is a check after each update of the hydrostatic equation and the pT retrieval stops if an inversion is detected with the following message printed to the screen and log file
F-SWPHYD: Negative altitude increment, retrieval abandoned
It's not elegant, but it'll do for now.; Affects: All versions of MORSE; Fixed: v17MAR05
Bug#13: Reported: 08 Mar 05 by Alastair Burgess; Symptom: "NaN" ("Not A Number") appears in MORSE output files and program gets into an infinite loop (although only on some machines).; Cause: Ray tracing inaccuracies near the tangent point can occasionally lead to negative absorber amounts, hence negative absorption coefficients which, if large enough, can cause floating overflows in radiative transfer calculation. Actually an RFM Bug (#55).; Affects: All versions of MORSE; Fixed: v17MAR05
Bug#12: Reported: 08 Feb 05; Symptom: morse.atm file set to climatology profile rather than retrieved profile (as in morse.rtv) if the last microwindow turns out to contain no usable data (eg due to cloud).; Cause: Atmospheric profiles set to climatology at start of each microwindow and usually partly overwritten with retrieval state vector but if the microwindow is not used the first step is done but not the second.; Affects: All versions of MORSE; Fixed: v17MAR05
Bug#11: Reported: 07 Feb 05; Symptom: Program stops with message
F-FWDWID: Unrecognised lineshape value; Cause: If a species uses LUTs for one microwindow but not the next, the SHPGAS variable in gascom.inc which is used to flag either the LUT or line shape to be used (defined in shpcon.inc) was not being correctly reset in micgas.for; Affects: All versions of MORSE; Fixed: v17MAR05
Bug#10: Reported: 03 Feb 05; Symptom: morse.res always being written, irrespective of whether RES flag is set in driver table or not; Cause: No check on RES flag in micswp.for before calling swpres.for to output residuals; Affects: versions from v25NOV04 onwards; Fixed: v03FEB05
Bug#09: Reported: 02 Feb 05 by Chiara Piccolo; Symptom: When using sweeps with non-standard nominal tangent heights can stop with Fatal Error message
F-MSEATM: retrieval level above top of forward model; Cause: Incorrect insertion of additional nominal levels into forward model altitude profile; Affects: all versions; Fixed: v03FEB05
Bug#08: Reported: 18 Jan 05 by Vivienne Payne; Symptom: When using Cloud detection and retrieving Offset, program can sometimes stop with Fatal Error message
F-MICAPR: No sweeps used for this MW'; Cause: Offset A Priori uncertainty based on average NESR value from usable sweeps, but average cannot be calculated if no usable sweeps (eg flagged as cloudy); Affects: all versions; Fixed: v19JAN05
Bug#07: Reported: 10 Dec 04; Symptom: Random profile levels occasionally flagged as cloudy; Cause: Cloud detection based on Cloud Index < 1.8 and altitude < 30km. However in some cases where the data is noisy the l1c.pro program calculates negative cloud indices which are set to 0.0 in the L1C file and these are misinterpreted as cloud.; Affects: all versions; Fixed: v10DEC04
Bug#06: Reported: 07 Dec 04 by Joanne Walker; Symptom: Floating Point Exception (ie array bound error) if MORSE is run with microwindows wider than 3cm-1; Cause: No check on required size of number of widemesh intervals NWID against array dimension MAXWID in rfmsiz.inc; Affects: all versions; Fixed: v07DEC04
Bug#05: Reported: 01 Dec 04 by Vivienne Payne; Symptom: RES Flag: incorrect residuals being written to morse.res, actually just the radiance spectra; Cause: Forward model values not being saved for output in module fwdvec.for; Affects: v25NOV04 - v29NOV04; Fixed: v01DEC04
Bug#04: Reported: 19 Nov 04 by Vivienne Payne; Symptom: Stops with message
F-FWDSAV: Logical error; Cause: Unexpectedly large differences between iterations for small absorber amounts when saving absorption coefficients.; Affects: all versions up to v18NOV04; Fixed: v19NOV04
Bug#03: Reported: 19 Nov 04 by Vivienne Payne; Symptom: Fails with floating zero divide if retrieving continuum with a priori correlation length set to 0 (CORREL=0 in *ASD section of driver table); Cause: no check on size of CORREL parameter before dividing in subroutine micapr.for; Affects: all versions up to v18NOV04; Fixed: v19NOV04
Bug#02: Reported: 18 Nov 04 by Alastair Burgess; Symptom: Fails on some machines with error message
F-MSEINI: Fatal error calling ACCLOG; Cause: unitialised logical FAIL in module acclog.for; Affects: v11NOV04 - v16NOV04; Fixed: v18NOV04
Bug#01: Reported: 16 Nov 04 by Chiara Piccolo; Symptom: Fails to load climatological profile for isotopic retrievals with error message of form, for example
F-DRVCLI: No o3 climatology; Cause: incorrect handling of ALLISO flag in module drvcli.for (and also in drvl2p.for); Affects: v08NOV04 - v11NOV04; Fixed: v16NOV04

Back to MORSE Index