v8.4.0

• Simulation output remains the same as the previous version, but output of vegetation establishment is now correct. Additionally, output of the example differs slightly from previous example output because of the following changes to the example inputs: updated potential evaporation coefficients; updated rooting profiles.
• Rooting profiles can now be estimated with an equation and parameters for each vegetation type; soil impermeability represents (semi-)restrictive soil layers that reduce rooting profiles (#459; @dschlaep).
• Potential evaporation coefficients can now be estimated from soil properties (#409; @dschlaep).
• The weather generator is now functional for all modes (#497, #498; @N1ckP3rsl3y). However, nc-based SOILWAT2 cannot input site-based coefficients and is restricted to domain-wide coefficients.
• Error and warning messages now consistently identify program stage, simulation unit, and date when the problem occurred; netCDF-related errors additionally identify the specific file and/or variable, and the final summary log consistently reports the number of simulation units with warnings and with errors, even if zero (#509; @dschlaep, @N1ckP3rsl3y).
• Increased test coverage of SOILWAT2 features and use cases by the "ncTestRuns" framework and the "allOutputChecks.sh" script (@dschlaep).
• For netCDF output, relative positions of vertical and time coordinate values can now be selected to left, center, or right align within bounds (#518; @dschlaep).
• Surface albedo can now be estimated with dynamic snow, soil and vegetation components (#522; @dschlaep).

Bugfixes

• Output of vegetation establishment is no longer reported one output time period too early; the bug occurred only for events on the first day of an output time period (#494; @N1ckP3rsl3y).
• The mpi-mode of "ncTestRuns" now works again also for the reference run (#500, @N1ckP3rsl3y).
• Output of nc-based runs no longer produce incorrect values if the last simulated year is incomplete (#503; @N1ckP3rsl3y).
• Simulations with nc-based inputs now only require forcing data for the simulation time period; in particular, forcing data are no longer needed past an early end date until the end of the last calendar year (#506; @N1ckP3rsl3y, @dschlaep).
• Weekly, monthly and yearly output are no longer incorrectly impacted by a shortened output time window. The functionality of shortened output time windows is removed (#185; @dschlaep).
• Several fixes for mpi-mode (#512, #513, #515, #516; @N1ckP3rsl3y, @dschlaep)
  • preventing the stalling of the program,
  • avoiding communication errors in specific workload configurations (#512),
  • no longer incorrectly marking sites/gridcells as failed when the program is terminated before the entire simulation set completed (#513),
  • correctly reading parameters of soil water release curve from nc-inputs for all values of N_SUID_ASSIGN (#515),
  • correctly handling daily weather inputs in leap years from nc-inputs with "noleap" calendars for all values of N_SUID_ASSIGN (#516),
  • correctly reading latitude of all grid cells -- including when mean monthly windspeed is read from nc-inputs (#516).
• Improve compliance of "netCDF" output with CF 1.10 (#520; @dschlaep).
• Output of nc-based runs no longer produces an out-of-bounds error if a user requests only a subset of the four output time steps (daily, weekly, monthly, yearly) (@dschlaep).
• The output directory specified via "outsetup.in" is now used correctly even if a user does not include a trailing "/" (@N1ckP3rsl3y).
• Fixed off-by-one errors in an internal size and a format-specifier mismatch used for reading nc-input variables and attributes (@dschlaep).
• Several fixes for spatial index files that are used to map simulation domains to nc-inputs (@N1ckP3rsl3y, @dschlaep)
  • construction and querying of KD-trees no longer fail if coordinates contain duplicates,
  • index file directories are now created automatically if they do not already exist,
  • the simulation domain input file is no longer incorrectly checked against itself when determining index file use.

Changes to inputs

• New input via "siteparam.in" to select the input option for potential evaporation coefficients: either provided as inputs by "soils.in" or estimated from soil properties.
• New input via "siteparam.in" to select the input option for rooting profiles: either provided as inputs by "soils.in" or estimated with an equation and parameters for a vegetation type.
• New input via "siteparam.in" to select the albedo method and values for ground surface roughness length, fresh snow albedo, soil albedo at zero moisture and at saturation, shape parameter for soil albedo darkening with moisture, and shape parameter for the snow fractional cover relationship.
• New nc-inputs via "inSite" for soil albedo at zero moisture, soil albedo at saturation, and a shape parameter for soil albedo darkening with moisture.
• New inputs via "veg.in" to provide values for the three parameters of the rooting profile equation for each vegetation type.
• New inputs via "veg.in" for values of the extinction coefficient for canopy albedo for each vegetation type.
• "START" and "END" day of year from "outsetup.in" are now ignored.
• New input via "desc_nc.in" to select position of vertical and time coordinate values relative to bounds.

Changes to outputs

• All modes (including txt-based) now output values of soil evaporation at each soil layer.
• All modes (including nc-based) now output only complete weeks, months, and years, i.e., time periods that are not affected by a delayed simulation start (after January 1 of the first year) or an early simulation end (before December 31 of the last year).
• New variable "pft_labels" for "netCDF" output now provides vegetation type names along the "pft" dimension; this replaces the variable "pft" which previously provided vegetation type names via attributes "flag_meaning" and "flag_values".
• New output group "ENERGYAVG" which currently includes surface albedo.

Full Changelog

v8.3.0...v8.4.0