https://github.com/mankoff/sankey
Sankey diagram for ice sheet mass flow
https://github.com/mankoff/sankey
antarctica greenland mass sankey tikz
Last synced: 2 months ago
JSON representation
Sankey diagram for ice sheet mass flow
- Host: GitHub
- URL: https://github.com/mankoff/sankey
- Owner: mankoff
- Created: 2024-04-10T04:30:28.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2024-06-06T12:54:35.000Z (11 months ago)
- Last Synced: 2024-06-11T17:34:22.756Z (11 months ago)
- Topics: antarctica, greenland, mass, sankey, tikz
- Language: TeX
- Homepage:
- Size: 40.4 MB
- Stars: 3
- Watchers: 2
- Forks: 1
- Open Issues: 13
-
Metadata Files:
- Readme: README.org
Awesome Lists containing this project
README
# #+PROPERTY: header-args:bash+ :session *sankey-shell*
# #+PROPERTY: header-args:jupyter-python+ :dir (file-name-directory buffer-file-name)
* On Ice Sheet Mass Flow
* Table of contents :toc_3:noexport:
- [[#on-ice-sheet-mass-flow][On Ice Sheet Mass Flow]]
- [[#introduction][Introduction]]
- [[#abstract--methods][Abstract & Methods]]
- [[#results][Results]]
- [[#processes-and-flow][Processes and flow]]
- [[#implementation][Implementation]]
- [[#greenland][Greenland]]
- [[#reset][Reset]]
- [[#discharge][Discharge]]
- [[#basal-melt][Basal melt]]
- [[#gz-retreat][GZ retreat]]
- [[#smb][SMB]]
- [[#shelf-melt-and-freezing][Shelf melt and freezing]]
- [[#mb][MB]]
- [[#antarctica][Antarctica]]
- [[#export-to-csvs][Export to CSVs]]
- [[#grounded-vs-marine-mass-loss][Grounded vs Marine mass loss]]
- [[#reset-1][Reset]]
- [[#masks-east-west-peninsula-islands-grounded-and-shelves][Masks: East, West, Peninsula, Islands, Grounded and Shelves]]
- [[#smb-mar][SMB (MAR)]]
- [[#basal-melt-1][Basal melt]]
- [[#discharge-1][Discharge]]
- [[#antarctic-ice-shelves][Antarctic Ice shelves]]
- [[#grace][GRACE]]
- [[#misc][Misc]]
- [[#export-tables-to-csvs][Export tables to CSVs]]
- [[#convert-pdfs-to-png][Convert PDFs to PNG]]
* Introduction
Sankey diagrams for mass flow in Greenland and Antarctica.
* Abstract & Methods
See [[./ms.tex]]
* Results
#+BEGIN_QUOTE
[!NOTE]
All figures are width-proportional within and between each other.
#+END_QUOTE
#+CALL: pdfs2png()
#+ATTR_ORG: :width 800px
[[./fig_aq_gl.png]]
#+ATTR_ORG: :width 800px
[[./fig_aq_parts.png]]
* Processes and flow
#+BEGIN_SRC dot :file flowchart.png :exports results
digraph G {
cd[label="Condensation"]
dp[label="Deposition"]
rf[label="Rainfall"]
sf[label="Snowfall"]
smbin[label="SMB\ninput"]
frontadv[label="Frontal\nadvance"]
shelffreeze[label="Sub-shelf\nfreeze-on"]
IO[label = ""]
smbout[label="SMB\noutput"]
su[label="Sublimation"]
ev[label="Evaporation"]
ru[label="Runoff"]
dyn[label="Dynamics"]
# submelt[label="Submarine\nmelt"]
discharge[label="Discharge"]
calvGL[label="Calving"]
frontmeltGL[label="Frontal\nmelt"]
shelfmeltGL[label="Sub-shelf\nmelt (GL)"]
calvAQ[label="Calving"]
# frontmeltAQ[label="Frontal\nmelt"]
shelfmeltAQ[label="Sub-shelf\nmelt (AQ)"]
frontret[label="Frontal\nretreat"]
gzret[label="Grounding\nline retreat", style="dashed"]
bmb[label="Grounded\nbasal mass\nbalance"]
subgraph cluster_GL{
rank="same"
label = "Greenland"
labelloc = "b"
calvGL
frontmeltGL
}
discharge -> calvGL
discharge -> frontmeltGL
frontmeltGL -> shelfmeltGL [style="dashed"]
subgraph cluster_AQ{
rank="same"
label = "Antarctica"
labelloc = "b"
calvAQ
shelfmeltAQ
}
discharge -> calvAQ
discharge -> shelfmeltAQ
cd -> smbin
dp -> smbin
rf -> smbin
sf -> smbin
smbin -> IO
frontadv -> IO
shelffreeze -> IO
IO -> smbout # [label="su + ev + ru"]
smbout -> su
smbout -> ev
smbout -> ru
IO -> dyn # [label="smb_in - smb_out"]
dyn -> discharge
dyn -> frontret
dyn -> gzret
dyn -> bmb
# ml[label="Mass\nloss", penwidth=3, color=red]
# ml -> Output
}
#+END_SRC
#+RESULTS:
[[file:flowchart.png]]
* Implementation
** Greenland
#+NAME: gl_baseline
| ID | Term | Value | Unc | IO | Period | Source |
|-------------+------------------------+----------+-----+-----+-----------+------------------------------------------------|
| RF | Rainfall | 40 | 15 | I_g | 2010-2019 | fettweis_2020 |
| CD | Condensation | 5 | 15 | I_g | 2010-2019 | fettweis_2020 |
| DP | Deposition | 10 | 15 | I_g | 2010-2019 | fettweis_2020 |
| SF | Snowfall | 680 | 15 | I_g | 2010-2019 | fettweis_2020 |
