Changes in version 1.4-4                        

New Features

  - The entire Cressman algorithm code has been refactored, which has
    accelerated its execution speed.
  - The “n_round” functionality has been added to the Cressman
    algorithm, which allows the final layers to be rounded.
  - The IDW algorithm has been redesigned, ensuring faster execution
    time and resource consumption.
  - A new parameter called “mask” has been added to the IDW algorithm,
    allowing the final file to be cropped and masked based on the input
    polygon.
  - The Altitude “Z” parameter is no longer required when running IDW or
    Cressman.
  - The ordinary kriging method has been added.

Bug Fixed

  - An error that occurred if the “Z” column was not passed in BD_Coords
    when it was not mandatory has been corrected.
  - Typos in the IDW documentation have been corrected.

                 Changes in version 1.3-4 (2025-05-02)                  

New Features

  - We added new goodness-of-fit metrics, including: Mean Squared Error
    (MSE), pearson's Correlation, Coefficient of Determination
    (R-squared)
  - We performed a refactoring of the project for its future maintenance
    and addition of new functionalities.
  - We have removed hydroGOF dependencies for the calculation of
    goodness-of-fit metrics. The metrics will now be hosted as internal
    functions.
  - We have added a new function to create a file similar to BD_Obs from
    several csv files.
  - We have restructured the RFmerge logic, reducing the execution time
    by 50%.
  - We have restructured the RFplus logic, reducing the execution time
    by 50%.
  - We have removed the RFplus validation that prevented you from
    limiting the Date to be only of “Date”, “IDate” or “Posixct” format.
    Now, you can enter any format for the date.
  - We add the function “create_data” to create the BD_Obs file from
    multiple individual files.

Bug Fixed

  - Fixed several bugs within the RFmerge tests
  - We have fixed a bug in RFplus and RFmerge that caused the algorithm
    not to compile when the layer “DEM” was entered with a name other
    than “DEM”. After fixing this bug, the user will be able to set any
    name for this layer.
  - We have corrected bugs in the RFmerge and RFplus algorithms that
    prevented interpolations to a scale other than daily. Now, these
    algorithms allow interpolation to any time scale (from minutes) to
    years or other time scales.
  - We have fixed a bug in the IDW and Cressman algorithm, which caused
    BD_Obs data to be transformed to daily format, regardless of the
    scale entered. This bug did not allow interpolations to hourly or
    smaller scales.

                 Changes in version 1.2-0 (2025-04-02)                  

New Features

  - in the validation mode, the 'mod_validation' parameter has been
    removed, as it was unnecessary. Now, if the 'stat_validation'
    parameter is entered, it will validate with the stations entered. If
    'stat_validation' is NULL a random validation will be used.
  - We have added the option to manually insert stations for validation,
    option 'stat_validation', when this parameter is entered it will be
    used to validate the stations entered by the user, otherwise when
    stat_validation = NULL, the validation will be performed randomly.
  - We have optimized the RFmerge algorithm and its validation algorithm
    has been connected to 'validation_module'.
  - We have optimized the RFplus algorithm and its validation algorithm
    has been connected to 'validation_module'.
  - We have created the 'validation_module' which will contain all the
    logic for point to pixel validation. All integration methods that
    require validation will be connected to this module.
  - Add a new interpolation method. RFplus proposed by Landi et al.
    (2025).
  - Add a new interpolation method. RFmerge proposed by Baez-Villanueva
    et al. (2020).
  - Add interpolation method Inverse Distance Weighted Interpolation
    (IDW).

Bug Fixed

  - An error has been corrected when manually selecting stations for
    validation, since if the training was not different from 1 the
    validation was not performed.
  - We have corrected a bug in the validation methods, when the 'random'
    method was set, the IDs did not match each station in situ. We have
    made a new approach in the allocation, correcting this problem.