B2SHARE Server Forschungszentrum Jülich
Not a member yet
379 research outputs found
Sort by
Forecasting near-surface ozone using temporally decomposed input variables and deep neural networks
No full textPoster "Forecasting near-surface ozone using temporally decomposed input variables and deep neural networks" and preprint "O3ResNet: A deep learning based forecast system to predict local ground-level daily maximum 8-hour average ozone
TOAR-II Data User Workshop June/July 2022
No full textThe Tropospheric Ozone Assessment Report (TOAR) is an initiative of the International Global Atmospheric Chemistry (IGAC) project. TOAR-II is the second phase of TOAR. It builds on the successful completion of the first comprehensive assessment on tropospheric ozone and will last from 2020 to 2024. The TOAR-II Data User Workshop 2022 introduced the new TOAR database and its tools for accessing and analysing TOAR data. This publication contains all lectures that were held during the on-site event.
Furthermore, users' workflows and data analysis problems have been tackled in hands-on sessions. The feedback from this workshop will also influence the further development of the TOAR infrastructure. All workshop material including the hands-on codes can be found at https://gitlab.jsc.fz-juelich.de/esde/toar-public/toar-data-user-workshop-2022
Vehicle measured irradiance data
No full text- "gps time utc": unix timestamp
- "timestr": time string in the format "%Y-%m-%d_%H:%M:%S" with UTC time zone (reformatted "gps time utc")
- "latitude", "longitude": gps coordinates
- "gps altitude (m)", "gps speed (m/s)", "gps course (deg)", "gps climb (m/min)": altitude, speed, bearing and climb rate
- "gps EPX Estimated * Error": gps estimated errors
- "Magnetic Bearing [deg.]", "Magnetic Tilt [deg.]": magnetic compass bearing and tilt
- "module", "sensor_azimuth", "sensor_zenith": side of the irradiance sensor ("module" attains values: "roof", "left", "right", "rear"). "sensor_azimuth" and "sensor_zenith" define orientation of the irradiance sensor. azimuth computed from magnetic compass, zenith 0 corresponds to the sky orientation
- "irradiance_Wm2": irradiance sensor irradiance and temperature readings
- "GHI", "DHI", "Clear sky GHI", "Clear sky DHI": copernicus irradiance values
- "geohash": region id which is used in training dataset split
The "train.csv.gz" contains the actual measurement data for the "irradiance_Wm2" field. The "submission_*.csv.gz" contains "GHI" values for the "irradiance_Wm2", which provides are very simple baseline model that ignores the topography
FastSurferVINN
No full textTraining checkpoints for FastSurferVINN (https://github.com/Deep-MI/FastSurfer) - please cite the paper when using this resource (https://doi.org/10.1016/j.neuroimage.2022.118933).Leading neuroimaging studies have pushed 3T MRI acquisition resolutions below 1.0 mm for improved structure definition and morphometry. Yet, only few, time-intensive automated image analysis pipelines have been validated for high-resolution (HiRes) settings. Efficient deep learning approaches, on the other hand, rarely support more than one fixed resolution (usually 1.0 mm). Furthermore, the lack of a standard submillimeter resolution as well as limited availability of diverse HiRes data with sufficient coverage of scanner, age, diseases, or genetic variance poses additional, unsolved challenges for training HiRes networks. Incorporating resolution-independence into deep learning-based segmentation, i.e., the ability to segment images at their native resolution across a range of different voxel sizes, promises to overcome these challenges, yet no such approach currently exists. We now fill this gap by introducing a Voxel-size Independent Neural Network (VINN) for resolution-independent segmentation tasks and present FastSurferVINN, which (i) establishes and implements resolution-independence for deep learning as the first method simultaneously supporting 0.7–1.0 mm whole brain segmentation, (ii) significantly outperforms state-of-the-art methods across resolutions, and (iii) mitigates the data imbalance problem present in HiRes datasets. Overall, internal resolution-independence mutually benefits both HiRes and 1.0 mm MRI segmentation. With our rigorously validated FastSurferVINN we distribute a rapid tool for morphometric neuroimage analysis. The VINN architecture, furthermore, represents an efficient resolution-independent segmentation method for wider application
Exploring decomposition of temporal patterns to facilitate learning of neural networks for ground-level daily maximum 8-hour average ozone prediction: Data
