SPEI 1.8.0, November 2022

This is a major release that adds a range of substantial new features and fixes a number of major and minor bugs.

Breaking changes

All functions now accept 3-d arrayed data, enabling the possibility of using them on gridded climate data. A new data set (cruts4) has been added to test this new function.
Function <spei> now accepts non-monthly data (that is, data with frequency other than 12).
A thorough set of test functions have been implemented to help developing new functions without compromising the old results (Thanks to Doug Freedman who wrote the seminal set of tests).

More thorough data and options checks have been implemented, and a single error / warning message is produced containing all the errors upon failure.
Different verbosity levels are now implemented in all functions: now they print information about the options being used. A new argument (verbosity) can be used to override this behavior if not wanted.
Different versions of the Penman-Monteith ETo calculation have been implemented in function <penman>.
A new option to include CO2 concentration data is now available in function <penman>.
A new option for when no wind data are available in function <penman>.
<plot.spei> has been completely rewritten based on ggplot2, solving some bugs and enabling more flexibility.

Several minor bugs in <thornthwaite>, <hargreaves>, and <penman> functions have been solved (output difference is less than 0.1%).
A bug in <spei> has been solved that resulted in bad cumulative data when using a non-rectangular kernel, leading to incorrect SPEI values.

Corrected an error in <spei> function, which was not working when distribution was Gamma or PeasonIII and using user provided parameters. (Thanks to Emanuele Cordano, emanuele.cordano@gmail.com – ‘ecor’)
Added probability of monthly precipitation = 0 (pze) when using user provided parameters. (Thanks to Emanuele Cordano).

Corrected a bug in the function which resulted in a multiplicative kernel instead of an additive one such the one expected in the and functions.
Plotting methods for objects completely rewritten, now using . Additionally, a bug causing a bad representation of time series starting in months other than January has been corrected.

Corrected an error in the function which resulted in wrong potential evapotranspiration estimates when a multivariate time series was used as input.
Corrected an error in the function which resulted in wrong handling of zero precipitation months when using the Gamma or PearsonIII distribution.
Minor fixes to the and functions to correctly handle objects when they result from a call to .
Modification to the function, which now yields kernel coefficients averaging one.
Corrected an error in the functions and which resulted in ub-pwm method being used irrespective of the value of the parameter used, when using the ‘Gamma’ or ‘PearsonIII’ distributions.
Minor changes to the documentation.

Optimization of function , now using embed() for accumulating the data at the desired time scale.

Minor fixes to functions and .
Documentation of the function defined by mistake ed as the saturation vapor pressure, while it should read ‘actual vapor pressure’.
Function added to display basic information about the SPEI package at startup.
Function added to display the NEWS file.

Minor fixes to functions and .
Added new option for user-supplied SPEI parameters in the function. This overrides the fitting of a probability function to the data.
Added new dataset from Allen et al. (1998).

Fixed a bug causing several functions to fail when a time series not belonging to matrix class was provided.
Function now distinguishes between calls to and and labels the axis accordingly.

Functions and now yield an object of class “spei”.
New functions for summarizing and plotting “spei” objects are provided.
An option to establish a reference period for the computation of the indices has been implemented in functions and .

First release of the SPEI package.