Bugfix in a constructor (that was making an error on some systems) and updated the Stefan-Boltzmann constant according to http://physics.nist.gov/cuu/index.html

Mostly code cleanup: trying to find more logical names for the buffering code in BufferedIOHandler, more modularity in InterpolationAlgorithms. Since when computing a variogram from time series it is possible to end up with significantly lower values than usual that lead the krigging in a totally wrong direction (should the variogram be normalized? Is it linked with the stationarity hypothesis?), now the empirical variogram is used instead of the full time series.

Two new methods for data binning have been added to libinterpol1D. The krigging now works properly with either an empirical variogram or a variogram computed on the last 3 days of data (the later being currently enabled). But a proper caching of meteo grids will have to be implemented!

This is a tentative improvement for the krigging, by properly building the variogram from past time series of variances. Unfortunatelly, this makes no difference with the previously implemented empirical variogram. Therefore the simpler version is still active. Stay tuned for upcoming improvements!

The Liston & Elder wind interpolation algorithm now works like in the paper. It has been renamed as "LISTON_WIND" and shows acceptable/encouraging preformances.

Working on wind spatial interpolation: both Ryan and Liston have been reviewed and improved. They can now both be used for VW and/or DW. 

The Winstral spatial interpolation algorithm is getting into shape: all three modes of operation have been implemented and tested and it is now properly documented. A new data generator has been created that computes a relative humidity from either a dew point temperature or a specific humidity.  Adding a "check_attribute" to libncpp.

My previous optimizations had introduced a bug in SimpleDEMWindInterpolate, this has been fixed. The optimizations are actually now done at another location. The (Ryan, 1977) wind direction interpolation has been properly checked, fixed and documented.

Yet another step for the MINGW date handling bug workaround... And an optimization: precipitation grids will be directly filled with zeroes if no station sees any precipitation (to avoid complicated algorithms in Winstral, etc)

More refinements to the Winstral spatial interpolation: it is now possible to provide a reference station to take the wind direction from, or to provide a fixed wind direction. 

Fixed a stupid bug in my previous commit and added the possibility to provide a fallback constant when using the CST spatial interpolation:
RH::algorithms = Cst
RH::cst                = 0.3

This would behave as before, except that if no data at all would be available at a given time step, then the value 0.3 would be taken to fill the grid.

The example io.ini file was missing some new changes. The Winstral algorithm had a bug when operating on dems with holes (nodata). It now takes its wind direction as argument (the base algorithm also has to be provided).

The first algorithm implementing (Winstral, 2002) is now being tested. This seems to work but the mass balance still has to be validated. The IOManager can now be constructed directly with a io.ini file name. Some code clean up has been performed in InterpolationAlgorithms.

Finally a good documentation on ordinary kriging... This should be a reasonnably standard (since everybody seems to enjoy using his own notation for the various matrices) and more importantly, really show what the whole thing is about, how the matrices are filled and what is solved.

Preparing the way for altitude thresholds on lapse rates: for example, preventing HNW to get crazy at elevations much higher than the highest station. This would be achieved by using max_altitude instead of the true elevation of a high point when doing detrending/retrending. See issue 358 and issue 359.

Fixing a documentation bug (thanks to Matteo), small optimization in Sun and a few extra optimizations in MathOptim

A new spatial interpolation algorithm has been implemented: "NONE". This just fills the grid with nodata. Otherwise, an unnecessary check has been removed in Meteo2DInterpolator and some comments editing, white spaces...

The LIDW_LAPSE method is now back. It still works the same, ie it can produce very unatural results depending on the stations set that is used!

The version number has been changed to 2.4.1 (this will most probably be an internal release shipped with Snowpack).

Bringing back the LIDW spatial interpolation method that perfoms IDW on a subset of the stations. At least it compiles...

Slightly improved documentation on spatial interpolations and now really allowing linear regression on only 2 points

Redesigned algorithm for redistribution of solid precip on steep slopes. This is done by pushing precip from the top to the bottom of the slope.
Curvature correction has been turned off. Now this redesign will be validated with laser-scanner data. 

The info string for the spatial interpolations was not being cleared between calls, leading to ever increasing content... Otherwise, using an ostringstream instead of stringstream in PNGIO and fixed a typo in the documentation.