| EV | Evaporation | 5 | 15 | O_g | 2010-2019 | fettweis_2020 |
| RU | Runoff | 435 | 15 | O_g | 2010-2019 | fettweis_2020 |
| SU | Sublimation | 60 | 15 | O_g | 2010-2019 | fettweis_2020 |
| RFZ | Refreezing | 200 | 15 | | 2010-2019 | fettweis_2020 |
| BMB | Grounded basal melting | 20 | 20 | O_g | steady | karlsson_2021 |
| DISCH | Discharge | 485-20+5 | 10 | | 2010-2019 | mankoff_2020_solid,kochtitzky_2023,bollen_2023 |
| CALV | Calving | 235 | 30 | O_o | | rignot_2010 |
| FRONTMELT | Frontal melting | 235 | 30 | O_o | | rignot_2010 |
| SHELFMELT | Sub-shelf melting | 25 | 40 | O_o | 2013-2022 | wang_2024 |
| SHELFFREEZE | Sub-shelf freeze-on | 5 | 40 | I_o | 2013-2022 | wang_2024 |
| GZRET | Grounding line retreat | 5 | ? | O_g | | |
| FRONTRET | Frontal retreat | 50 | 4 | O_o | 2010-2020 | kochtitzky_2023 |
| FRONTADV | Frontal advance | 0 | | I_o | 2010-2020 | kochtitzky_2023 |
#+CALL: orgtbl2csv(tbl="gl_baseline")
#+RESULTS:
: Exported gl_baseline to ./dat/gl_baseline.csv
*** Reset
#+BEGIN_SRC bash :exports both :results verbatim
trash G_GL
[[ -e ./G_GL ]] || grass -e -c EPSG:3413 ./G_GL
#+END_SRC
*** Discharge
**** From total mass balance product
#+BEGIN_SRC jupyter-python :exports both
!date
fname="https://thredds.geus.dk/thredds/fileServer/MassBalance/MB_SMB_D_BMB.csv"
df = pd.read_csv(fname, index_col=0, parse_dates=True)[['D','D_err']]
df = df\
.resample('1D').interpolate().resample('YS').sum()
print("2010s")
print(df['2010-01-01':'2019-12-31'].mean())
#+END_SRC
#+RESULTS:
: Wed Dec 18 07:11:05 AM EST 2024
: 2010s
: D 485.373414
: D_err 44.951388
: dtype: float64
Then, subtract 15 from 475 based on citet:kochtitzky_2023 who report, in Section 3.3, 17 +- 6.8 and 14.5 +- 5.8 but that "[b]ecause our fluxgates were typically located tens to hundreds of meters lower than those in the similar studies (King et al., 2018; Mankoff et al., 2020), the melt correction for these studies would be higher than values presented herein, although it is beyond the scope of the current study to determine what those values would be."
**** Peripheral discharge (Bollen 2023)
***** Where are these glaciers
#+BEGIN_SRC bash :exports both :results verbatim
grass ./G_GL/PERMANENT
g.mapset -c Bollen_2023
cat "${DATADIR}/Bollen_2023/GreenlandGIC_discharge_timeseries - Ellyn Enderlin.csv" \
| cut -d, -f1-3 \
| v.in.ascii input=- output=bollen_2023 separator=, skip=1 x=2 y=3 z=1
#+END_SRC
***** How much do they contribute?
#+BEGIN_SRC jupyter-python :exports both
import pandas as pd
data_root='/home/kdm/data'
path='Bollen_2023'
fname='GreenlandGIC_discharge_timeseries - Ellyn Enderlin.csv'
df = pd.read_csv(f"{data_root}/{path}/{fname}", index_col=0, header=[0])
df = df.sum(axis='rows')
df = df / 1E9 # per email from Ellyn, units are m^3/year. Convert to Gt.
df = df['2010':'2018']
df.mean()
#+END_SRC
#+RESULTS:
: 5.209345977852399
*** Basal melt
+ 21 Gt/yr from Karlsson (2021) http://doi.org/10.1038/s41467-021-23739-z
+ Assume steady state
*** TODO GZ retreat
From Millan (2022) http://doi.org/10.5194/tc-16-3021-2022
+ Gz retreat is ~0.13 km/yr (Fig. 3a)
+ Ice velocity is ~1200 m/yr (Fig. 3b) (not needed)
+ 20 km wide
Rates are higher per Ciraci (2023) http://doi.org/10.1073/pnas.2220924120, but
+ Ice surface close to flotation near GZ, and shelf is ~500 m thick, so estimate 600 m ice.
Therefore, gz retreat in Gt/year is width * thick * retreat rate * density
#+BEGIN_SRC bash :exports both :results verbatim
frink "0.13 km/yr * 20 km * 600 m * 917 kg/m^3 -> Gt/yr"
#+END_SRC
#+RESULTS:
: 1.43052
Assume similar from other ice shelves too, for a total of ~5 Gt/yr GZ retreat in Greenland.
*** SMB
#+BEGIN_SRC bash :exports both :results verbatim
g.mapset -c MAR
ncdump -v TIME dat/MARv3.12-GRD-15km-annual.nc4 # 30-39 = 2010-2019
ncra --overwrite -d TIME,30,39 dat/MARv3.12-GRD-15km-annual.nc4 tmp/MAR_GL.nc
ncdump -v X10_110 tmp/MAR_GL.nc # 101
ncdump -v Y20_200 tmp/MAR_GL.nc # 181
g.region w=$(( -645000 - 7500 )) e=$(( 855000 + 7500 )) s=$(( -3357928 - 7500 )) n=$((-657928 + 7500 )) res=15000 -p
var=SF # debug
for var in SF RF RU SU ME SMB EVA CON DEP SUB MSK AREA; do
r.in.gdal -o input=NetCDF:tmp/MAR_GL.nc:${var} output=${var}
r.region -c map=${var}
done
r.mapcalc "GL_ice_all = (MSK > 50) & ((x()-y()) > 520000)" # Limit to ice and remove Canada
