Profile

Forecast at the selected point

What am I seeing?

→ Live forecasts of optical turbulence over the Netherlands

Optical turbulence forecasts are produced by postprocessing the output of the Global Forecast System (GFS) with OTProf. GFS forecasts are available live and for free (see Licenses for details). A new GFS forecast is produced four times a day (00 UTC, 06 UTC, 12 UTC, 18 UTC), with 16 days lead time. Here, we utilize only the first five forecast days (120 hours). The website updates automatically whenever a new GFS run becomes available, and the Forecast run selector allows switching between them.

Note, GFS's shape is compatible with the ERA5 pressure level training data (0.25 degree, 30+ pressure levels), but we did not perform a GFS-specific performance analysis. GFS is produced by a different numerical weather model with different errors and biases. We expect qualitative performance to be acceptable, but do not give any warranty.

How it is made

OTProf and this website are developed by Maximilian Pierzyna. This website shows live forecasts and a climatology of optical turbulence profiles () and column-integrated parameters (Fried parameter and scintillation index ) over the Netherlands, estimated by the deep learning model OTProf. The OTProf system is composed of the following three packages:

  1. otprof-wrf: Code and config files to generate training data by post-processing output from the Weather Research and Forecasting (WRF) model with the variance-based parameterization of He and Basu (2015).
  2. otprof: Code for training and inference of the OTProf deep learning model.
  3. otprof-ui: Code of the web UI to show OTProf-based forecasts and climatology, i.e. this website, and of the pipelines that compute them from GFS and ERA5.

The method and its performance are described in 📄 Pierzyna, Basu, Saathof (2026, preprint).

How to cite

If you use OTProf products, please cite Pierzyna, Basu, and Saathof (2026). The products are licensed under CC BY-SA 4.0, so please credit them and share derived work under the same license. See Licenses for details.

Details

Licenses

Disclaimer: Experimental research product

The optical turbulence (OT) forecasts and climatology on this page are experimental research products generated by OTProf, a machine-learning model. They serve as technology demonstrators and are intended to give a qualitative, order-of-magnitude picture of the spatio-temporal evolution of OT () in the Netherlands.

The OTProf products are provided "as is", without warranty of any kind. Use for operational decisions is entirely at your own risk; the authors and Technische Universiteit Delft accept no liability. Validation against local observations is strongly encouraged. Forecasts may be delayed, incomplete, or discontinued without notice, and methods may change between versions.

Known limitations

  • Trained on model output, not observations. OTProf learns from Weather Research and Forecasting (WRF) model simulations and inherits WRF's known biases relative to observations. Fine-tuning on observations is future work.
  • No horizontal consistency. OTProf is a 1D column model. Each grid column is predicted independently. Interpret spatial patterns with caution.
  • Different input data for forecasts. The climatology is derived from ERA5, matching the setup evaluated in our publication. The forecasts, however, use GFS input, which comes from a different weather model with its own errors and biases. We have not yet assessed performance on GFS input but expect it to be qualitatively similar.
  • Performance. Model skill and limitations are documented in our publication. Users are responsible for familiarizing themselves with them.

The OTProf products are the optical turbulence forecasts and the optical turbulence climatology on this website, delivered as interactive plots and as NetCDF and CSV files. The OTProf products are licensed under CC BY-SA 4.0, which excludes any warranty. They are derived from two third-party datasets listed below, which permit redistribution of derivatives.

The source code of this website and its data pipelines, OTProf UI, is licensed under the GNU General Public License v3.0 or later if not stated otherwise.

The source code of the deep learning model used to estimate optical turbulence from the input datasets, OTProf, is licensed under the GNU General Public License v3.0 or later if not stated otherwise. The architecture of the OTProf model (SQF-H) is derived from the Kaggle notebook "LEAP: Training 1" by Ron Shlomo, licensed under the Apache License 2.0. The weights of the pretrained models are licensed under CC BY-SA 4.0 and are available on Hugging Face.

The OTProf training data are generated using the Weather Research and Forecasting (WRF) model v4.5.2. The code and config files to run WRF for this purpose are licensed under the GNU General Public License v3.0 or later and published on GitHub as OTProf WRF. The WRF output (i.e., the training data) is available on Zenodo under CC BY-SA 4.0 (DOI: 10.5281/zenodo.23733024).

Technische Universiteit Delft hereby disclaims all copyright interest in the programs "OTProf" and "OTProf UI" written by the Author(s). Henri Werij, Faculty of Aerospace Engineering, Technische Universiteit Delft.

© 2026, Maximilian Pierzyna (excluding third-party code and data listed below)

Third-party input data

OTProf optical turbulence forecasts are derived from data provided by the NOAA Global Forecast System (GFS), taken from the AWS Registry of Open Data every 6 hours whenever a new GFS cycle becomes available. The original NOAA data have been altered by OTProf: the forecasts are not original NOAA data, and NOAA does not endorse them.

The OTProf-based optical turbulence climatology is derived from ERA5 pressure level data (Hersbach et al., 2023: ERA5 hourly data on pressure levels from 1940 to present, DOI: 10.24381/cds.bd0915c6), provided by the Copernicus Climate Change Service (C3S) Climate Data Store (CDS) under the CC BY 4.0 licence. The climatology contains modified Copernicus Climate Change Service data (2026): OTProf estimates optical turbulence from the ERA5 data, and the estimates are aggregated into monthly statistics.

Third-party assets

OTProf UI ships the following third-party assets:

Download OTProf products

Disclaimer: Experimental research product

The optical turbulence (OT) forecasts and climatology on this page are experimental research products generated by OTProf, a machine-learning model. They serve as technology demonstrators and are intended to give a qualitative, order-of-magnitude picture of the spatio-temporal evolution of OT () in the Netherlands.

The OTProf products are provided "as is", without warranty of any kind. Use for operational decisions is entirely at your own risk; the authors and Technische Universiteit Delft accept no liability. Validation against local observations is strongly encouraged. Forecasts may be delayed, incomplete, or discontinued without notice, and methods may change between versions.

Known limitations

  • Trained on model output, not observations. OTProf learns from Weather Research and Forecasting (WRF) model simulations and inherits WRF's known biases relative to observations. Fine-tuning on observations is future work.
  • No horizontal consistency. OTProf is a 1D column model. Each grid column is predicted independently. Interpret spatial patterns with caution.
  • Different input data for forecasts. The climatology is derived from ERA5, matching the setup evaluated in our publication. The forecasts, however, use GFS input, which comes from a different weather model with its own errors and biases. We have not yet assessed performance on GFS input but expect it to be qualitatively similar.
  • Performance. Model skill and limitations are documented in our publication. Users are responsible for familiarizing themselves with them.

By downloading OTProf products you accept that they are experimental and come without warranty of any kind; you use them at your own risk. They are licensed under CC BY-SA 4.0; see Licenses for the terms of use.