Welcome to the Last Millennium Reanalysis Online (LMR Online) framework code base-- a toolset for climate field reconstructions using ensemble data assimilation. This is an offshoot ofthe original LMR Framework (Hakim et al. 2016). This repository contains the code used for reconstructions in "Coupled Atmosphere--Ocean Reconstruction of the Last Millennium Using Online Data Assimilation, Perkins and Hakim, submitted 2020"
This package is based off of the original LMR framework (Hakim et al. 2016) with enhancements to include climate forecasts via a linear inverse models, (pyLIM from Perkins & Hakim 2020).
The original LMR documentation page has some information on components, but is currently somewhat outdated for this version.
- Free software provided under the modified BSD License
- Tested and running on Python 3.6+
The installation requires the conda package manager. A minimal Miniconda installation is sufficient to install the environment.
To install an environment that replicates the linux environment used to perform the reconstructions, run:
./create_lmr_py3_env.sh
To change the name of the conda environment, simply adjust the ENV variable in that script. It defaults to lmr_env.
Additionally, you will need to install the pyLIM package. Outside of the LMROnline directory:
git clone https://www.github.com/frodre/pyLIM
cd pyLIM
conda activate <ENV>
python setup.py install
For convenience, we provide a prepared directory on Zenodo with pre-processed data ready to run experiments from the coupled atmosphere-ocean reconstruction paper.
Copy any of the configs to the LMROnline directory
cp configs\config_production_ccsm4_pagesv2_wCoral_gis_seaslinobj_offline_past1000.yml LMROnline/config.yml
cp LMROnline/config_templs/LMR_config_template.py LMROnline/LMR_config.py
Adjust the opening lines of the configuration to match desired output directories
wrapper:
iter_range: !!python/tuple [0, 10]
multi_seed: 89
datadir_output: /tmp/user/lmr_output
archive_dir: /path/to/final/output
nexp: production_ccsm4_pagesv2_wCoral_gis_seaslinobj_offline_past1000
core:
lmr_path: /path/to/LMR/input/data
Note lmr_path must point to the directory of the pre-processed data parent directory.
From the LMROnline directory, start a reconstruction with
python LMR_wrapper.py config.yml