# r.clump input=GL_ice output=clumps --o
# main_clump=$(r.stats -c -n clumps sort=desc | head -n2 | tail -n1 | cut -d" " -f1)
# r.mapcalc "GL_ice = if(clumps == ${main_clump}, 1, null())"
# r.mask raster=GL_ice --o
r.mapcalc "MASK = if(GL_ice_all == 1)" --o
# if only X % of a cell is ice, scale by that.
r.mapcalc "scale_mask = (GL_ice_all * MSK) / 100"
# scale
## units are mm.w.eq. per grid cell. Grid cell areas are in km^2
## + mm.w.eq. -> m w.eq.: /1E3
## + m w.eq -> kg: *1E3
## + area in km^2 -> m^2: *1E3*1E3
## + kg -> Gt: /1E12
# ds = ds/1E3 * 1E3 * ds['AREA']*1E3*1E3 / 1E12
for var in SF RF RU SU ME SMB EVA CON DEP SUB; do
r.mapcalc "${var} = (${var}/1000) * 1000 * (AREA * 1000*1000) * scale_mask / exp(10,12)"
done
r.mask -r
r.mapcalc "RFZ = ME + RF - RU"
#+END_SRC
#+BEGIN_SRC bash :exports both :results verbatim :session "*projects/sankey-shell*"
for var in SF RF RU ME SMB EVA CON DEP SUB RFZ; do
echo ${var} $(r.univar -g ${var} | grep sum)
done
#+END_SRC
#+RESULTS:
#+begin_example
[?2004lSF sum=678.472341306034
RF sum=41.0073369748482
RU sum=433.411271134275
ME sum=594.819117205514
SMB sum=232.245706856329
EVA sum=7.43645901936729
CON sum=2.02922271273767
DEP sum=12.3770587084991
SUB sum=60.0712550947222
RFZ sum=202.41518304609
#+end_example
*** Shelf melt and freezing
#+BEGIN_SRC bash :exports both :results verbatim
grass ./G_GL/PERMANENT
g.mapset -c Wang_2024
tif_list=$(find ~/data/Wang_2024 -name "????.tif")
t=$(echo $tif_list | tr ' ' '\n' | head -n1) # debug
for t in ${tif_list}; do
dirname=$(basename $(dirname ${t}))
fname=$(basename ${t})
fname=${fname%.*}
tname=g_${dirname}_${fname} # add g_ because "79N" is not a valid name
r.in.gdal input=${t} output=${tname}
done
g.region raster=$(g.list type=raster sep=,) -pa
r.series input=$(g.list type=raster sep=,) output=melt method='average'
r.colors -a map=melt color=viridis
r.mapcalc "area = area()"
## Melt data is m/year
## Multiply by area to get m/m^2 or grams, then 1000 to get kg
r.mapcalc "melt = melt * 1000 * area / exp(10,12)" --o
r.mapcalc "melt_on = if(melt > 0, melt, null())"
r.mapcalc "freeze_on = if(melt < 0, melt, null())"
#+END_SRC
**** Stats
#+BEGIN_SRC bash :exports both :results verbatim :session *projects/sankey-shell*
echo "NET"
r.univar -gt map=melt | cut -d"|" -f11
echo ""
echo "FREEZE_ON"
r.univar -gt map=freeze_on | cut -d"|" -f11
echo ""
echo "MELT_OFF"
r.univar -gt map=melt_on | cut -d"|" -f11
#+END_SRC
#+RESULTS:
#+begin_example
[?2004lNET
[?2004lsum
33.4127947245078
[?2004l
[?2004lFREEZE_ON
?2004lsum
-2.68199438110646
[?2004l
[?2004lMELT_OFF
[?2004lsum
36.094789105614
#+end_example
*** MB
**** GRACE ESA
+ https://data1.geo.tu-dresden.de/gis_gmb/
#+begin_src jupyter-python :exports both
import xarray as xr
ds = xr.open_dataset("~/data/Dohne_2023/GIS_GMB_grid.nc")
ds['dm'] = ds['dm'] * ds['area']
ds = ds.sel({'time':slice('2010-01-01','2019-12-31')})
ds = data=ds['dm'].to_dataset()
ds = ds['dm'].sum(dim=['x','y'])/1E12
ds = ds - ds.values[0]
_ = ds.plot()
ds = ds.resample({'time':'YS'}).mean()
ds = ds.diff(dim='time')
print(ds.mean())
#+end_src
#+RESULTS:
:RESULTS:
: Size: 8B
: array(-250.12027707)
[[file:./figs_tmp/911c045e76e18fc0fb23bf799dc621683309edb9.png]]
:END:
Results processed by Thorben Döhne are: -226 +- 14.5 Gt/yr
**** GRACE JPL
#+BEGIN_SRC jupyter-python :exports both
import numpy as np
import pandas as pd
from datetime import datetime, timedelta
from uncertainties import unumpy
df = pd.read_csv("~/data/GRACE/greenland_mass_200204_202410.txt",
comment="H", parse_dates=True, sep="\\s+", header=None,
names=['year','mass','err'])
# Function to convert year.frac to ISO format (YYYY-MM-DD)
def year_frac_to_iso(year_frac):
year = int(year_frac)
frac = year_frac - year
start_of_year = datetime(year, 1, 1)
days_in_year = (datetime(year + 1, 1, 1) - start_of_year).days
date = start_of_year + timedelta(days=frac * days_in_year)
return pd.to_datetime(date.strftime('%Y-%m-%d'))
# Apply the conversion to the 'Year' column
df['date'] = df['year'].apply(year_frac_to_iso)
df = df.drop(columns=['year'])
df = df.set_index('date')
df = df[['mass','err']]
# df.resample('D').mean().interpolate()
df = df['2010-01-01':'2019-12-31']
df['mass'] = df['mass'] - df['mass'].max()
# arr = unumpy.uarray(df['mass'].values, df['err'].values)
_ = df['mass'].plot() # <-- traditional plot
# df.resample('YS').mean().diff().plot()
print(df['mass'].resample('YS').mean().diff().mean(), '+-', df['err'].resample('YS').mean().mean())
#+END_SRC
#+RESULTS:
:RESULTS:
: -265.0541666666667 +- 23.828999999999997