More size_t fixes (ie trying to be consistent) and a bug was found when spatially interpolating using a few specific algorithms that would not reset their internal vectors between two calls (thanks Matteo for helping uncover this bug!)

Now the toString() method of Meteo2DInterpolator works with the new structure and is called by IOManager.toString(). Small cleanup (white spaces, etc) in 2D_interpolations and ResamplingAlgorithms.

One more step closer to the goal: the dem is now only passed when calling calculate() and the simpleDEMWindInterpolate now computes the curvatures if necessary.

Small optimizations in IOHandler.

Getting closer to the goal of persistent spatial interpolation objects. There are no "initialize" and "getQualityRating" anymore, the initializations are performed by "getQualityRating". The arguments that used to be passed to the constructors are now split between constructor and getQualityRating. A few hacks will have to be removed: we should be able to get rid of some of the callbacks, the plain copy of Date and Dem will have to be avoided (most probably by working with pointers) and the map of objects that has been implemented (but commented out) will be enabled. So, a few hacks remain, but this is already a nice, big step forward!

The Krigging is now working. The current limitations are: 1)there is no maximum range, so each available station would take part to the variogram fit. 2)the covariance is currently NOT computed on past time series, limiting the relevance of the variogram. Practically, the variogram fit that always ends up being used is LINVARIO.

All matrix and fit methods that used to return void but could also throw exceptions now return a boolean to indicate if things went well (ie a matrix could be inverted, etc). They still throw exceptions for logic errors (incompatible dimensions, etc).

Following the latest changes, this is a general code cleanup in all the classes involved in spatial interpolations. The LIDW method has been commented out since this works very poorly and would need to be properly redone.

And now the code that uses the previously upgraded infrastructure in order to compute spatial interpolations using a trend/residuals approach. The tests work, all results should remain identical to what it was before. Lots of now unnecessary code will be removed from the libinterpol2D in the next few days...

Some code cleanup and micro-optimizations. The documentation has been improved (configuration of doxygen as well as some duplicate section labels).

Putting the code into shape for major changes in the spatial interpolations as well as pushing the minor version nunmber up

After compiling with -Weffc++, it appeared that lots of things could be done in initialization lists. Some of it has been done for this commit (but quite a few remain).

New warnings have been tried to detect variable shadowing (one variable declared with the same name as another one), alignement issues, etc. This produces very verbose output (not all warnings are legitimate) but allowed to fix lots of cases of misalignement and shadowing issues. These options have been left in the CMakeLists but NOT enabled (since these warnings are not so easy to interpret).

Two new members have been added to Grid3DObject, to store the heights of the 3D levels (ie: level 1 = 10m above ground, etc). This is similar as what is done by ARPS, GRIB, etc A boolean specifies if these are absolute or relative heights (ie: above DEM).

The unventilated temperatures filter and unheated rain gauge filter have been renamed. The "soft" option has been added to Unheated_RainGauge. And the usual documentation fixes...

IOManager: added another interpolate procedure that takes as arguments a Date, a DEMObject, the parameter to interpolate, a vector of Coords objects and a vector of doubles. The Coords represent different points in the DEM. The interpolation is only performed for these points and for each point one double value is stored in a vector. 

Furthermore IOManager now has a member variable called interpolater of type Meteo2DInterpolator. A circular inclusion problem has been solved between IOManager and Meteo2DInterpolator.

libinterpol2D bugfix: The SimpleDEMWindInterpolate algorithm had  a small bug in it, it wasn't interpolating values in the last row or last column of the 2D grid.

Some variable names starting with "_" have been fixed. The Crawford and Duchon ILWR model has been implemented in Atmosphere and checked on Weissflujoch data.

The Fit1D class constructor can now read a fit model name as a string. This was necessary for parsing a user input coming from io.ini. 

The ordinary kriging interpolation algorithm is now active. It is far from perfect (the variogram is computed in a very cheap way that contributes to make it very fragile), but it is time for the first users to start playing with it. The variogram model to use can be provided as an optional argument and in case of fit failure, an appropriate error message is given (but no fallback mechanism exists). The next step will be to compute the variogram on time series, which should produce more reliable data sets for fitting the variogram model on.

Some error messages have been improved in the matrix class.