The parallel version of the radiation fields was not working properly (wrong geolocalization for the domain band). The documentation has also been updated.

Finally, this is the implementation of issue 340: distributed cloudiness (through atmospheric loss factor and splitting coefficient). This mostly follows the lines of what has been recently implemented in MeteoIO with a few tricks to perform domain decomposition when running in parallel (either MPI or OpenMP).  as well as benefit from Snowpack's albedo feedback. In the first tests, when commenting out this albedo feedback and forcing horizontal components, it performs exactly the same as MeteoIO's implementation.

Since there is no meteo1d object anymore, Snowdrift had to be modified so it can extract what it needs from a vector of MeteoData. An hopefully better implementation of Snowdrift without slatation is also included.

Fixed some documentation issues, lowered the thresholds to turn on the katabatic flows correction, swapped the linking order of Snowpack and MeteoIO in order to easily allow to use a system-wide MeteoIO with a locally installed Snowpack

The latest changes in Snowpack have been reflected in SnowpackInterface. The new way of setting the compiler options has been imported and tweaked from meteoIO. In order to mave closer to the "efficient c++" recommendations, several changes have been made in the rest of the classes (mostly defining a few copy constructors and "=" operators).

More documentation fixes, improvements to the getSynop script, small bug fix in EnergyBalance (more useful error message to help debug simulations)

Finally, everything seems to work properly, the documentation has been updated and several bugs fixed. The structure will still have to be improved (too many callback to my taste).

Paving the way for the restructuring: all modules should now be able to declare which grids they need, provide a timer and be constructed and managed from AlpineMain / AlpineControl. The next goal is to make each module call SnowpackInterface to get its grids and remove the other way around as is currently the case. Some deprecated methods have been removed.

Trying to address some of the "HACKS", removing dead code, hoping to make the new gridded outputs fully functional and safe. Let's see how ctest will digest it tonight!

This is a big code cleanup: the "TYPE_DOUBLE2D" has been removed (it was only an alias) and the definitions from Alpine3D.h moved into AlpineMain.h so Alpine3D.h is now a proper include for "library" usage.

Trying to get non-confusing information for the points of interest outputs: the enforced/measured snow height is now set to undefined. Touch up on the information messages for more consistency.

The runoff module now has a call back to SnowpackInterface (to be used in an upcoming version).

This fixes a bug with the MPI version of SimpleTerrainRadiation (uninitialized values summed through all_reduce_sum as well as possibly negative sky view factors and the usage of the diffuse radiation without multiplying it by the sky view factor when using terrain radiation.

Some unecessary MPI barriers have been removed. A few unused typedef have been removed. The copyright notices have been uniformized and edited when necessary.

A bug has been found in the MPI version: when reading the meteo data, only the master was able to pre-read the data but it was never broadcasted... thus forcing all workers to re-read the data themselves. The MPI outputs have also been cleaned up in order to reduce the deluge of messages in a parallel simulation. Whenever possible, only the master prints the messages. Multiple copies have been kept when this would help understand what is going on in a parallel simulation.

The MPI serialize() method was not returning the size of the serialized object which made it impossible to properly allocate the memory after the call. The implementation of the new psum_ph parameter in MeteoObj had been totally butchered and was returning an empty grid at the first time step (thus leading to a crash when trying to distribute this grid to workers). 

More consistency in the error messages, more try/catch blocks in order to properly report problems in constructors, code cleanup and documentation improvements (things that were either missing or unclear). Turning off the warnings about pragams, since we need to keep openmp pragmas even when not compiling for openmp!

Warnings cleanup, improvements to the build system (suppress some warnings that will take quite a while to address, etc)

This is still a commit that BREAKS EVERYTHING, so don't update to this one either! The code now compiles but fails at linking.

The goal is to make Alpine3D's structure very similar to Snowpack or MeteoIO. Therefore this commits deletes obsolete files and mostly moves files around.

More documentation, updated the run.sh file for greater clarity and flexibility. An optional key in Ebalance has been made mandatory because it was always forgotten...

The view factor computation was NOT initialized, leading to wrong results. This has been fixed. Another implementation has been written for sky view factors that is much faster. This has been checked against the view factors as computed by SAGA (and/or QGis). The implementation should be identical but the results are not identical. However the SAGA sky view factors seem too high when analyzing the terrain. The original view factors are in any case way too high compared to any other implementation and show very limited spatial variations compared to other implementations.

It is now possible to generate gridded runoff without having to define catchments. The format and location of the runoff grids is now configured with specific keys. The documentation has been expanded to cover each modules.

Some missing includes have been added for Windows. The necessary changes in order to be compatible with the latest MeteoIO (that got rid of the grid.ncols syntax) have also been performed.

The configuration keys that have a default value are now properly using it. The terrain_radiation algorithms now provide a variable that contain their name and this is used in information messages. 

Birthday bugfix: direct and diffuse radiation where given in the wrong order as arguments to getRadiation, cleaned up MPI parallelization of TerrainRadiationSimple algorithm. Bugfix also in SnowpackInterface - the albedo_grid needed to be reduced before handing it over to EnergyBalance.

TerrainRadiation: Use TerrainRadiationSimple for the time being. The MPI version of TerrainRadiation.cc does not deliver the same results as the sequential version. This needs to be investigated further and cleaned up. 

Moved the allreduce_sum of the reflected radiation into TerrainRadiation.cc:getRadiation(). The TerrainRadiationAlgorithms are responsible for their own parallelization, they operate on the full DEM and receive the full diffuse and direct radiation components as parameters to their getRadiation() functions.

Fixed the simple terrain radiation ("+" instead of "*"). Fixed a size_t kind of warning for 64 bytes. Otherwise, white spaces...

SnowpackInterface: now runs with MPI and scales very well - however two issues remain: interdependecy of modules (runoff is called from within snowpack) and the writing of data, that is distributed on multiple MPI nodes (writeOutputSNO, writing runoff data)

EnergyBalance: Bugfix, the indices for the OpemMP threads were not mapped onto the correct DEM portion.

EnergyBalance: Reintroducing OpenMP - now both OpenMP and MPI can coexist in one binary peacefully. Every MPI process gets a certain slice of the DEM and divides that DEM locally according to the number of workers configured.