[[file:./figs_tmp/64b7a155973c517d77c396f87b15a6a4f5d91932.png]]
:END:
**** Mankoff 2021
#+BEGIN_SRC jupyter-python :exports both
!date
fname="https://thredds.geus.dk/thredds/fileServer/MassBalance/MB_SMB_D_BMB.csv"
df = pd.read_csv(fname, index_col=0, parse_dates=True)[['MB','MB_err']]
df = df\
.resample('1D').interpolate().resample('YS').sum()
print("2010s")
print(df['2010-01-01':'2019-12-31'].mean())
#+END_SRC
#+RESULTS:
: Wed Jan 22 02:16:48 PM PST 2025
: 2010s
: MB -246.172157
: MB_err 94.196209
: dtype: float64
** Antarctica
#+NAME: aq
| ID | Term | East_g | West_g | Peninsula_g | East_s | West_s | Peninsula_s | Unc | IO | Period | Source |
|-------------+------------------------+--------+--------+-------------+--------+--------+-------------+---------+----+-----------+----------------------------------------------------------------|
| RF | Rainfall | 1 | 1 | 2 | 1 | 1 | 2 | 15 | I | 2010-2019 | fettweis_2020 |
| CD | Condensation | 1 | 1 | 1 | 1 | 1 | 1 | 15 | I | 2010-2019 | fettweis_2020 |
| DP | Deposition | 37 | 24 | 6 | 6 | 6 | 2 | 15 | I | 2010-2019 | fettweis_2020 |
| SF | Snowfall | 1392 | 724 | 282 | 172 | 180 | 57 | 15 | I | 2010-2019 | fettweis_2020 |
| RFZ | Refreezing | 15 | 5 | 19 | 26 | 10 | 32 | 15 | | 2010-2019 | fettweis_2020 |
| EV | Evaporation | 1 | 1 | 1 | 1 | 1 | 1 | 15 | O | 2010-2019 | fettweis_2020 |
| RU | Runoff | 1 | 1 | 2 | 2 | 1 | 4 | 15 | O | 2010-2019 | fettweis_2020 |
| SU | Sublimation | 151 | 33 | 13 | 23 | 9 | 4 | 15 | O | 2010-2019 | fettweis_2020 |
| BMB | Grounded basal melting | 47 | 19 | 3 | 0 | 0 | 0 | 30 | O | | van-liefferinge_2013 |
| DISCH | Discharge | 1147 | 902 | 292 | 0 | 0 | 0 | 5 -- 50 | | 2008-2019 | davison_2023 (to shelves) + rignot_2019 (grounded + islands) |
| CALV | Calving | 223 | 46 | 139 | 694 | 567 | 104 | 5 | O | 2010-2019 | greene_2022 + rignot_2019 discharge (grounded + islands) |
| FRONTMELT | Frontal melting | 0 | 0 | 0 | 0 | 0 | 0 | | O | | |
| SHELFMELT | Sub-shelf melting | 0 | 0 | 0 | 527 | 684 | 164 | 150 | O | 2010-2017 | paolo_2023 |
| SHELFFREEZE | Sub-shelf freeze-on | 0 | 0 | 0 | 208 | 147 | 11 | 300 | I | 2010-2017 | paolo_2023 |
| GZRET | Grounding line retreat | 1 | 45 | 1 | 0 | 0 | 0 | 15 | O | 1997-2021 | davison_2023 (only Pine Island, Thwaites, Crosson, and Dotson) |
| FRONTRET | Frontal retreat | 0 | 0 | 0 | 69 | 206 | 125 | 5 | O | 2010-2021 | greene_2022 |
| FRONTADV | Frontal advance | 0 | 0 | 0 | 192 | 2 | 1 | 5 | I | 2010-2021 | greene_2022 |
*** Export to CSVs
Split AQ table above to east,west,peninsula,all CSVs, combining shelf and grounded
#+BEGIN_SRC jupyter-python :exports both :var aq=aq :colnames no
import numpy as np
import pandas as pd
aq = np.array(aq)
df = pd.DataFrame(aq[1:,1:], index=aq[1:,0], columns=aq[0,1:])
df.index.name = 'ID'
cols = ['East_g','East_s','West_g','West_s','Peninsula_g','Peninsula_s']
df[cols] = df[cols].astype(int)
df['All'] = df[cols].sum(axis='columns')
df['E'] = df[['East_g','East_s']].sum(axis='columns')
df['W'] = df[['West_g','West_s']].sum(axis='columns')
df['P'] = df[['Peninsula_g','Peninsula_s']].sum(axis='columns')
df = df.drop(columns=['IO', 'Period', 'Source'])
df = df.drop(columns=cols)
def custom_round(x, base=5):
if (x > 0) and (x < base): x = base
return int(base * round(float(x)/base))
cols = ['All','E','W','P']
for c in cols: df[c] = df[c].apply(lambda x: custom_round(x, base=5))
for c in cols:
df[['Term',c]].rename(columns={c:'Value'}).to_csv('./dat/aq_' + c + '.csv')
df
#+END_SRC
#+RESULTS:
| ID | Term | Unc | All | E | W | P |
|-------------+------------------------+---------+-------+------+-----+-----|
| RF | Rainfall | 15 | 10 | 5 | 5 | 5 |
| CD | Condensation | 15 | 5 | 5 | 5 | 5 |
| DP | Deposition | 15 | 80 | 45 | 30 | 10 |
| SF | Snowfall | 15 | 2805 | 1565 | 905 | 340 |
| RFZ | Refreezing | 15 | 105 | 40 | 15 | 50 |
| EV | Evaporation | 15 | 5 | 5 | 5 | 5 |
| RU | Runoff | 15 | 10 | 5 | 5 | 5 |
| SU | Sublimation | 15 | 235 | 175 | 40 | 15 |
| BMB | Grounded basal melting | 30 | 70 | 45 | 20 | 5 |
| DISCH | Discharge | 5 -- 50 | 2340 | 1145 | 900 | 290 |
| CALV | Calving | 5 | 1775 | 915 | 615 | 245 |
| FRONTMELT | Frontal melting | | 0 | 0 | 0 | 0 |
| SHELFMELT | Sub-shelf melting | 150 | 1375 | 525 | 685 | 165 |
| SHELFFREEZE | Sub-shelf freeze-on | 300 | 365 | 210 | 145 | 10 |
| GZRET | Grounding line retreat | 15 | 45 | 5 | 45 | 5 |
| FRONTRET | Frontal retreat | 5 | 400 | 70 | 205 | 125 |
| FRONTADV | Frontal advance | 5 | 195 | 190 | 5 | 5 |
