ISMIP6.org

Merging Observed & ISMIP6 Freshwater Anomalies

Table of contents

• Introduction
  • Summary
  • Issues
• Greenland observations
  • Mankoff 2021
  • Greenland observations: code
  • Greenland observations: table
  • Greenland observations: plot
• Greenland ISMIP
  • Load ISMIP SSP 245
  • Table
  • ISMIP annual contribution
• Greenland merge
  • Code
  • Table
  • Plot
• Antarctic observations
  • Data source
  • Load data
  • Antarctic obs table
  • Antarctic obs plot
  • Uncertainty
• Antarctic ISMIP
  • Load ISMIP SSP 245
  • Table
  • ISMIP annual contribution
• Antarctic merge
  • Code
  • Table
  • Plot
• Summary
  • Output
  • Questions & To Do
• Appendix
  • References
  • About This Document
• LaTeX Header
  • Beamer
  • References
  • Hyperref
  • Tweak References
  • Background Block
  • Code export
• Local Variables

Introduction

Summary

This workbook

• Computes freshwater anomaly from Greenland and Antarctica for the NASA GISS ModelE.
  • Anomaly is defined as mass loss from 1990 through 2100
• We use observed anomalies ~1990 through ~2020
  • Greenland: 1990 – 2020
  • Antarctica: 1994 through 2017; Data from 1990 – 1993 is extrapolated.
• We use ISMIP6 projections from 2020 through 2100

Issues

• {{{WARNING}}} Projections in Antarctica are only grounded ice, not ice shelf calving and melting, which is the majority of the observed mass loss.

Greenland observations

Mankoff 2021

citet:mankoff_2021 provide main ice sheet mass loss from 1840 through next week.

• Years with mass gain subtracted from following years $\to$ all mass change is $\le$ 0.
• hi and lo years are $\pm$ published uncertainty from Mankoff (2021)

{{{SKIPLINE}}}

• Paper: https://doi.org/10.5194/essd-13-5001-2021
• Data: https://doi.org/10.22008/FK2/OHI23Z v439

Greenland observations: code

import pandas as pd
obs_GL = pd.read_csv('./out/mass_change_no_gain_greenland.csv',
                 index_col=0,
                 parse_dates=True,
                 comment='#',
                 delim_whitespace=True,
                 names=['Gt/yr'])
obs_GL.index.name='Date'

Greenland observations: table

Head

Date	Gt/yr
1990-01-01 00:00:00	137.6
1991-01-01 00:00:00	76.7
1992-01-01 00:00:00	0
1993-01-01 00:00:00	3.6
1994-01-01 00:00:00	113.8
1995-01-01 00:00:00	211.9
1996-01-01 00:00:00	0
1997-01-01 00:00:00	0
1998-01-01 00:00:00	102
1999-01-01 00:00:00	47
2000-01-01 00:00:00	77.1
2001-01-01 00:00:00	26.1
2002-01-01 00:00:00	142.5
2003-01-01 00:00:00	167.2
2004-01-01 00:00:00	165.8

Tail

Date	Gt/yr
2005-01-01 00:00:00	168.4
2006-01-01 00:00:00	239.8
2007-01-01 00:00:00	257.3
2008-01-01 00:00:00	201.2
2009-01-01 00:00:00	243
2010-01-01 00:00:00	376.8
2011-01-01 00:00:00	336.2
2012-01-01 00:00:00	429.3
2013-01-01 00:00:00	107.9
2014-01-01 00:00:00	184.6
2015-01-01 00:00:00	213.9
2016-01-01 00:00:00	256
2017-01-01 00:00:00	102.6
2018-01-01 00:00:00	75.8
2019-01-01 00:00:00	426

Greenland observations: plot

Greenland ISMIP

Load ISMIP SSP 245

import xarray as xr
ds = xr.open_dataset('./ISMIP6/emulandiceAIS_GIS_ssp245_globalsl_GT.nc')
ismip_GL = ds['GIS_sea_level_change']\
    .to_dataframe()\
    .unstack().T

ismip_GL.index = [pd.to_datetime(_[1], format='%Y') for _ in ismip_GL.index]
ismip_GL.index.name = 'Date'

Table

Date	50	17	83
2020-01-01 00:00:00	1800	1440	2520
2030-01-01 00:00:00	3960	2880	5400
2040-01-01 00:00:00	6480	4680	9000
2050-01-01 00:00:00	10080	6840	13320
2060-01-01 00:00:00	12960	8640	18000
2070-01-01 00:00:00	16920	10800	23760
2080-01-01 00:00:00	20880	12240	30600
2090-01-01 00:00:00	25200	13320	37800
2100-01-01 00:00:00	27720	12600	45720

