v1.6.1 (August 27, 2023)#
A new minor release of the OGGM with several improvements and bug fixes. We recommend all users currently using 1.6.0 to switch to this version if they are still in the testing/learning phase. If you rely on your results staying stricly the same, you should stick to the version you are currently running.
OGGM 1.6.1 should be fully compatible with 1.6.0 code. We have updated the pre-procecessed directories however, and recommend users to switch to the new OGGM version if possible.
The new pre-processed directories are available in the 2023.3 version: https://cluster.klima.uni-bremen.de/~oggm/gdirs/oggm_v1.6/L3-L5_files/2023.3
These incorporate all the changes listed below.
Corrected a small bug in the W5E5 climate files, which led to some glaciers getting climate data from a grid point further away than they should. This should affect the results of a few thousand glaciers in a minimal way. Fix (PR1547, PR1557) by Lilian Schuster
Added more flexibility to
compile_run_output. It is now possible to compile runs with different data variables (the default is NaN). It is needed to compile different spinup strategies together, as some include special data variables (e.g.
is_fixed_geometry_spinup) (PR1563). By Patrick Schmitt
Fixed a problem in the hydro outputs where on some occasions
melt_on_glacierwould be negative. We changed this term to become a positive term for snowfall on glacier instead (PR1584). By Lilian Schuster
Fixed an issue with volume below water level computations with lake-terminating glaciers (PR1584). This affected only diagnostic computations of the
volume_below_watervariable and should be insignificant. By Fabien Maussion
There is now the possibility to compute distributed area and thickness changes from the flowline projections (PR1576, PR1585, PR1619, PR1623). The functionality is currently in the sandbox but is documented in the tutorials. By Anouk Vlug, Patrick Schmitt and Fabien Maussion
Added three new flowline diagnostic variables: thickness change in one year (
dhdt), forcing climatic mass balance (
climatic_mb) and flux divergence (
flux_divergence). All variables are in units meter of ice per year (PR1595). By Patrick Schmitt
Added more flexibility to
run_dynamic_spinup. Users can now specify a target year and a desired value to match. The default is still the same, matching area or volume at the RGI date (PR1600). By Patrick Schmitt
Added possibility to use MultiPolygon outlines together with elevation bands. That can be useful when working with local glacier inventories with multiple outlines (e.g. older outline single polygon but newer outline multi polygon for the same glacier) (PR1604). By Patrick Schmitt
The pre-processed directories are now run with dynamical spinup and calibration as the standard option. Dynamical calibration is run with a lower error tolerance than before, improving results in all regions (PR1558). By Fabien Maussion
OGGM now provides “standard” projections, attached to a specific OGGM version (PR1627). This will be a huge asset for many users and will help to track important changes as OGGM continues to improve. By Lilian Schuster
v1.6.0 (March 10, 2023)#
A new major release of the OGGM with several important changes. We recommend all users to switch to this version only if they are ready for a new study, or are prepared to rerun their simulations with changed results.
we removed the
init_glacier_regionstask, which was deprecated since a few OGGM versions. Similarly, other old functions (e.g.
process_cmip5_data) were also removed.
several default parameters were updated to new values. See “migrating guide” (in construction) to navigate through these changes.
the calibration of the mass balance models with the \(t^*\) (“T star”) method is no longer supported. The new calibration scheme is considerably more flexible, but relies on new parameter names.
as a result, old workflows and old glacier directories cannot be used (after Level 3) in OGGM v1.6 anymore.
Because of the many changes, the list below is not exhaustive at all. We preferred to focus on what is new in the tutorials, and recommend all returning users to go over the new tutorials to familiarize themselves with the changes.
Added a new entity task
run_dynamic_melt_f_calibration. This task dynamically calibrates the temperature sensitivity mu star to a geodetic mass-balance observation. There are different options available how this is done, the default incorporates an inversion and a dynamic spinup in each iteration (PR1425). By Patrick Schmitt
Added two new output variables in
volume_m3_min_h) which are needed for a dynamic mu star calibration which should include the minimum ice thickness argument of an dynamic spinup (needed as a filter for interannual changes of especially the area). Also included
cfg.PARAMS['dynamic_spinup_min_ice_thick']to be able to globally define the used minimum ice thickness for the dynamic spinup (PR1425). By Patrick Schmitt
Rearanged the prepro levels. Level 4 now adds a historical run (previously done in level 5) and a spinup historical run (using dynamic mu star calibration). Level 5 now replaces level 4 and creates the minigdirs (where only the files for a model run are kept and no inversion is possible anymore) (PR1425). By Patrick Schmitt
Added trapezoidal downstream line (PR1491). Can be selected with
cfg.PARAMS['downstream_line_shape'], with the options
init_present_time_glacier(gdir). By Patrick Schmitt
Added option to extend the plot limits when plotting multiple gdirs. Could be used with
graphics.plot_modeloutput_map(gdirs, extend_plot_limits=True)(PR1508). By Patrick Schmitt
New release of the OGGM model, setting the ground for a major update to be released soon. Several major improvements available for testing, and will become the default in a future major release:
MassRedistributionCurveModelwhich uses the Huss curves to parameterize glacier retreat.