*** Grounded vs Marine mass loss
#+begin_src jupyter-python :exports both :var aq=aq :colnames no
import numpy as np
import pandas as pd
aq = np.array(aq)
df = pd.DataFrame(aq[1:,1:], index=aq[1:,0], columns=aq[0,1:])
df.index.name = 'ID'
df = df.drop(columns=['Source', 'Period', 'Unc'])
df = df.drop(['RFZ'])
cols = ['East_g','East_s','West_g','West_s','Peninsula_g','Peninsula_s']
df[cols] = df[cols].astype(int)
for roi in ['East','West','Peninsula']:
df.loc['DISCH',roi+'_g'] = df.loc['DISCH',roi+'_g'] - df.loc['CALV',roi+'_g']
# df.loc['CALV', 'West_s'] = df.loc['CALV', 'West_s'] + df.loc['CALV', 'West_g']; df.loc['CALV', 'West_g'] = 0
# df.loc['CALV', 'East_s'] = df.loc['CALV', 'East_s'] + df.loc['CALV', 'East_g']; df.loc['CALV', 'East_g'] = 0
# df.loc['CALV', 'Peninsula_s'] = df.loc['CALV', 'Peninsula_s'] + df.loc['CALV', 'Peninsula_g']; df.loc['CALV', 'Peninsula_g'] = 0
# # df.loc['CALV', 'West_s'] = df.loc['CALV', 'West_s'] + df.loc['CALV', 'West_g'];
# df.loc['CALV', 'West_g'] = 0
# # df.loc['CALV', 'East_s'] = df.loc['CALV', 'East_s'] + df.loc['CALV', 'East_g'];
# df.loc['CALV', 'East_g'] = 0
# # df.loc['CALV', 'Peninsula_s'] = df.loc['CALV', 'Peninsula_s'] + df.loc['CALV', 'Peninsula_g'];
# df.loc['CALV', 'Peninsula_g'] = 0
# disch = df.loc['DISCH']['All_g'] - df.loc['CALV']['All_g']
df['All_g'] = df[['East_g','West_g','Peninsula_g']].sum(axis='columns')
df['All_s'] = df[['East_s','West_s','Peninsula_s']].sum(axis='columns')
df['All'] = df['All_g'] + df['All_s']
df['East'] = df['East_g'] + df['East_s']
df['West'] = df['West_g'] + df['West_s']
df['Peninsula'] = df['Peninsula_g'] + df['Peninsula_s']
def custom_round(x, base=5):
if (x > 0) and (x < base): x = base
return int(base * round(float(x)/base))
cols = ['All', 'All_g', 'East', 'East_g', 'West', 'West_g', 'Peninsula', 'Peninsula_g']
# df.loc['Net'] = df[cols]
da = df[df['IO'] == 'I'][cols].sum() - df[df['IO'] == 'O'][cols].sum()
for i in da.index:
if i[-1] == 'g': da[i] = da[i] - (df.loc['DISCH',i] + df.loc['DISCH',i[:-1] + 's'])
# if i[-1] == 's': da[i] = da[i] + df.loc['DISCH',i[:-1]+'g']
for i in ['All','East','West','Peninsula']:
da[i + '_s'] = da[i] - da[i + '_g']
da = da.sort_index()
da = da.apply(lambda x: custom_round(x, base=5))
df = pd.DataFrame(index = ['Antarctica', 'East', 'West', 'Peninsula'],
columns = ['Grounded', 'Marine', 'Total'])
df.loc['Antarctica'] = da[['All_g','All_s','All']].values
df.loc['East'] = da[['East_g','East_s','East']].values
df.loc['West'] = da[['West_g','West_s','West']].values
df.loc['Peninsula'] = da[['Peninsula_g','Peninsula_s','Peninsula']].values
df
#+end_src
#+RESULTS:
| | Grounded | Marine | Total |
|------------+------------+----------+---------|
| Antarctica | -190 | -260 | -450 |
| East | 85 | 190 | 270 |
| West | -250 | -275 | -525 |
| Peninsula | -20 | -175 | -195 |
*** Reset
#+BEGIN_SRC bash :exports both :results verbatim
trash G_AQ
[[ -e ./G_AQ ]] || grass -e -c EPSG:3031 ./G_AQ
#+END_SRC
*** Masks: East, West, Peninsula, Islands, Grounded and Shelves
#+BEGIN_SRC bash :exports both :results verbatim
grass ./G_AQ/PERMANENT
v.in.ogr input=${DATADIR}/NSIDC/NSIDC-0709.002/1992.02.07/IceBoundaries_Antarctica_v02.shp output=basins
g.region vector=basins res=10000 -pas
v.db.select map=basins|head
v.db.select -c map=basins columns=Regions | sort | uniq # East West Peninsula Islands
v.db.select -c map=basins columns=TYPE | sort | uniq # FL GR IS (float, ground, island)
v.to.rast input=basins output=east use=val val=1 where='(Regions == "East")'
v.to.rast input=basins output=west use=val val=2 where='(Regions == "West")'
v.to.rast input=basins output=peninsula use=val val=3 where='(Regions == "Peninsula")'
v.to.rast input=basins output=islands use=val val=4 where='(Regions == "Islands")'
r.patch input=east,west,peninsula,islands output=basins
r.category basins separator=":" rules=- << EOF
1:East
2:West
3:Peninsula
4:Islands
EOF
r.colors map=basins color=viridis
v.to.rast input=basins output=ground use=val val=1 where='(TYPE == "GR") or (TYPE == "IS")'
v.to.rast input=basins output=ground_noisland use=val val=1 where='(TYPE == "GR")'
#+END_SRC
**** Label islands to nearest region (east,west,peninsula)
Rignot 2019 provides discharge for Islands, but not by region. Here, determine island *area* per region, and percent of islands within each region. Then, for other values that are reported for all islands, split by area percent. This assumes all islands have the same flux (volume flow rate per unit area) for whatever property is divided up using this method.
