Commits · efc25e92e6f85d9aab0bfdb82733f65726711a4e

ImisIO: Added the USEANETZ functionality. Many IMIS stations don't have an own precipitation measurement, but there has been some work done to associate the measurements from nearby ANETZ stations with the IMIS stations, thus allowing to calculate a regression for the precipitation for the IMIS stations. If you want to use this feature, then add the following key to your io.ini

[INPUT]
USEANETZ = 1

Small fixes: the MAGNUSS interpolation has been renammed in HNW_SNOW (since we will add avalanche snow redistribution in the future and potentially other things) and the documentation has been reworked.

A severe bug has been found in the spatial interpolations: the algorithm arguments were always the ones from the last evaluated algorithm (therefore not necessarily matching the chosen algorithm).

The MAGNUSS spatial interpolation algorithm now takes as an optional argument the name of the algorithm to use to initialize the grid. 

It is now possible to specify a lapse rate to be used for the CST_LAPSE algorithm (optional argument). If the "soft" argument is added, then a lapse rate is calculated from the data, and if it fails the user provided lapse rate is used instead. Without "soft", only the user provided lapse rate is used. Without any option, only the lapse rate calculated from the data itself is used.

This is the first implementation of the precipitation distribution algorithm discribed in (Magnusson, 2010). This basically initializes the precipitation with an IDW_LAPSE, then removes some or all precipitation according to the local slope, applies a factor depending on the curvature (Huss, 2008) and applies a factor so that the mean precipitation matches the original mean (this is definitely sub-optimal for this part).

Small adjustments to accumulate resampling filter: return IOUtils::nodata in all cases where accumulation is not possible

Fixed a compiler warning in Matrix ("warning: type qualifiers ignored on function return type" , cf http://stackoverflow.com/questions/1607188/why-is-type-qualifier-on-return-type-is-meaningless for full explanations). Small hack/bug fix in SMETIO 8related to issue 87)

The matrix class now uses the standard matrix notation: A(i,j) represents the element j of line i, with index starting at 1 and going to nrows/ncols. The documentation is not yet available...

The matrix class is now independent of Array2D (this makes more sense since arrays and matrices are quite different)

More improvements into the matrix class: the +,-,*,/ operators for doubles have been defined. The +,-,*,==,!= operators also exist for matrix as well as a isIdentity() for checking if a matrix is an identity matrix.

The embryo of a matrix class has been implemented. It is still based on Array2D and therefore the indexing is still 0 to n-1 instead of being 1 to n and the coordinates are swapped compared to normal matrices.

New regression models have been added, all based on the linear regression code but using variable transforms to achieve their goal. 

The global header file (MeteoIO.h) was incomplete. It has therefore been reorganized and now contains everything that it should...

This is the initial commit of the horizon functionnality for DEMs as coded by Fabienne Lanini. Some cleanup will still be required though...

The comparison operator of Coords has been reviewed and improved. The << operator that is not available for a parclass (for popc) can be replaced with IOHandler.toString() for POPC

The DEM object now calculates slope and azimuth in degrees. This is more consistent with the usage of bearings in degrees and of trigonometric angles in radiants.

A missing file has been added to the autotools compilation. A better comparison is performed for Coords objects.

Temporary fix for the << operator and POPC: a par class can not have a "friend" method... (so IOHandler can not redefine "<<" the way we used to do). Therefore, it has been disabled for popc

In order to solve compilation problems with VC++, we need to rename all publicly accessible variables whose name start with a "_". This is the begining of this effort...

More debugging help: the "<<" operator has been redefined for MeteoFilter, MeteoProcessor and BufferedIOHandler. Since the Config object is copied all over the place, it is not always printed.

Several 1D statistical methods has been implemented while the existing ones have been made nodata safe.

Added a redefinition of the << operator for IOPlugin and IOHandler. This should be convenient for debugging!

Some small optimizations have been brought into the spatial interpolations following some tests with cachegrind. Nothing huge, but small improvements.

Added a void write(const string& filename) procedure to the Config class. The Config object can thus be written into an ini file.

Added two functions to MeteoData: bool param_exists(string) and void reset()
reset() sets all the values of the MeteoData object to IOUtils::nodata (note: not the date), and param_exists(string) returns true if a certain parameter exists and false otherwise in a MeteoData object.

The Meteo1DInterpolator now correctly creates MeteoData objects that have more parameters than the default ones (like ta, hnw, ...)

Code tagging and some usefull methods for debugging and the begining of a true accumulator for time series (for any sampling rates).

Added a soft min/max filter to Meteo2DInterpolator. It is explicitly called for RH grids and HNW grids to make sure values stay in the range of [0;1] (for RH) or [0; inf) (for HNW).

The linear regression implementation is now able to remove a variable number of "invalid" points based on the initial size of the data set. Some extra weighting methods have been implemented (but are not yet used).

Several speed optimizations have been brought to the 2D interpolations (using the const keyword when possible, etc). The Quake3 1/x^2 approximation has been implemented for the IDW_core and brings a slight loss of precision (1e-6 in relative error in real life tests) while boasting a huge reduction of computational time (factor of 3 when called within Alpine3D with its own overhead).

The documentation can now be build and not installed (instead of linking building and installation as previously done)

The array memory alignement has been fixed for Array2D (this was the only that was wrongly aligned). This is a fix to issue 43

The bufferAlways method of BufferedIOHandler has been renamed as setBufferProperties. The A3DIO, IMISIO, SMETIO and SNIO plugins now correctly support time zones. The plugin howto documentation now warns the developer to take care of time zones issues.

The DEMObject was not properly buffered (the update flag was not honored), this has been fixed. The A3DIO plugin now implements readStationData. A few speed improvements have been brought to A3DIO and SMETIO by using the "reserve" method of the vectors (so that the vectors don't have to reallocate memory constantly).

The DEMObject subset constructor was broken, it has been fixed. It involved creating a subset method in the arrays classes.

The Coord class now supports using the given UTM zone when converting from lat/long to UTM. If the computed zone does not match the one requested in the io.ini, it prints a warning but uses the user-supplied zone.

A first implementation of readStationData for SMETIO has been done. 

Some documentation has been added for plugin development

Added a deleteKey function to Config object: both deleteKey("testkey", "mysection") or just deleteKey("testkey2") are valid (defaulting to the GENERAL section).