mu_star_calibration_from_geodetic_mbtask which now calibrates each glacier individually
run_dynamic_spinuptask which used the ice dynamics model to spinup the model (instead of the default equilibrium assumption)
and much more! See below.
In the process of adding new output diagnostic files (PR1308), the signature and return output of
FlowlineModel.run_until_and_storechanged. We hope that this change won’t affect too many of our users but if it does, it should be relatively straightforward to update your code: users now control the number of outputs with the
geom_pathkwargs. Most users will probably have used the
run_*tasks anyway, and won’t be affected by this change (except maybe for the point below). By Fabien Maussion
cfg.PARAMS['store_model_geometry']is now set to
Falseper default. If you were relying on these files (e.g. for a run with spinup or similar), your code will fail with
Trueshould solve the issue.
Copernicus DEM 90m is now called
'COPDEM'and also uses the 2021 release with additional data corrections (PR1364).
Enhancements & bug fixes#
MassRedistributionCurveModelunder the FlowlineModel interface, which uses the Huss curves to parameterize glacier retreat. There is some parameterisation for advance as well, but a very coarse one (PR1288). By Fabien Maussion
Added a new
mu_star_calibration_from_geodetic_mbtask which now calibrates each glacier individually from the reference geodetic MB data. This is a rather quick solution for now, but it opens new avenues (PR1286). A bug in the new feature was later corrected (PR1351). By Fabien Maussion and Lilian Schuster
Added a new
utils.get_geodetic_mb_dataframewhich returns the reference geodetic MB data, currently from Hugonnet et al 2021. Also changed the behavior of
cfg.DATAto be shared across processes (PR1285). By Fabien Maussion
Added a new output diagnostic files containing variables along the flowlines instead of aggregated at the glacier level (PR1308). These files are stored in the glacier directory (
gdir.get_filepath('fl_diagnostics')) and are not saved per default. Set
cfg.PARAMS['store_fl_diagnostics'] = Trueto activate it. By Fabien Maussion
Added a new keyword argument to
fixed_geometry_spinup_yrwhich allows to “start” a simulation at an earlier date than the RGI date (PR1327). In practice, it computes the glacier volume change from SMB only (fixed geometry), ignoring glacier area change in this period. Therefore, it is only valid for short periods of times (years, not decades). By Fabien Maussion
Added new keyword arguments to
run_with_hydrowhich allow to select which glacier area should be used as reference for the hydro computations (PR1331). Furthermore, it also allows to use a previous geometry file for the computations, i.e. allowing for continuous historical to projections outputs (if needed). By Fabien Maussion
The package “descartes” is no longer a required dependency for oggm.
Added a new entity task
run_dynamic_spinup. This task dynamically spinup the glacier to match the area or volume at the RGI date. To do so the glacier is simulated from the recent past (default 1980) to the RGI date. The unknown glacier geometry at the start of the simulation is iteratively changed with a short constant climate run with a varying temperature bias (PR1342, PR1361). By Patrick Schmitt
Added the 30m version of the Copernicus DEM. This DEM can be set with
source='COPDEM30'and can be useful for smaller sized glaciers. An account with Copernicus is required to access the DEM (free for academics). (PR1364). By Matthias Dusch
Added a new merge_consecutive_run_outputs entity task which allows to merge two output files together (PR1379). This is useful to merge e.g. spinup + historical or historical + projection runs in post-processing. By Fabien Maussion
Very minor release to remove the dependency on the
This is a minor release of OGGM, containing mostly bugfixes and a few new features.
This version should be fully backwards compatible.
This a new update of the OGGM model. It should be largely compatible with OGGM v1.4.0. The main addition in this release is the computation of hydrological diagnostics. Check-out the new tutorial at https://oggm.org/tutorials !