#+begin_src bash :exports both :results verbatim
r.patch input=east,west,peninsula output=main_ice
r.colors map=main_ice color=viridis
r.grow.distance input=main_ice value=main_ice_grow
r.mapcalc "islands_near = int(if(islands, main_ice_grow))"
#+end_src
**** Find area of islands within each region
#+begin_src bash :exports both :results verbatim :session "*projects/sankey-shell*"
r.stats --q -A -r -c -N input=islands_near
#+end_src
#+RESULTS:
: 1 417
: 2 803
: 3 174
: [Raster MASK present]
Total Cells = 417 + 803 + 174 = 1394
East = 417 / 1394 % = 29.9139167862 ~= 30
West = 803 / 1394 % = 57.6040172166 ~= 60
Peninsula = 174 / 1394 % = 12.4820659971 ~= 10
**** Make masks for all grounded (including islands) or only shelf (excluding island)
#+BEGIN_SRC bash :exports both :results verbatim
r.mask --o raster=ground
r.patch input=islands_near,main_ice output=grounded_with_islands
r.mask --o -i raster=ground
r.mapcalc "shelf_without_islands = main_ice"
r.mask -r
r.category map=basins | r.category map=shelf_without_islands rules=-
r.category map=basins | r.category map=grounded_with_islands rules=-
#+END_SRC
*** SMB (MAR)
#+BEGIN_SRC bash :exports both :results verbatim
g.mapset -c MAR
ncdump -v TIME dat/MARv3.12-ANT-35km-annual.nc4 # 30-39 = 2010-2019
ncra --overwrite -d TIME,30,39 dat/MARv3.12-ANT-35km-annual.nc4 tmp/MAR_AQ.nc
ncdump -v X tmp/MAR_AQ.nc # 176
ncdump -v Y tmp/MAR_AQ.nc # 148
g.region w=$(( -3010000 - 17500 )) e=$(( 3115000 + 17500 )) s=$(( -2555000 - 17500 )) n=$(( 2590000 + 17500 )) res=35000 -p
var=SF # debug
for var in SF RF RU ME SMB EVA CON DEP SUB MSK AREA; do
r.in.gdal -o input=NetCDF:tmp/MAR_AQ.nc:${var} output=${var}
r.region -c map=${var}
done
r.mapcalc "MASK = if(MSK > 50)" --o
r.mapcalc "scale_mask = MSK / 100" # if only X % of a cell is ice, scale by that.
# scale
## units are mm.w.eq. per grid cell. Grid cell areas are in km^2
## + mm.w.eq. -> m w.eq.: /1E3
## + m w.eq -> kg: *1E3
## + area in km^2 -> m^2: *1E3*1E3
## + kg -> Gt: /1E12
# ds = ds/1E3 * 1E3 * ds['AREA']*1E3*1E3 / 1E12
for var in SF RF RU ME SMB EVA CON DEP SUB; do
r.mapcalc "${var} = (${var}/1000) * 1000 * (AREA * 1000*1000) * scale_mask / exp(10,12)"
done
r.mapcalc "RFZ = ME + RF - RU"
#+END_SRC
**** Stats
***** SMB components grounded and shelf
#+BEGIN_SRC bash :exports both :results verbatim :session *projects/sankey-shell*
for mask in grounded_with_islands shelf_without_islands; do
echo $mask
r.mask --o raster=${mask}@PERMANENT --q
for var in RF CON DEP SF RFZ EVA RU SUB; do # SF RF RU EVA CON DEP SUB ME; do
echo -n "${var} ${mask}"
r.univar -gt map=${var} zones=${mask}@PERMANENT | cut -d"|" -f2,13 | column -s"|" -t | sed 's/label.*//'
r.univar -g ${var} | grep sum
echo "#"; echo "#"
done
done
r.mask -r --q
#+END_SRC
#+RESULTS:
#+begin_example
grounded_with_islands
RF grounded_with_islands
East 0.53462477161335
West 0.2532230323633
Peninsula 2.22781624112255
[01;31m[Ksum[m[K=3.0156640450992
CON grounded_with_islands
East 0.00144321189675
West 0.00241510084115
Peninsula 0.01323398293865
[01;31m[Ksum[m[K=0.01709229567655
DEP grounded_with_islands
East 36.9861991237577
West 23.8279628054373
Peninsula 5.8151846089547
[01;31m[Ksum[m[K=66.6293465381494
SF grounded_with_islands
East 1392.4748276417
West 723.601551820622
Peninsula 281.709413065019
[01;31m[Ksum[m[K=2397.78579252734
RFZ grounded_with_islands
East 14.6234218646823
West 5.16798014233454
Peninsula 19.3083482881789
[01;31m[Ksum[m[K=39.0997502951956
EVA grounded_with_islands
East 0.6060187163407
West 0.2013515636148
Peninsula 0.6982005019075
[01;31m[Ksum[m[K=1.505570781863
RU grounded_with_islands
East 1.53022074184155
West 0.0059355454226
Peninsula 1.8878783909651
[01;31m[Ksum[m[K=3.42403467822925
SUB grounded_with_islands
East 150.9735683004
West 32.9662640970179
Peninsula 12.5062719218602
[01;31m[Ksum[m[K=196.446104319277
shelf_without_islands
RF shelf_without_islands
East 0.842541426357001
West 0.4498711186449
Peninsula 2.307238481508
[01;31m[Ksum[m[K=3.59965102650989
CON shelf_without_islands
East 0.0031724865901
West 0.0019547481581
Peninsula 0.03522553443475
[01;31m[Ksum[m[K=0.04035276918295
DEP shelf_without_islands
East 5.70103389655939
West 6.05853236904585
Peninsula 1.5062480433876
[01;31m[Ksum[m[K=13.2658143089928
SF shelf_without_islands
East 172.41137746281
West 180.27593549343
Peninsula 56.6841289993761
[01;31m[Ksum[m[K=409.371441955615
RFZ shelf_without_islands
East 25.7408112537284
West 9.64777465721551
