This readme document should give an overview of the project and its main components. For a full documentation of the newly developed pollen code to assimilate real-time pollen date into COSMO/ICON please refer to this confluence page: https://service.meteoswiss.ch/confluence/x/dYQYBQ (access has to be requested outside of Meteoswiss domain) There is e second git-repo available here: https://github.com/sadamov/cosmo.git that contains the newly developed Fortran Subroutines to assimilate the real-time pollen data in COSMO/ICON.
The analysis was conducted in R 4.2.1. The vignettes are Rmarkdown notebooks. The project is set up as a minimal R-package to assure maximum reproducibility (https://r-pkgs.org/index.html).
All necessary libraries and R-packages were installed using conda (https://docs.conda.io/en/latest/miniconda.html). Make sure you have conda installed before attempting to reproduce the steps below.
So to reproduce the full analysis, carry out these steps:
# Clone git repo
git clone git@github.com:sadamov/realtime_calibration.git
# Step into directory
cd realtime_calibration
# Create conda environment with all required dependencies
conda env create -n realtime_cal -f environment.yml
# Activate Conda Environment
conda activate realtime_cal
# Either work in an interactive R-Session (e.g. within RStudio or VSCode) or knit the document directly
R -e "rmarkdown::render('vignettes/analysis.Rmd')"
There is one branch in this repo:
- Main: Protected main branch containing the latest fully functional version of all vignettes and scripts.
There is quite a lot of external data required for this analysis and only people working at MeteoSwiss will have access to all of it. In the folder /ext-data scripts are stored that will retrieve and preprocess the data from external sources. The ready-made data is then stored in the /data folder and accessed by various scripts. To rerun the analysis only the data inside the /data folder is required.
- dwh: Text files of Pollen-Measurements (Concentrations 1/m^3) averaged daily and hourly. Daily surface temperatures. The data is retrieved with the ruby script dwh_retrieve(). Used in various vignettes.
- cosmo: Text file of Pollen Concentrations (1/m^3) predicted by COSMO for one specific hour. The data is retrieved with Fieldextra.
- other: Dataframes containing names and abbreviations of Swiss pollen stations and species. Manually typed (information available in various MeteoSwiss documentations) and used in various vignettes.
- Analysis.Rmd: The whole analysis in one location, loading the data from the /data folder and calculating all metrics and plots. Can be knitted into html or pdf format for a nicer reading experience.
- Analysis.R: The same content as the Rmd-file, can be alternatively used in old-school R setups.