Various name changes (with deprecation cycle, i.e. old code should still work):
Mass balance models now properly refer to
prcp_fac(was incorrectly named
prcp_bias) (PR1211). Additionally, the
oggm.core.flowlinecan now also adjust the precipitation factor for sensitivity experiments. By Lilian Schuster
Users can now choose the variables that are written in diagnostics files and can also choose not to write the glacier geometry files during the run (PR1219). The respective global parameters are
store_model_geometry. By Fabien Maussion
run_with_hydrotask which computes hydrological diagnostics after a standard dynamical run (PR1224). This is highly experimental, will likely change in the future. By Fabien Maussion
This a new major update of the OGGM model. It it the result of one year of development, with several non-backwards compatible changes.
We recommend all users to update to this version.
Major new release 1.4!
There have been a large number of additions too long to be summarized, and the list below is far from complete (we have waited way too long for this release). Here are the highlights:
new option to compute centerlines: “elevation band flowlines”
new option to calibrate OGGM mass balance regionally to geodetic estimates
new option to calibrate the creep parameter Glen A to match the ice thickness to the Farinotti et al. (2019) consensus
users can now choose from a variety of pre-processed directories, including with new climate data (e.g. ERA5)
OGGM now has a calving parameterization (switched off per default)
OGGM shop, to download several new input datasets
Historical runs (“spin-up”) are now available per default and can be readily used for projections
and much much more….
Note: not all changes since v1.3.1 are documented below. We’ll try to be better with documenting changes in the future.
The calving parameterization has been extended and made more useful by logging its output to compiled files. See PR996 for code changes and watch out for upcoming changes in the documentation and notebooks. By Fabien Maussion
Added a new “glacier directory initialization” global task: init_glacier_directories (PR983, GH965). It replaces init_glacier_regions and covers all its functionality, except that it does not process the DEM data (this was a confusing “feature” of init_glacier_regions). The old task init_glacier_regions is officially deprecated but without warnings for now. Since it is a very widely used task, we prefer to deprecate it in a slow cycle: first, change the documentation, deprecate later. By Fabien Maussion
More climate datasets are now supported in OGGM (PR1036). A new task (historical_climate_qc) has been added to quality check the climate timeseries. This has not been tested intensively yet and is still a WIP. By Fabien Maussion
The inversion tasks now can invert for trapezoid shapes (PR1045). This has non-trivial consequences for the model workflow. First and foremost, the decision about using a trapezoid bed (instead of parabolic when the parabola is too “flat”) is taken at the inversion step and not afterwards. The forward model and the inversion are therefore much more consistent. Furthermore, the task filter_inversion_output was simplified to take the estimated downstream bedshapes into account and now preserves glacier area, not volume. This also is a step towards more physical consistency between inverse and forward model. By Fabien Maussion
Added Copernicus DEM GLO-90 as optional DEM. Requires credentials to
spacedata.copernicus.eustored in a local
.netrcfile. Credentials can be added on the command line via
$ oggm_netrc_credentials(PR961). By Matthias Dusch.
Added support for “squeezed” flowlines from Huss & Farinotti 2012 (PR1040). The corresponding tasks are
tasks.fixed_dx_elevation_band_flowline(). By Fabien Maussion
Minor release with small improvements but an important and necessary change in multiprocessing.
After a recent change in multiprocessing, creating a pool of workers became very slow. This change was necessary because of race conditions in GDAL, but these conditions are rarely relevant to users. We now make this change in multiprocessing optional (PR937)
various improvements and changes in the dynamical model - mass balance model API. These were necessary to allow compatibility with the PyGEM model (PR938, PR946, PR949, PR953, PR951). By Fabien Maussion and David Rounce.
The time stepping scheme of OGGM has been fixed for several flaws. This blog post explains it in detail. We expect some changes in OGGM results after this release, but they should not be noticeable in a vast majority of the cases.
We recommend all users to update to this version.
The adaptive time stepping scheme of OGGM has been fixed for several flaws which lead to unstable results in certain conditions. See the blog post for a full description. The API didn’t change in the process, but the OGGM results are likely to change slightly in some conditions. (GH731, GH860, PR931). By Fabien Maussion and Alex Jarosch.
OGGM is released under a new license. We now use the BSD-3-Clause license.
Minor release of the OGGM model with several small improvements. We don’t expect major changes in the model results due to this release.
Important: this will be the last release under a GPL license. The next release (v1.2) will be done without modifications but under a BSD-3-Clause License.
Added ASTER v3 as optional DEM. Requires credentials to
urs.earthdata.nasa.govstored in a local
.netrcfile. Credentials can be added on the command line via
$ oggm_netrc_credentials(PR884). By Matthias Dusch.