Peninsula 32.3899504973317
[01;31m[Ksum[m[K=67.7785364082756
EVA shelf_without_islands
East 0.70256713317005
West 0.2422464141299
Peninsula 0.70164322473235
[01;31m[Ksum[m[K=1.6464567720323
RU shelf_without_islands
East 1.5089940427256
West 0.0304982132294
Peninsula 4.35827769837335
[01;31m[Ksum[m[K=5.89776995432835
SUB shelf_without_islands
East 23.4661462650309
West 8.5418917438099
Peninsula 3.94097993607855
[01;31m[Ksum[m[K=35.9490179449194
[Raster MASK present]
[?2004l
#+end_example
*** Basal melt
Van Liefferinge (2013) http://doi.org/10.5194/cp-9-2335-2013
Convert MAT file to XYZ for importing into GRASS
#+BEGIN_SRC jupyter-python :exports both
import scipy as sp
import numpy as np
import pandas as pd
mat = sp.io.loadmat('/home/kdm/data/Van_Liefferinge_2023/Melt_Mean_Std_15exp.mat')
X = mat['X'].flatten() * 1E3 # convert from km to m
Y = mat['Y'].flatten() * 1E3
m = mat['MeanMelt'].flatten() / 10 # cm to mm
e = mat['StdMelt'].flatten() / 10 # cm to mm
melt = pd.DataFrame(np.array([X,Y,m,e]).T, columns=['x','y','melt','err']).dropna()
melt.to_csv('./tmp/melt.csv', header=False, index=False)
melt.head()
#+END_SRC
#+RESULTS:
| | x | y | melt | err |
|--------+-----------+------------+-------------+-------------|
| 148741 | 1.045e+06 | -2.14e+06 | 1e-09 | 1.71243e-25 |
| 149859 | 1.03e+06 | -2.135e+06 | 0.00146608 | 0.000148305 |
| 149860 | 1.035e+06 | -2.135e+06 | 0.000266042 | 0.000389444 |
| 149861 | 1.04e+06 | -2.135e+06 | 1e-09 | 1.71243e-25 |
| 149862 | 1.045e+06 | -2.135e+06 | 0.00045698 | 0.000668948 |
#+BEGIN_SRC bash :exports both :results verbatim
grass ./G_AQ/PERMANENT
g.mapset -c liefferinge_2023
r.in.xyz input=./tmp/melt.csv output=melt sep=, --o
r.in.xyz input=./tmp/melt.csv output=err z=4 sep=, --o
#+END_SRC
#+BEGIN_SRC bash :exports both :results verbatim :session *projects/sankey-shell*
echo "All: " $(r.univar -g map=melt | grep sum)
echo "All: " $(r.univar -g map=err | grep sum)
# echo ""
r.univar -gt map=melt zones=basins | cut -d"|" -f2,13 | column -s"|" -t
#+END_SRC
#+RESULTS:
#+begin_example
All: sum=69.3982306335468
[?2004lAll: sum=20.0261054475124
[?2004l
[?2004llabel sum
East 46.7540492694752
West 18.8528624157926
Peninsula 3.18704264192471
Islands 0.279139711405429
#+end_example
Uncertainty % is 20/69 = 0.289855072464
*** Discharge
+ Discharge is "grounded discharge"
+ Input to ice shelves where ice shelves exist
+ Calving (similar to Greenlandic discharge) where ice shelves do not exist.
**** Rignot 2019 (Shelf, non-shelf, and Island)
***** Load
#+NAME: load_rignot
#+BEGIN_SRC jupyter-python :exports both
import pandas as pd
df = pd.read_excel("~/data/Rignot_2019/pnas.1812883116.sd01.xlsx", index_col=0)
##############################################################################
###
### cleanup
###
df = df.loc[df.index.dropna()]
for i in [0,0,0]: # drop Excel rows 2,3,4
df = df.drop(index=df.index[i])
# Drop super-shelves and rename indented sub-shelves
super_shelf = ["LarsenB", "Wordie", "Ronne", "Ross West", "Ross East", "Amery_Ice_Shelf", "Filchner", "AP", "WAIS", "EAIS", "TOTAL SURVEYED"]
df = df.drop(index=super_shelf)
for i in df.index:
if i[0] == ' ': df = df.rename(index={i: i.strip()})
for c in df.columns: # Drop extra columns
if 'Unnamed' in str(c):
df = df.drop(columns=c)
df = df.drop(columns=["Basin.1", "σ SMB", "σ D", "D type"]) # Drop unused columns
##############################################################################
# Green color = no ice shelf
noshelf = ["West_Graham_Land", "Eastern_Graham_Land", "Hektoria_Headland", "Evans_Headland", "Drygalski_Headland", "LarsenA", "Rydberg_Peninsula", "Zonda_Eureka", "Cape_Jeremy", "Wilkins_George_VI", "Wilkins_Island", "Thomson", "Fox", "Cooke", "Walgreen_Coast", "Lucchitta_Velasco", "Jackson-Perkins", "Frostman-Lord-Shuman-Anandakri", "Shirases_Coast", "Saunders_Coast", "Ross_East1", "Ross_East2", "Ross_East3", "Ross_East4", "Ross_East5", "Dry_Valleys", "Icebreaker-Fitzgerald", "Victoria_Land", "Oates_Coast", "Wilkes_Land", "Adelie_Coast", "Sabrina_Coast", "Clarie_Coast", "Law_Dome", "Budd_Coast", "Knox_Coast", "Ingrid_Christensen_Coast", "Wilhelm_II_Coast", "Enderby_Land", "Prince_Olav_Coast", "Mawson_Coast"]
df['shelf'] = 1
df.loc[noshelf, 'shelf'] = 0
# Sum numeric columns
df.loc['Sum'] = np.nan
for c in df.columns: # convert to numbers
try: df[c] = pd.to_numeric(df[c])
except: df.loc['Sum',c] = 'All'
cols = df.select_dtypes(include=[np.number]).columns.drop('shelf')
