NEWS

InterpolateR 1.4-4

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.

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.

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.

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.

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).

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.