No full textThis record contains data for the manuskript "Exploring decomposition of temporal patterns to facilitate learning of neural networks for ground-level daily maximum 8-hour average ozone prediction" by L. H. Leufen, F. Kleinert and M. G. Schultz.We provide the complete experiment folders of the best trained networks (_network_daily). These contain the data used (data), the forecasts created (forecasts), the neural network used (model), graphics about the data and the evaluation (plots) as well as the exact results of the error analysis (latex_report). In addition, each experiment folder contains a start script (start_script.txt) with disabled train options, which can be used to restart the evaluation without overwriting or training the model. We have added an example file (run_Example.sh) how such a call could look like. To start a new experiment it required to update the start script accordingly. Furthermore, we provide the results of the uncertainty estimate of the MSE (analysis_data, stored in leufen-data-0.tar.gz) for the FCN experiments but also for the follow-up experiments with different NN architectures (FCN, RNN, CNN). All mentioned data are grouped and packed into several tar.gz archives (leufen-data-.tar.gz).To have a better insight into the data and experiment, we provide a ready-to-run jupyter notebook to load and visualize our data and results. To run this notebook we rely on docker. Download the docker file (leufen-docker.tar.gz) and follow the instructions (instructions.md and instructions.pdf) to load data, models, and the notebook. Note that changes made by the user to the notebook well be removed on exit as long the option "--rm" is present. When following the instructions, it is *not* required to unpack the data file (leufen-data-*.tar.gz). If you encount issues with disk space limits and docker, it is always possible to use a reduced number of data files or to unpack data for a single experiment and just parse them to the docker container. When using a windows host system, some commands provided in the instructions might slightly deviate.All experiments were carried out with the software MLAir in version 2.0.0 . A detailed description is given in https://doi.org/10.5194/gmd-14-1553-2021 and the source code can be found at https://gitlab.jsc.fz-juelich.de/esde/machine-learning/mlair
Exploring decomposition of temporal patterns to facilitate learning of neural networks for ground-level daily maximum 8-hour average ozone prediction: Data
No full textThis record contains data for the manuskript "Exploring decomposition of temporal patterns to facilitate learning of neural networks for ground-level daily maximum 8-hour average ozone prediction" by L. H. Leufen, F. Kleinert and M. G. Schultz.We provide the complete experiment folders of the best trained networks (_network_daily). These contain the data used (data), the forecasts created (forecasts), the neural network used (model), graphics about the data and the evaluation (plots) as well as the exact results of the error analysis (latex_report). In addition, each experiment folder contains a start script (start_script_no_train.txt) with disabled train options, which can be used to restart the evaluation without overwriting or training the model. We have added an example file (run_example.sh) how such a call could look like. To start a new experiment it required to update the start script accordingly. Furthermore, we provide the results of the uncertainty estimate of the MSE (analysis_data, stored in leufen-data-0.tar.gz) for the FCN experiments but also for the follow-up experiments with different NN architectures (FCN, RNN, CNN). All mentioned data are grouped and packed into several tar.gz archives (leufen-data-.tar.gz).To have a better insight into the data and experiment, we provide a ready-to-run jupyter notebook to load and visualize our data and results. To run this notebook we rely on docker. Download the docker file (leufen-docker.tar.gz) and follow the instructions (instructions.md and instructions.pdf) to load data, models, and the notebook. Note that changes made by the user to the notebook well be removed on exit as long the option "--rm" is present. When following the instructions, it is *not* required to unpack the data file (leufen-data-*.tar.gz). If you encount issues with disk space limits and docker, it is always possible to use a reduced number of data files or to unpack data for a single experiment and just parse them to the docker container. When using a windows host system, some commands provided in the instructions might slightly deviate.All experiments were carried out with the software MLAir in version 2.0.0 . A detailed description is given in https://doi.org/10.5194/gmd-14-1553-2021 and the source code can be found at https://gitlab.jsc.fz-juelich.de/esde/machine-learning/mlair .To rerun an experiment,
(1) it is required to install MLAir according to the installation instructions provided in the source code repository,
(2) download the file north_german_plain_background.json
(3) unpack the experiment folder
(4) adjust the placeholder in run_example.sh with the correct folder name, e.g. MBFCN_LT_ST_network_daily