ISMIP annual contribution

Greenland merge

Code

obs_GL_17_50_83 = pd.concat([obs_GL]*3, axis='columns', ignore_index=True)
obs_GL_17_50_83 = obs_GL_17_50_83.rename(columns={0:17, 1:50, 2:83})
df_GL = pd.concat((obs_GL_17_50_83, ismip_GL_ann)).interpolate()

Table

Date	17	50	83
2010-01-01 00:00:00	376.8	376.8	376.8
2011-01-01 00:00:00	336.2	336.2	336.2
2012-01-01 00:00:00	429.3	429.3	429.3
2013-01-01 00:00:00	107.9	107.9	107.9
2014-01-01 00:00:00	184.6	184.6	184.6
2015-01-01 00:00:00	213.9	213.9	213.9
2016-01-01 00:00:00	256	256	256
2017-01-01 00:00:00	102.6	102.6	102.6
2018-01-01 00:00:00	75.8	75.8	75.8
2019-01-01 00:00:00	426	426	426
2020-12-31 00:00:00	285	321	357
2021-12-31 00:00:00	144	216	288
2022-12-31 00:00:00	144	216	288
2023-12-31 00:00:00	144	216	288
2024-12-31 00:00:00	144	216	288
2025-12-31 00:00:00	144	216	288
2026-12-31 00:00:00	144	216	288
2027-12-31 00:00:00	144	216	288
2028-12-31 00:00:00	144	216	288
2029-12-31 00:00:00	144	216	288

Plot

Antarctic observations

Data source

• From Slater (2021)

Load data

import pandas as pd

S2021_kw = {'parse_dates':True, 'index_col':0}
shelf_calving = pd.read_csv('~/data/Slater_2021/iceshelves_calving_cumul_1994_2017_annual.csv', **S2021_kw)\
                  .rename(columns={"Cumulative mass change (Gt)":"Ice Shelf Calving"})
shelf_thinning = pd.read_csv('~/data/Slater_2021/iceshelves_thinning_cumul_1994_2017_annual.csv', **S2021_kw)\
                   .rename(columns={"Cumulative mass change (Gt)":"Ice Shelf Thinning"})
landice_AQ = pd.read_csv('~/data/Slater_2021/AIS_cumul_1994_2017_annual.csv', **S2021_kw)\
               .rename(columns={"Cumulative mass change (Gt)":"Antarctica"})

S2021_kw = {'left_index':True, 'right_index':True, 'how':'outer'}
obs_AQ = shelf_calving.merge(shelf_thinning, **S2021_kw)\
                      .merge(landice_AQ, **S2021_kw)
obs_AQ.index.name = 'Date'

obs_AQ = -1 * obs_AQ.diff()

obs_AQ['Total'] = obs_AQ.sum(axis='columns')

Antarctic obs table

\fontsize{4pt}{4}\selectfont

obs_AQ

Date	Ice Shelf Calving	Ice Shelf Thinning	Antarctica	Total
1994-01-01 00:00:00	nan	nan	nan	0
1995-01-01 00:00:00	71.375	1.6598	37.17	110.205
1996-01-01 00:00:00	71.4792	-17.3615	28.78	82.8977
1997-01-01 00:00:00	126.945	-4.85001	32.31	154.405
1998-01-01 00:00:00	126.946	38.3959	22.43	187.771
1999-01-01 00:00:00	336.403	84.7815	78.73	499.915
2000-01-01 00:00:00	307.921	91.948	28.27	428.139
2001-01-01 00:00:00	332.46	76.967	20.98	430.407
2002-01-01 00:00:00	139.388	108.552	28.99	276.93
2003-01-01 00:00:00	139.388	189.182	96.01	424.58
2004-01-01 00:00:00	139.388	269.945	78.91	488.243
2005-01-01 00:00:00	139.387	297.37	16.89	453.647
2006-01-01 00:00:00	139.388	312.36	95.35	547.098
2007-01-01 00:00:00	139.388	332.43	207.72	679.538
2008-01-01 00:00:00	139.388	347.33	113.88	600.598
2009-01-01 00:00:00	139.388	306.2	89.15	534.738
2010-01-01 00:00:00	139.388	216.62	229.38	585.388
2011-01-01 00:00:00	139.388	129.18	147.99	416.558
2012-01-01 00:00:00	139.388	95.05	209.4	443.838
2013-01-01 00:00:00	139.387	90.42	269.75	499.557
2014-01-01 00:00:00	139.388	55.86	261.69	456.938
2015-01-01 00:00:00	139.387	-12.93	254.84	381.297
2016-01-01 00:00:00	139.388	-49.23	67.02	157.178

