MT-Probabilistic_Meteorological_Forecasting

This repository contains the codes developed during an internship in Airbus and used in the Master Thesis, also in this repository.

Airbus Copyright

All the code here is owned by Airbus

What is on the code?

01-Request_NOAA_data-GFS_GENS.ipynb

This file contains a script capable of download GFS or GENS data from NOAA's website. Some smaller modifications may be necessary to adapt it to your necessity (days, forecast hours, etc).

02a-grib_Parser-to-PANDAS.ipynb

A code capable of parsing the .grib data download using the previous notebook. The data will be parsed to a DataFrame (which is not optimal depending on how you intends to use the data).

02b-grib_Parser-to-Matrix_numpy.ipynb

Similar to the previous one, but the resulting data will be a numpy matrix.

03a-data-extracting_and_visualizing.ipynb

Code dedicated to visualize the data punctually.

03b-Analysis of Dependence.ipynb

A brief analysis of how the value of the wind is dependent from the value close to it. It evaluated de dependencies in space and also in time.

04a-data-verifying_punctual_metric(slow).ipynb

Application of the punctual metric and evaluation of some distributions using it (Gaussian, KDE, Histogram and best fitting). (slow) -> since it uses DataFrame in a non-efficient way, the calculations are slow.

04b-data-verifying_CRPS(slow).ipynb

Application of the CRPS metric and evaluation of the Gaussian distribution and the Histogram. (slow) -> since it uses DataFrame in a non-efficient way, the calculations are slow.

05-simplified_model_validation-sigma_non-variant_in_time.ipynb

Application and analysis of the simplified model, i.e., $\sigma$ constant in time. Evaluation of the results using CRPS.

06a-Autoencoder.ipynb

Autoencoder application.

06b-GADO.ipynb

Generative Adversarial Model first implementation