Skip to content

Project in progress. A collection of tools to statistically downscale climate data.

Notifications You must be signed in to change notification settings

chrystali2002/Linear_downscaling

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Linear_downscaling

Katrina Wheelan

July 31, 2020

NCAR

(IN PROGRESS)

This repository contains custom Jupyter notebooks, Python scripts, and batch scripts for linearly downscaling climate data.

Directories:

  • notebooks/ - Jupyter notebooks for interactive/customizable downscaling (these need to be updated)

    • Precip_DownScaling.ipynb - A Jupyter notebook that linearly downscales precipitation data using large scale predictors. The notebook employs a logistic model to predict yes/no precip and then a linear model to predict intensity. It has an optional stochastic component to correct the distribution.

    • Temp_Downscaling_MonthlyModel.ipynb - similarly downscales maximum temperature data using large scale predictors. The model controls by month, employing 3 linear methods: (1) manual linear regression, using the same predictors but different coefficients for each month; (2) LASSO linear regress using a single model; (3) LASSO regression conditioned on month. The chosen model also includes an optional stochastic component (an approximation of the "variance inflation" for SDSM).

    • Graphs.ipynb - Shows various metrics to compare different models

  • pythonScripts - A folder to store scripts called by bash scripts

    • regress.py - Optimizable Python file to do regression for downscaling
    • regression_tools.py - A module of functions to perform the necessary steps for downscaling
    • plotting.py - A module of plotting functions for the modeled data
    • SDSM_process.py - A short script I wrote to plot SDSM data.
    • settings.txt - Currently does nothing; I want to edit regress.py to read in settings from this file
  • bashScripts/ - Contains batch scripts for submission to the NCAR supercomputer

  • SDSM_p1/ - Plots and raw data from running SDSM for daily maximum temperature on p1 (lat = 38.125, lon = -101.875) with variance inflation = 1

    • plots/
      • Compare - A few plots comparing the Python output to SDSM output for predictions
      • SDSM/ - Plots of just the SDSM output
    • betas_python.txt - the coefficients generated by the python program
    • betas.txt - the coefficients generated by SDSM (a txt version of the PAR file)
    • tmax_p1_output.OUT - SDSM OUT file of weather generation (variance inflation = 1)
    • tmax_p1_output.SIM - SDSM metadata file
    • tmax_p1.PAR - SDSM PAR file of betas

About

Project in progress. A collection of tools to statistically downscale climate data.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 92.9%
  • Python 6.9%
  • Shell 0.2%