Minor release of the OGGM model, with several substantial improvements, most notably:
update in the inversion procedure for calving glaciers (Recinos et al., 2019)
new glacier evolution model based on Marzeion et al., 2012
We don’t expect major changes in the model results due to this release.
run_untilnow makes sure that the years (months) are not crossed by the adaptive time-stepping scheme (GH710).
run_until_and_storeshould now be consistent. The change is unlikely to affect the majority of users (which used
run_until_and_store), but the results or
run_untilcan be affected (PR726). By Matthias Dusch.
gridded_mb_attributestasks to add raster glacier attributes such as slope, aspect, mass balance… to the glacier directory (PR725). This can be useful for statistical modelling of glacier thickness. By Matteo Castellani.
Added support for another evolution model: the volume-area-scaling based model of Marzeion et al., 2012 (PR662). This is a major enhancement to the code base as it increases the number of choices available to users and demonstrates the modularity of the model. By Moritz Oberrauch.
Changed the way the calving flux is computed during the ice thickness inversion. This no longer relies on an iteration over mu*, but solves for h instead. The new function is likely to replace the “old” calving loop (PR794). By Fabien Maussion.
the flowline model gives access to new diagnostics such as ice velocity and flux along the flowline. The numerical core code changed in the process, and we will monitor performance after this change (PR853). By Fabien Maussion.
Minor release of the OGGM model, with several bugfixes and some improvements.
We don’t expect any change in the model results due to this release.
Added (and updated) an official task to find the calving flux based on the mass-conservation inversion (inversion.find_inversion_calving). This is still in experimentation phase! (PR720). By Beatriz Recinos.
This is a major new release of the OGGM model, with substantial improvements to version 1. We recommend to use this version from now on. It coincides with the publication of our publication in Geoscientific Model Development.
New contributors to the project:
Matthias Dusch (PhD student, University of Innsbruck), added extensive cross-validation tools and an associated website.
Philipp Gregor (Master student, University of Innsbruck), added options to switch on lateral bed stress in the flowline ice dynamics
Nicolas Champollion (PostDoc, University of Bremen), added GCM data IO routines.
Sadie Bartholomew (Software Engineer, UK Met Office), added ability to replace colormaps in graphics with HCL-based colors using python-colorspace.
The utils.copy_to_basedir() function is changed to being an entity task. In addition gcm_data files, when present, will from now on always be copied when using this task (GH467 & PR468). By Anouk Vlug.
The method used to apply CRU and GCM anomalies to the climatology has changed for precipitation: we now use scaled anomalies instead of the standard anomalies (PR393). The previous method might have lead to negative values in some cases. The corresponding reference t* have also been updated (PR407). This change has some consequences on the the model results: cross-validation indicates very similar scores, but the influence on global model output has not been assessed yet. By Fabien Maussion.
It is now possible to run a simulation with spinup in the standard workflow (PR411). For this to happen it was necessary to clean up the many *filesuffix options. The new names are more explicit but not backwards compatible. The previous filesuffix is now called output_filesuffix. The previous input_filesuffix is now called climate_input_filesuffix. The random_glacier_evolution task is now called run_random_climate for consistency with the other tasks See the PR linked above for more info. By Fabien Maussion.
Rework of the 2d interpolation tasks for ice thickness in the context of ITMIX2. The new interpolation are better, but not backwards compatible. Aside of me I don’t think anybody was using them (PR465). By Fabien Maussion.
Diagnostic variables (length, area, volume, ELA) are now stored at annual steps instead of monthly steps (PR488). The old behavior can still be used with the
store_monthly_stepkwarg. Most users should not notice this change because the regionally compiled files were stored at yearly steps anyways. By Fabien Maussion.
The list of reference t* dates is now generated differently: instead of the complex (and sort of useless) nearest neighbor algorithm we are now referring back to the original method of Marzeion et al. (2012). This comes together with other breaking changes, altogether likely to change the results of the mass balance model for some glaciers. For more details see the PR: PR509 By Fabien Maussion.
The ice dynamics parameters (Glen A, N, ice density) are now “real” parameters accessible via
cfg.PARAMS(PR520, GH511 and GH27). Previously, they were also accessible via a module attribute in
cfg, which was more confusing than helping. Deprecated and removed a couple of other things on the go, such as the dangerous `
optimize_inversion_paramstask (this cannot be optimized yet) and the useless
volume_inversionwrapper (now called
mass_conservation_inversion) By Fabien Maussion.
The temperature sensitivity mu* is now flowline specific, instead of glacier wide. This change was necessary because it now allows low-lying tributaries to exist, despite of too high glacier wide mu*. This change had some wider reaching consequences in the code base and in the mass balance models in particular: PR539. This will also allow to merge neighboring glaciers in the future. By Fabien Maussion.
The shapefiles are now stored in the glacier directories as compressed tar files, addressing GH367 & GH615. This option can be turned off with cfg.PARAMS[‘use_tar_shapefiles’] = False. By Fabien Maussion.