(5) either execute run_example.sh or copy it's content and call the python command directly
TOAR-II data-gathering event, 21-Jan-2022
No full textThe Tropospheric Ozone Assessment Report (TOAR) is an initiative of the International Global Atmospheric Chemistry (IGAC) project. TOAR-II is the second phase of TOAR. It builds on the successful completion of the first comprehensive assessment on tropospheric ozone and will last from 2020 to 2024. A data-gathering event was organized via Zoom on January 21st, 2022 to inform the scientific community about the plans for TOAR-II and how to contribute data to the TOAR data portal and the TOAR database. This publication contains all presentation files from this event together with the video and audio recording, as well as transcripts of the question and answer sessions
FastSurferVINN
No full textTraining checkpoints for FastSurferVINN (https://github.com/Deep-MI/FastSurfer) - please cite the paper when using this resource (https://doi.org/10.1016/j.neuroimage.2022.118933).Leading neuroimaging studies have pushed 3T MRI acquisition resolutions below 1.0 mm for improved structure definition and morphometry. Yet, only few, time-intensive automated image analysis pipelines have been validated for high-resolution (HiRes) settings. Efficient deep learning approaches, on the other hand, rarely support more than one fixed resolution (usually 1.0 mm). Furthermore, the lack of a standard submillimeter resolution as well as limited availability of diverse HiRes data with sufficient coverage of scanner, age, diseases, or genetic variance poses additional, unsolved challenges for training HiRes networks. Incorporating resolution-independence into deep learning-based segmentation, i.e., the ability to segment images at their native resolution across a range of different voxel sizes, promises to overcome these challenges, yet no such approach currently exists. We now fill this gap by introducing a Voxel-size Independent Neural Network (VINN) for resolution-independent segmentation tasks and present FastSurferVINN, which (i) establishes and implements resolution-independence for deep learning as the first method simultaneously supporting 0.7–1.0 mm whole brain segmentation, (ii) significantly outperforms state-of-the-art methods across resolutions, and (iii) mitigates the data imbalance problem present in HiRes datasets. Overall, internal resolution-independence mutually benefits both HiRes and 1.0 mm MRI segmentation. With our rigorously validated FastSurferVINN we distribute a rapid tool for morphometric neuroimage analysis. The VINN architecture, furthermore, represents an efficient resolution-independent segmentation method for wider application
Representing chemical history in ozone time-series predictions - a model experiment study building on the MLAir (v1.5) deep learning framework: Experiments and source code
No full textHere we provide the source code and required data sets to reproduce all results related to "Representing chemical history in ozone time-series predictions - a model experiment study building on the MLAir (v1.5) deep learning framework" by F. Kleinert, L. H. Leufen, A. Lupascu, T. Butler and M. G. Schultz, submitted to GMD(D) (gmd-2022-122).
This record contains the MLAir source code, reference (competitor) experiments and forecasts.
As we could not make the full 4D (400x360x35) hourly wrf-chem model fields available for technical reasons, we decided to upload the subdomain of interest only. WRF-Chem model fields (YYYY-MM) are available from:
2009-01 to 2009-04: https://doi.org/10.34730/c799f04beb644e38a575fa20c2dd8d40
2009-05 to 2009-08: https://doi.org/10.34730/d5f34ae6a8e34d4c8ac33f75b993e8a9
2009-09 to 2009-12: https://doi.org/10.34730/a423ec9003194209989726a95a1a490c
2010-01 to 2010-03: https://doi.org/10.34730/718262bd2c894fd6aadce19a08040f69
This record contains the MLAir (v1.5) python framework used to conduct the experiments (mlair.tar.gz). The version is also available from the projects gitlab repository (tag Kleinert_etal_2022_initial_submission): https://gitlab.jsc.fz-juelich.de/esde/machine-learning/mlair/-/tree/Kleinert_etal_2022_initial_submission
Additionally, this record contains the competitor experiment directories (trained neural networks, input & target data, etc.) and our competitor models' forecasts. After extraction, you can use this forecast path as an external parameter in `run_wrf_dh_sector3.py`. Thus, MLAir can use the reference forecasts to calculate skill scores etc.
The `coords.nc` file contains the time-independent coordinates related to the model fields linked above and can be linked to MLAir by specifying the absolut path within the `run_wrf_dh_sector3.py` runscript (external_coords_file=coords.nc)
Global_Aerosol_OPP_profile_reanalysis_from_MERRA-2, vol.2.
No full textSee Global_Aerosol_OPP_profile_reanalysis_from_MERRA-2, vol.1 for full description