NEWS
climatol 2026-05-20
- outrename: The output files are copied with the new names rather than just
being renamed as in former versions.
- rclimdex2climatol: Fixed bug that exchanged lat-lon coordinates since
version 4.3-0.
climatol 2026-01-23
- homogen: The detection of excessive sequences of identical values is now
based on the Poisson probability distribution.
- Default niqd is now c(5,6,3), where the third value refers to the quantile
1-10^-3 (i.e., 0.999) of the Poisson probability distribution rather than to
any interquartilic distance as in the former algorithm.
- The deletion of spatial outliers now takes place only in the first iteration
to avoid too many false positives.
- Fixed bug that generated corrupted pdf output with onlyQC=TRUE.
climatol 2026-01-02
- homogen: removed debugging code that saved some data into aux0.Rdat file.
- niqd is increased from c(4,6,1) to c(5,6,1.5) to reduce false positives in
the initial quality control.
- Colors are now specified by name to avoid differences between systems.
climatol 2025-12-16
- datrestore: now restores true outliers in all homogenized series, not only
in those reconstructed from the last homogeneous subperiod.
- homogen: Fixed bug that sometimes gave errors when metad=TRUE.
- rclimdex2climatol: Bugs fixed.
- weekendaccum: improved quantiles for the detection of weekends or other days
with false zero precipitation which actually has been accumulated.
- Parameter expl=TRUE has been added, to avoid changing any value by default.
- Output files now have consistent names *_wkn.*
climatol 2025-10-12
- homogen: Removed code added to version 4.0 intended to lower the process time
but was worsening the reconstruction of short series.
- Removed inconsistency which did not allow to ouput graphics when using cumc
and onlyQC together.
- Fixed error when stage 1 was skipped.
- Old parameter tol reintroduced to allow different tolerance to simultaneous
splits in the series.
- rclimdex2climatol: readapted to the last RClimDEX/RClimPACT files format.
- rean2climatol: new function to append a series from reanalysis to the input
files.
climatol 2025-05-28
- csv2climatol: Now enabled to process subdaily data.
- dahstat: Added option stat='mseries' to output homgenized monthly aggregates
to a CSV file.
- No longer fails with an error when trusted reference series are not at the
end of the datasets.
- dd2m: Series with less than 12 monthly data are removed from the output.
- homogen: A new initial check removes series with less than mindat data items.
- The initial QC implements detection and deletion of spikes in the series,
except if their probability distribution is very skewed. (niqd now defaults
to c(4,6,1).)
- The correlogram disregards correlations calculated with less than 10 data
pairs.
- Reciprocal outliers (same date and opposite sign in nearby stations) are
reinstated by means of the datrestore function.
- Deletes fragments with zero mean if std=2 (normal ratio normalization).
- Zeroes are not excluded by default from the detection and deletion of long
runs of identical values when series are not daily or subdaily.
- raway is now a factor to apply to distances involving reanalysis series
previously increased by the median distance between observational series.
- onlyQC no longer saves the original data as *-QC_*, but as *-ori_*, as
users sometimes thought they should continue working with the *-QC_* files.
- read.dat: Fixed bug in this internal function that failed to identify daily
data.
- weekendaccum: New function to check for weekends or other days with false
zero precipitation which actually has been accumulated. (Applicable to daily
precipitation series only.)
climatol 2024-03-23
- climatol2rclimdex: Issues message error in the absence of common station
codes in the three variables involved.
- csv2climatol: Lines without observations are now disregarded.
- dahstat: Fixed bug when stat='series' and only one series was requested.
- daily2climatol: Columns of the data files may be only two if the first one
contains dates in Date or POSIX format. Fixed bug in removing lines with no
data.
- datrestore: New function to restore deleted outliers checked as true into
the dah matrix of homogenized series contained in the *.rda output results.
- diagwl: Is now resistant to missing data in the input series, and the
parameter shem is now guessed. Corrected inconsistencies between parameters
yeari, yearf and per.
- fix.sunshine: Changes negative values to zeros.
