Shortwave Intermittency

This is a set of scripts for testing shortwave intermittency in the ACRANEB2 radiative transfer scheme using the acra2 SCM (Single Cell Model) for the master's degree thesis Broadband approach as a framework for implementation of radiative transfer scheme with selective intermittency: Cost versus accuracy study.

• Python 2.6 or newer
• R
• numpy
• jinja2
• rjson

1. Install requirements (Ubuntu/Debian):

    apt-get install python python-numpy r-base
   

2. Install python packages:

    pip install --user jinja2
   

3. Install R packages:

    echo "install.packages('rjson', repos='http://cran.rstudio.com/')" | R --no-save
   

Create a symbolic link to acra2 (SCM model executable) in the scripts directory:

cd scripts
ln -s /path/to/scm/bin/acra2

All scripts are contained in the scripts directory.

Prepare input for run.py (see below). Calculates zenithal angles according to chosen intermittency constants specified inside the script.

input.py <start> <stop> <interval>

• start

  Starting angle in degree.

• stop

  Stop angle in degrees.

• interval

  Intermittency interval in degrees.

Run the SCM model multiple times for different values of zenithal angle, optionally supplying given optical thicknesses to the model.

run.py <namelist> <input>

• namelist

  A template of Fortran namelist for the SCM model. The template is formatted using jinja2, supplied with variables:

  • mu0 – zenithal angles
  • optical_thickness_downward – downward optical thickness (if present in input)
  • optical_thickness_upward – upward optical thickness (if present in input)

• input

  Input specification. Example:

    {
    	"mu0": [0, 0.5, 1]
    	"optical_thickness_downward": [
    		[...],
    		[...]
    	],
    	"optical_thickness_upward": [
    		[...],
    		[...]
    	]
    }
  

  mu0 specifies zenithal angles at which to run the model. optical_thickness_downward and optical_thickness_upward are optional and specify given optical depths with which the model is forced to run.

Plot datasets from a product (output of run.py). The script produces a PDF plot for each dataset in the product.

plot_product.R <product> <dir>

• product

  Product containing datasets as output by run.py.

• dir

  Output directory.

Perform linear interpolation of optical depths. Outputs JSON suitable as input for run.py.

interpolate.R <product> <input>

• product

Product file as output by run.py. Optical depths from the product are used as interpolation points.

• input

Input file as in run.py. Determines points where to interpolate.

Perform comparison of a product with respect to reference.

Outputs a product containing datasets which are differences between the corresponding datasets in product and reference.

validation.R <product> <reference>

• product

  Product to be validated.

• reference

  Product to be used as a reference.

Note: Before running the example, add the directory containing scripts to PATH: export PATH="$PATH:/path/to/scripts".

# Prepare input for run.py:
# Angles from 0 to 90 deg by 15-deg steps (coarse).
input.py 0 90 15 > input_15deg.json

# Angles from 0 to 90 deg by 1-deg steps (dense).
input.py 0 90 1 > input_1deg.json

# Run SCM with the coarse input.
run.py nam.template input_15deg.json > product.json

# Run SCM with the dense input as reference.
run.py nam.template input_1deg.json > product_ref.json

# Plot product to directory product.
plot_product.R product.json product
plot_product.R product_ref.json product_ref

# Perform interpolation of optical depths at 1-degree intervals.
interpolate.R product.json input_1deg.json > input_linear_interp.json

# Run SCM with the interpolated optical depths.
run.py nam.template input_linear_interp.json > product_linear_interp.json

# Plot product to directory product_linear_interp.
plot_product.R product_linear_interp.json product_linear_interp

# Compare interpolation to reference.
validation.R product_linear_interp.json product_ref.json > validation.json

# Plot comparison.
plot_product.R validation.json validation

Under the directory analysis you can find analyses of a number of cases:

• clear: Clear-sky atmosphere, 87 layers
• cloud: Cloudy atmosphere, 87 layers
• cases/l87/: Additional cases, 87 layers
• cases/l122: Additional cases, 122 layers

Additional cases:

• c25: Tropical
• c27: Mid-latitude summer
• c29: Mid-latitude winter
• c31: Subarctic summer
• c33: Subarctic winter

Namelists of the respective cases can be found in the directory namelists.

Each directory contains files as in the example above, especially:

• product_ref: Plots of the reference (non-intermittent) run.
• product_linear_interp: Plots of the interpolated run (linear interpolation).
• validation: Comparison of the reference and interpolated runs.

This software is available under the terms of the MIT license.