Database Assembly using cfr’s built in PAGES 2k database

Database Assembly using cfr’s built in PAGES 2k database#

Author#

Tanaya Gondhalekar

This notebook walks through the process of building a proxy database containing only the PAGES 2k (v2.0.0) proxy records. This is the most straighforward and shortest way to obtain a ProxyDatabase to use for an emulation of LMRv2.1.

Key steps include:

  • Filtering for annual and subannual records with sufficient temporal coverage

  • Optionally filtering by the archive types used in Tardif et al. (2019)

The final result is saved as a NetCDF file that can be used in subsequent data assimilation notebooks.

import cfr

octave not found, please see README

Use cfr’s .fetch() to pull PAGES2kv2 from the cloud#

This is a ProxyDatabase object

pdb = cfr.ProxyDatabase().fetch('PAGES2kv2')

Use ProxyDatabase.filter() to filter by resolution#

We want only annual and subannually resolved records, hence dt ≤ 1.2 (a little higher than 1.0 as buffer)

filtered_pdb = pdb.filter(by='dt', keys= (0.0,1.2))

(Optional) Use the same filtering method for archive type#

Instead of dt, we use ptype to specify which archives we want used. This eliminates certain records despite being annual/subanually resolved.

filtered_pdb_2= filtered_pdb.filter(by='ptype', keys=['coral', 'tree', 'bivalve', 'ice', 'lake.varve_thickness', 'lake.varve_property', 'lake.accumulation'])

print(len(filtered_pdb.records))
print(len(filtered_pdb_2.records))

571
558

Save as a netCDF file to use in the data assimilation workflow#

filtered_pdb.to_nc('./prev_data/filtered_res_cfr_pages2kv2.nc')

>>> Warning: this is an experimental feature.
>>> Warning: this is an experimental feature.


100%|██████████| 571/571 [00:19<00:00, 29.11it/s]

>>> Data before 1 CE is dropped for records: ['NAm_019', 'Arc_026', 'NAm_011', 'Ant_010', 'Eur_003', 'Ant_007', 'Ant_028', 'Arc_033', 'SAm_030', 'Arc_032', 'Arc_002', 'Aus_001', 'Arc_020', 'Arc_022', 'Arc_004', 'Eur_006'].
ProxyDatabase saved to: ./prev_data/filtered__res_cfr_pages2kv2.nc


filtered_pdb_2.to_nc('./prev_data/filtered_ptype_res_cfr_pages2kv2.nc')

>>> Warning: this is an experimental feature.
>>> Warning: this is an experimental feature.


100%|██████████| 558/558 [00:18<00:00, 30.24it/s]

>>> Data before 1 CE is dropped for records: ['NAm_019', 'Arc_026', 'NAm_011', 'Ant_010', 'Eur_003', 'Ant_007', 'Ant_028', 'Arc_033', 'Arc_032', 'Arc_002', 'Aus_001', 'Arc_020', 'Arc_022', 'Arc_004', 'Eur_006'].
ProxyDatabase saved to: ./prev_data/filtered__ptype_res_cfr_pages2kv2.nc

This notebook is intentionally short, showing how easy it is to this with cfr. However, that particular database is frozen in time. To take into account more recent updates to the PAGES 2k paleotemperature database, notebooks C01_c and C01_d showcase how to use the LiPDverse and LiPD Graph, as well as associated Python tools, before feeding the data to cfr.

Contents