- homogen:
- Break-points in a single series with a single trusted reference are now
adjusted without the need to set force=TRUE.
- niqd defaults to c(4,1) instead of 3 to delete clearly wrong data and long
runs of identical data (with niqd>1).
- Fixed a problem that prevented to draw high resolution coastlines in the
map of stations.
- Fixed the bug of extra %d in sprintf("dahstat(...)#averages",...)
- Outliers not deleted because of having only one reference are not plotted
in red anymore in the final histogram of standardized anomalies.
- New consistency check when data seem sub-daily.
- boxplots and correlogram weight less and display faster in the output PDF.
- IDFcurves: Hourly data are now allowed and issues with the legend have been
fixed.
- rclimdex2climatol: Fixed bug when reading the stations file.
- windrose: New parameters allow to choose the number of wind directions and to
define the wind speed classes.
climatol 2023-04-09
- Only the three first parameters are needed in the homogen() function to
process any climatic variable. The rest of parameters are guessed or have
safe default values.
- This function begins now with a quality control checking for extreme values,
increments into successive data and number of consecutive data of identical
value.
- Precipitation data accumulated during several days can be dissaggregated.
- Optional data transformation has been removed because the ratio normalization
yields better results.
- A new test (Cucconi) has been added which has the ability to detect both
changes in mean and variance. SNHT is still de default test, but the
thresholds are now named 'inht'.
- New utility functions to help in preparing input files with the required
format: sef2climatol, xls2csv and csv2climatol
- The unsatisfactory experimental function homogsplit has been removed.
Series in big domains with different climate regimes can be split providing
code lists to datsubset().
- New tools: fix.sunshine to prune any excess in sunshine hours produced in
adjusted daily series; and QCthresholds to obtain monthly quantiles of
extreme values, increments between consecutive values and sequences of
identical values.
- Added graphic products to display two dimensional scaterplots,
Intensity-Duration-Frequency diagrams, diagrams of running trends on time
windows of different lengths, Months-Hours isopleths and meteograms.
climatol 2018-05-04
- Checking that station codes are not duplicated and do not include the '-'
symbol.
- Important bug correction.
climatol 2018-04-15
- New file conversion functions daily2climatol(), rclimdex2climatol and
climatol2rclimdex().
- New function datsubset() to subset data into new input files.
- Bug correction in homogsplit().
- Bigger labels in graphics.
climatol 2016-08-27
- Many novelties, with new functions (to prepare the input data from ODBC
accessible data bases, to obtain grids from the homogenized series, etc) and
new functionalities added to others.
- The package is now able to deal with daily data, and can honor stations
metadata. (Be aware that some parameters change their names to be more
mnemonic).
climatol 2013-11-01
- Bug fixed: a global color re-assignment resulted in abnormal halt under most
O.S.+latest R combination.
climatol 2013-10-01
- Code optimization allows the process to be completed in around half the time
of the previous version.
- Original series can be labeled as homogeneous to keep them unchanged.
- Some parameters have more meaningful names (the old are kept for backward
compatibility), and a couple default values have been changed.
- dahstat() has a new option to yield the homogenized series in individual text
files.
- A map is drawn in the station location figure if the package 'maps' is
installed.
climatol 2011-06-05
- Corrected bugs in distance and trend computations and an improved convergence
in the missing data recomputation yield much better homogenization results.
climatol 2011-01-25
- Too many to be explained here. The main novelty is the homogen() function,
for an automated homogenization of climatological series.
- Other functions have been changed or replaced, so you cannot expect
compatibility with the 1.x versions.
climatol 2006-10-06
- Change of '°C' by just 'C' to avoid non-ascii characters warnings in the R
checking process of the package.
- Update of the PDF poster of the documentation.
climatol 2006-05-26
- Workaround to avoid 'Error: NAs are not allowed in subscripted assignments',
produced when using option sqrtrans=TRUE in function depudm() since R version
2.0.0.
- Minor corrections in the documentation.
01-Nov-2004, version 1.0:
- First public release.