changes.md

NOTE: Changes in the code can be due to

• a) errors (ERROR)
• b) new features that offer major new options/functionality (NEW)
• c) program technical improvements with no effect on results (TECHNICAL).

The term in brackets is used to mark each change.

8 September 2016: (NEW, R. Hock)

• v3.0
  • Added new retreat parameteriation (V-A scaling, Huss-parameterization)
  • small non-significant changes in many files
  • adjusted input.txt (simplified dates of winter/summer balance; added name of thickhness file)

29 January 2015: (ERROR, TECHNICAL L. Gillispie)

• v2.1.2
  • The makefile now detects whether or not the underlying kernel is 32-bit, and compiles accoringly.
  • Moved the function statistdouble from snowmodel.c:52 to writerout.c:841, An existing, and older function statist with the same purpose has exitsed in writeout, having both in the same file makes morelogical sense.
  • Fixed a type-mistatch bug related to statist being called when statistdouble should have been.
  • Fixed a bug wherin the energy model was attempting to wrtieout unallocated grids, and was causing the enegry model to segfault.

21 January 2015: (ERROR, L. Gillispie)

• v2.1.1
  • fixes an un-allocated pointer error which caused segmentation faults in certain circumstances

15 January 2015: (NEW, C. Reijmer)

• v2.1.0
  • changes in subsurface model: renmaing variables and writing more variables to output so that the water and energy balance can be tracked better.

6 November 2013: (NEW, R. Hock)

• v2.0.2

• change in output file "modelperformance.txt", discharge volume discharge is set to -9999 if discharge measurements contain missing values during simulation period (if nstepsdis < nsteps)

5 November 2013: (ERROR, R. Hock)

• v2.0.1

• ERROR fixed in computing number of valid time steps of discharge data. Affects r2 slightly and cumulative measured discharge volume in discharge output. (Version number not changed).

10/31/2013 (NEW, ERROR, R. Hock)

• v2.0.0

• CHANGE IN input.txt: all parameters removed that had to do with parameter optimization (between discharge and snow model parameters)
• NEW OUTPUT: time series of glacierwide ice heat flux, mass balance and cumulative mass balance added to 'areamean.txt'
• NEW OUTPUT: file with modeled and measured point balances ('pointbalances.txt'). A file with measured balances (measuredpointbalances.txt) is read and the program computes the pointbalances for the same locations and periods each stake's mass balance refers to.
• ERROR FIXED: cumulative mass balances for stakes (cummassbal.txt) was reset to 0 at end of mass balance year. Fixed so that it is continuous.
• ERROR FIXED: precipitation correction factor was not applied when precipitation was read from file for energy balance model.
• TECHNICAL: filenames of default output files changes: melting.txt → cumablation.txt, specificmassbal.txt → seasonalmassbal.txt, res.out → modellog.txt
• TECHNICAL: NEW: Longwave incoming radiation grid array is used even if LWin is spatially constant

10/20/2013 (NEW, R. Hock)

• v1.1.0

• Added "model performance" output file which records one line of discharge r^2 and r^2 (ln) values and other parameters to the text file "model_performance.txt"

10/07/2013 (TECHNICAL, R. Hock)

• v1.0.1 Some values of the 'hidden options' in variab.h have been changed to reflect a more general case.

09/20/2013 (ERROR, C. Reijmer)

• v1.0.1 Bug fixed in the skin layer formulation responsible for the lack of closure of the surface energy balance as found in the output.

• Added MBsum in order to have mass balance output consistent with other mass balance parameters output.
• Made changes (together with Torbjorn Ostby) to functions meltlayermice and resetgridwinter to remove some bugs related to the sr50 output. Correspondence between mass balance output and sr50 has improved.
• Made changes to functions subsurf and surftempskin to correct a bug resulting in wrong surface temperature to be used to calculate the sensible and latent heat flux. As a result energy balance was not 0 for T<0. This is fixed now.

