Quickstart reconstruction¶

After getting Python installed and downloading the LMR sample data, you are now set to run a simple reconstruction of surface air temperature!

In this section we’ll start with the LMR_lite.py script which runs a reconstruction but hides many of the details. In this section we’ll run through a simple reconstruction that uses the downloaded sample data.

Note

Many of the utility functions used in this script and in LMR_lite_utils.py are useful for running LMR in jupyter notebook environments, which promotes rapid protyping with data stored in memory.

Configuration¶

To start off, the configuration files need to be copied into the main source code directory for LMR. Wherever you cloned/downloaded the source code (we’ll use the path /home/disk/foo/LMR_src for our code directory) there should be a config_templs/ folder which holds configuration templates. From the LMR_src directory, there are two files you need to copy to there to perform an experiment:

$ cp config_templs/config_lite_template.yml ./config_lite.yml
$ cp config_templs/LMR_config_template.py ./LMR_config.py

The config_lite.yml file holds a subset of the configuration parameters we’ll need for the reconstruction, while LMR_config.py is the main file for all configuration business logic.

Next, you’ll have to edit a line in config_lite.yml. In this file, edit the path lmr_path defined in the core section. It should be changed to point to the input data folder (e.g., /home/disk/foo/LMR_data or whatever you defined when unzipping the sample data)

core:
  nexp: test_lmr_recon
  lmr_path: /home/path/to/LMR/

Note

There are many more options that allow for fine control over a reconstruction. Please see LMR configuration for details. The sample files provided let you run this experiment out of the box. Other configuration changes may require you to recreate intermediate files associated with PSM calibration and estimated proxy values (Ye values) from the prior (climate model data).

Running LMR_lite¶

After making the configuration change the reconstruction is ready for launch!

If you installed an Anaconda environment, make sure that you have the correct one activated. E.g.,

$ source activate lmr_py3

Then you can run a reconstruction using:

$ (lmr_py3) python LMR_lite.py

The code will print out reconstruction progress in your terminal. After the reconstruction is finished you’ll see the total time elapsed, a few plot windows will open up if you are running locally, and you’ll see that a file of analyzed output will be saved in the source directory (analyses_1900_2000_1880_2000.npz). And that’s all for a simple reconstruction! For an in-depth description of configuring and running a reconstruction from start to finish please see Performing a full reconstruction.

Example end of reconstruction output¶

gmt=    0.364120 nhmt=    0.315032 shmt=    0.414933
gmt=    0.215862 nhmt=    0.181326 shmt=    0.251612
gmt=    0.142075 nhmt=    0.119087 shmt=    0.165872
gmt=    0.196176 nhmt=    0.187506 shmt=    0.205150
gmt=    0.310819 nhmt=    0.344414 shmt=    0.276043
gmt=    0.221263 nhmt=    0.284358 shmt=    0.155951
gmt=    0.403800 nhmt=    0.503617 shmt=    0.300477
gmt=    0.397144 nhmt=    0.444159 shmt=    0.348477
gmt=    0.540332 nhmt=    0.636867 shmt=    0.440407
gmt=    0.345345 nhmt=    0.314784 shmt=    0.376979
gmt=    0.555479 nhmt=    0.623369 shmt=    0.485203
gmt=    0.590704 nhmt=    0.662194 shmt=    0.516702
gmt=    0.332070 nhmt=    0.360107 shmt=    0.303047
gmt=    0.396883 nhmt=    0.546958 shmt=    0.241537
gmt=    0.595685 nhmt=    0.762036 shmt=    0.423490
gmt=    0.537395 nhmt=    0.650302 shmt=    0.420523
gmt=    0.455852 nhmt=    0.482466 shmt=    0.428303
gmt=    0.479458 nhmt=    0.575316 shmt=    0.380233
gmt=    0.603113 nhmt=    0.718212 shmt=    0.483970
gmt=    0.424851 nhmt=    0.564871 shmt=    0.279912
analyses_1900_2000_1880_2000.npz exists...loading it
returning global means...
failed to get the current screen resources
saving to .png

-----------------------------------------------------
Reconstruction completed in 9.480862776438395 mins
-----------------------------------------------------

Figure of global mean temperatures against analysis data produced by LMR_lite.py.