The old GlaThiDa file linking the total volume of glaciers (T database) to RGI has been updated to RGI Version 6 (PR403). Generally, we do not recommend to use these data for calibration or validation because of largely unknown uncertainties. By Fabien Maussion.
The computing efficiency of the 2D shallow ice model has been increased by a factor 2 (PR415), by avoiding useless repetitions of indexing operations. The results shouldn’t change at all. By Fabien Maussion.
Added optional shape factors for mass-conservation inversion and FluxBasedModel to account for lateral drag dependent on the bed shape (PR429). Accepted settings for shape factors are None, ‘Adhikari’ (Adhikari & Marshall 2012), ‘Nye’ (Nye, 1965; equivalent to Adhikari) and ‘Huss’ (Huss & Farinotti 2012). Thorough tests with applied shape factors are still missing. By Philipp Gregor.
Some amelioration to the mass balance models (PR434). Added a
repeatkwarg to the
PastMassBalancein order to loop over a selected period. Added an
UncertainMassBalancemodel which wraps an existing model and adds random uncertainty to it. By Fabien Maussion.
Added a new parameter (
PARAMS['use_rgi_area']), which specifies whether OGGM should use the reference area provided by RGI or the one computed from the local map and reprojected outlines (PR458, default: True). By Matthias Dusch.
simple_glacier_maskstasks allows to compute glacier rasters in a more robust way than the default OGGM method (PR476). This is useful for simpler workflows or to compute global statistics for external tools like rgitools. This task also computes hypsometry files much like RGI does. By Fabien Maussion.
process_histalp_datatask to run OGGM with HISTALP data automatically. The task comes with a list of predefined t* like CRU and with different default parameters (see blog). The PR also adds some safety checks at the calibration and computation of the mass balance to make sure there is no misused parameters (PR493). By Fabien Maussion.
process_cesm_datafunction has been split into two functions, to make it easier to run oggm with the climate of other GCM’s:
process_cesm_datareads the CESM files and handles the CESM specific file logic.
process_gcm_datais the general task able to handle all kind of data.
process_cesm_datacan also be used as an example when you plan make a function for running OGGM with another GCM (GH469 & PR582). Anouk Vlug.
Added a workflow
merge_glacier_taskswhich merges tributary/surrounding glaciers to a main glacier, allowing mass exchange between them. This is helpful/necessary/intended for growing glacier experiments (e.g. paleoglaciology) (PR624). By Matthias Dusch.
entity_taskdecorator now accepts a fallback-function which will be executed if a task fails and cfg.PARAMS[‘continue_on_error’] = True. So far only one fallback function is implemented for climate.local_t_star (PR663). By Matthias Dusch.
New process_gcm_data task to handle CMIP5 files. By Nicolas Champollion.
v1.0 (16 January 2018)#
This is the first official major release of OGGM. It is concomitant to the submission of a manuscript to Geoscientific Model Development (GMD).
This marks the stabilization of the codebase (hopefully) and implies that future changes will have to be documented carefully to ensure traceability.
New contributors to the project:
Anouk Vlug (PhD student, University of Bremen), added the CESM climate data tools.
Anton Butenko (PhD student, University of Bremen), developed the downstream bedshape algorithm
Beatriz Recinos (PhD student, University of Bremen), participated to the development of the calving parametrization
Julia Eis (PhD student, University of Bremen), developed the glacier partitioning algorithm
Schmitty Smith (PhD student, Northand College, Wisconsin US), added optional parameters to the mass balance models
v0.1.1 (16 February 2017)#
Minor release: changes in ITMIX to handle the synthetic glacier cases.
It was tagged only recently for long term documentation purposes and storage on Zenodo.
v0.1 (29 March 2016)#
Initial release, used to prepare the data submitted to ITMIX (see here).
This release is the result of several months of development (outside of GitHub for a large part). Several people have contributed to this release:
Michael Adamer (intern, UIBK), participated to the development of the centerline determination algorithm (2014)
Kévin Fourteau (intern, UIBK, ENS Cachan), participated to the development of the inversion and the flowline modelling algorithms (2014-2015)
Alexander H. Jarosch (Associate Professor, University of Iceland), developed the MUSCL-SuperBee model (PR23)
Johannes Landmann (intern, UIBK), participated to the links between databases project (2015)
Ben Marzeion (project leader, University of Bremen)
Fabien Maussion (project leader, UIBK)
Felix Oesterle (Post-Doc, UIBK) provided the AWS deployment script (PR25)
Christian Wild (master student, UIBK), participated to the development of the centerline determination algorithm (2014)