09/18/2013 (L. Gillispie)

• v1.0.0 Versioning started

09/11/2013 (TECHNICAL, L. Gillispie)

• Switched build system from CMake to Makefile.

08/26/2013 (TECHNICAL, L. Gillispie)

• File and path name-lengths have been made uniform. There are three preprocessor macros, located in 'variab.h' and 'variabex.h' called:
  • FNAME_LEN : Length of filenames. Set currently to 80.
  • PNAME_LEN : Length of paths/directory names. Set to 201.
  • PFNAME_LEN : Max length of fully qualified path+filenames, this should be set equal to FNAME_LEN + PNAME_LEN, currently set to 281. To declare a new string, for say a filename, do the following:

    char filename[FNAME_LEN];
    char pathtofile[PNAME_LEN];
    char fullfilename[PFNAME_LEN];

6/21/2013 (NEW, R. Hock)

• Added added glacierwide cumulative balances to areamean.txt; increase array sizes of names in variab.h Purpose: allow comparison e.g. with GRACE time series; no change in input.txt, column is added to areamean.txt; Before: mass balance only computed if if( ((winterbalyes == 1) && (summerbalyes == 1)) || (maxmeltstakes > 0) ), NEW: if condition removed.

5/23/2013 (NEW, C. Reijmer)

• Subsurface model: Added option to use skin-Temperature calculations in computing surface temperatures. Use of this option requires changing paramters in variab.h and recompile code ("hidden parameters").

11/4/2012 (TECHNICAL, L. Gillispie)

• Fixed mismatching declarations in headers.

8/31/2012

• Configuration filenmae changed from input.dat to input.txt.
• Added sample shading.txt shading configuration file.
• Added Matlab visualization functions inputreadagrid.m and grids2diff.m,
• Updated Matlab vis tool ``grid.m```

8/28/2012

• Added sample ```input.dat``
• Added Matlab grid visualization tool.

Major changes since 11/2011 (L. Gillispie)

• Github hosting!

• Executable rename:

  • degree is now known as the Distributed Enhanced Temperature Index Model (DeTIM), and has executable detim

  • meltmod is now know as the Distributed Energy Balance Model (DEBaM), and has executable debam

• Build system in place! MeltMod now uses cmake to configure its builds.

  • Adds portability, cmake generates makefiles with the current environment in mind.
  • Out of source builds: Cmake will build everything in a folder outside of the source tree, and thus keep the source tree unmuddied by built files.
  • All of the relevant tools: ascigrid/gridasci/shading can be build via the same cmake build system
  • That being said, there have been some issues with cmake. In particular, since meltmod's code contains many circular dependencies, these can cause the model to fail to build unexpectedly. (See issue #16 for some discussion of this)

• Multiple headers: every source file now has its own header file, which is included where needed.

  • Speeds up compile time when changes occur only in a few files
  • Makes the control structure of the program more clear
  • Allows us to make certain functionality more modular

• Bug fixes: closed the following issues

  • issue #10 : readprecipindexmap was reading the wrong file, this is fixed
  • issue #11 : shayes is turned off when executing the temperature index model
  • issue #12 : methodsurftemp is immediately set to 1 when running the temperature index model, this was happening later, and screwing up some initialization parameters
  • issue #13 : fixed typo in snow-cover initialization
  • issue #14 : fixed bug in simulation mode, wherein large arrays were getting allocated but never used

Which brings us to...

• Project Reorganization:

  • All model source files are now located in the src/ folder in the project root
  • Utility programs related to the project are located in the util/ folder
  • Tests and their related file are located in the test/ folder

• Code reorganization

  • All functions relating to matrix and tensor manipulation have been moved into it's own file: grid.c/h and included where necessary. Before, these were spread through a few different files, having them in one place makes sense.
  • issue #9: Redundant long ints nrl,nrh,ncl,nch,ndl and ndh used for matrix/tensor sizing were removed, matrix allocation and deallocation now take integers as arguments.