Antarctic obs plot

Uncertainty

From Slater (2021) Table 1

Component	Uncertainty [Gt yr-1]
Ice shelf calving	36
Ice shelf thinning	39
Antarctic land ice	24
mean	33
sum	99

• Assume these are 1 σ uncertainty
• lo/mid/hi estimates can use mean (33), max (39) or sum (99)

If fields are random

from uncertainties import unumpy
import numpy as np
np.sum(unumpy.uarray([1,1,1], [36,39,24]))

3.0+/-58.249463516842795

Antarctic ISMIP

Load ISMIP SSP 245

import xarray as xr
ds = xr.open_dataset('./ISMIP6/emulandiceAIS_GIS_ssp245_globalsl_GT.nc')
ismip_AQ = ds['AIS_sea_level_change']\
    .to_dataframe()\
    .unstack().T

ismip_AQ.index = [pd.to_datetime(_[1], format='%Y') for _ in ismip_AQ.index]
ismip_AQ.index.name = 'Date'

Table

Date	50	17	83
2020-01-01 00:00:00	2160	1440	3240
2030-01-01 00:00:00	3960	2520	5400
2040-01-01 00:00:00	6120	3600	8640
2050-01-01 00:00:00	8640	5040	12600
2060-01-01 00:00:00	11520	6120	16920
2070-01-01 00:00:00	14580	6120	22680
2080-01-01 00:00:00	18000	6480	29221
2090-01-01 00:00:00	23760	8640	38581
2100-01-01 00:00:00	28800	9720	50040

ISMIP annual contribution

Antarctic merge

Code

obs_AQ_17_50_83 = pd.concat([obs_AQ['Total']]*3, axis='columns', ignore_index=True)
obs_AQ_17_50_83 = obs_AQ_17_50_83.rename(columns={0:17, 1:50, 2:83})
df_AQ = pd.concat((obs_AQ_17_50_83, ismip_AQ_ann)).interpolate()

Table

\fontsize{5pt}{5}\selectfont

Date	17	50	83
2009-01-01 00:00:00	534.738	534.738	534.738
2010-01-01 00:00:00	585.388	585.388	585.388
2011-01-01 00:00:00	416.558	416.558	416.558
2012-01-01 00:00:00	443.838	443.838	443.838
2013-01-01 00:00:00	499.557	499.557	499.557
2014-01-01 00:00:00	456.938	456.938	456.938
2015-01-01 00:00:00	381.297	381.297	381.297
2016-01-01 00:00:00	157.178	157.178	157.178
2020-12-31 00:00:00	132.589	168.589	186.589
2021-12-31 00:00:00	108	180	216
2022-12-31 00:00:00	108	180	216
2023-12-31 00:00:00	108	180	216
2024-12-31 00:00:00	108	180	216
2025-12-31 00:00:00	108	180	216
2026-12-31 00:00:00	108	180	216
2027-12-31 00:00:00	108	180	216
2028-12-31 00:00:00	108	180	216
2029-12-31 00:00:00	108	180	216
2030-12-31 00:00:00	108	180	216
2031-12-31 00:00:00	108	216	324

Plot

Summary

Output

def writer(da, fname):
    dd = da.copy(deep=True)
    dd.index = dd.index.year
    dd.to_csv(fname, sep='\t', header=False)

writer(df_AQ[17], './out/AQ_17.csv')
writer(df_AQ[50], './out/AQ_50.csv')
writer(df_AQ[83], './out/AQ_83.csv')
writer(df_GL[17], './out/GL_17.csv')
writer(df_GL[50], './out/GL_50.csv')
writer(df_GL[83], './out/GL_83.csv')

Questions & To Do

• [ ] …

Appendix

References

\printbibliography[heading=none]

About This Document

This document is an Emacs Org Mode plain-text file with code and text embedded. If you are viewing:

• A PDF, HTML, or DOC file, then it was generated by exporting from Org. Not all of the Org parts (code, results, comments, etc.) were exported. The Org source file is available upon request, and may be embedded in the PDF. You can access files embedded in PDF files with from within your PDF viewer.
• A file with an org extension in something other than Emacs, then you are seeing the canonical version and the full source, but without any syntax highlighting, document structure, or the ability to execute the code blocks.
• An Org file within Emacs, then this is the canonical version. You should be able to fully interact and reproduce the contents of this document, although it may require 3rd-party applications (Python, etc.) and a similar Emacs configuration. This is available upon request.

LaTeX Header

Beamer

Title Page

Theme

Other

References

Hyperref

Tweak References

Background Block

Code export

Local Variables