df.loc['Sum', cols] = df[cols].sum(axis='rows')
cols = df.columns[0:10].to_list()
cols.insert(3,'shelf')
df[cols].tail(10)
#+END_SRC
#+RESULTS: load_rignot
| Glacier name | Basin | Region | Subregion | shelf | SMB | D | 1979 | 1980 | 1981 | 1982 | 1983 |
|-----------------+---------+----------+--------------------------+---------+-----------+-----------+-----------+------------+------------+------------+------------|
| Stancomb_Wills | K-A | East | Stancomb_Wills_Ice_Shelf | 1 | 22.03 | 21.39 | 25.261 | 24.7291 | 24.1972 | 23.6653 | 23.1333 |
| Princess_Martha | K-A | East | Princess_Martha_Coast | 1 | 0.21 | 0.21 | 0.21 | 0.21 | 0.21 | 0.21 | 0.21 |
| Coats_Coast | K-A | East | Coats_Coast | 1 | 6.43 | 6.43 | 6.43 | 6.43 | 6.43 | 6.43 | 6.43 |
| Academy | J"-K | East | Filchner_Ice_Shelf | 1 | 24.27 | 24.1 | 24.27 | 24.2505 | 24.231 | 24.2115 | 24.1921 |
| Support_Force | J"-K | East | Filchner_Ice_Shelf | 1 | 9.72 | 9.361 | 9.72 | 9.73057 | 9.74115 | 9.75172 | 9.7623 |
| Recovery | J"-K | East | Filchner_Ice_Shelf | 1 | 41.05 | 41.05 | 41.05 | 41.1242 | 41.1983 | 41.2725 | 41.3467 |
| Slessor | J"-K | East | Filchner_Ice_Shelf | 1 | 26.11 | 24.916 | 26.11 | 26.1256 | 26.1412 | 26.1568 | 26.1724 |
| Bailey | J"-K | East | Filchner_Ice_Shelf | 1 | 8.98 | 8.61 | 8.98 | 9.00126 | 9.02252 | 9.04378 | 9.06504 |
| Islands | nan | Islands | Islands | 1 | 76.9899 | 76.9899 | 76.9899 | 76.9899 | 76.9899 | 76.9899 | 76.9899 |
| Sum | All | All | All | nan | 2097.57 | 2236.96 | 2126.64 | 2133.49 | 2140.25 | 2147.01 | 2153.77 |
***** Shelf vs Non-shelf discharge
+ WARNING: Using shelf vs. non-shelf is important and can be done for all AQ, but Rignot "Island" discharge (~77 Gt/year) doesn't provide enough metadata to break down by east/west/peninsula.
+ Instead, for all islands in NSIDC-0709.002 product, find their region (east, west, peninsula), and calculate area of islands in each region, and then split values by area. That assumes all islands have the same flux (volume flow rate per unit area).
#+BEGIN_SRC jupyter-python :exports both
<>
c = np.arange(2010,2017+1)
dd = df.groupby(['shelf','Region']).sum().drop(columns=['Basin','Subregion'])[c].mean(axis='columns')
dd.loc['Non-shelf discharge'] = dd.loc[0,:].sum()
dd.loc['shelf discharge'] = dd.loc[1,:].sum()
dd['Total discharge'] = dd.loc[['Non-shelf discharge','shelf discharge']].sum()
dd
# df_shelf = df[df['shelf'] == 1][c].mean(axis='columns')
# df_noshelf = df[df['shelf'] == 0][c].mean(axis='columns')
# df_shelf
# print("Total discharge: ", df[df['shelf'] >= 0][c].mean(axis='columns').sum())
# print('Shelf discharge: ', df_shelf.sum())
# print('Non-shelf discharge: ', df_noshelf.sum())
#+END_SRC
#+RESULTS:
#+begin_example
shelf Region
0.0 East 177.140000
Peninsula 131.398873
West 23.003920
1.0 East 926.544384
Islands 76.989900
Peninsula 205.140504
West 768.695078
Non-shelf discharge 331.542793
shelf discharge 1977.369866
Total discharge 2308.912659
dtype: float64
#+end_example
Non-shelf discharge from Rignot is:
| Region | Values | Sum | Comment |
|-----------+--------------+-------+--------------------------------------------|
| East | 177 + 77*0.6 | 223.2 | East non-shelf discharge + 60 % of islands |
| West | 23 + 77*0.3 | 46.1 | West non-shelf discharge + 30 % of islands |
| Peninsula | 131 + 77*0.1 | 138.7 | Peninsula + 10 % islands |
#+TBLFM: $3=$2
**** Davison 2023 (Discharge to shelf)
This is steady-state discharge from grounded ice to shelves.
#+NAME: load_davison_discharge
#+begin_src jupyter-python :exports both
import numpy as np
import pandas as pd
fname = '~/data/Davison_2023/adi0186_table_s2.xlsx'
loc = pd.read_excel(fname, sheet_name='Total mass changes', index_col = 0, usecols = 'B,C,D', skiprows = 4)
loc = loc.drop('Antarctic Ice Shelves')
df = pd.read_excel(fname, sheet_name='Discharge', index_col = 1, skiprows = 3)
df = df[df.columns[1::2]]
df.columns = [np.floor(c).astype(int) for c in df.columns]
df = df.drop(index=df.index[0])
df = df.drop(index='Antarctic Ice Shelves')
df = df[np.arange(2010,2020)].mean(axis='columns')
df.name = 'Mass'
df
#+end_src
#+RESULTS: load_davison_discharge
#+begin_example
Abbot 32.473268
Ainsworth 0.157966
Alison 2.985331
Amery 78.564587
Andreyev 2.207105
...
Withrow 0.480019
Wordie 7.754308
Wylde 0.005026
Zelee 0.42351
Zubchatyy 0.469816
Name: Mass, Length: 162, dtype: object
#+end_example
#+begin_src jupyter-python :exports both
<>
df = loc.join(df)
import geopandas as gpd
fname = '~/data/NSIDC/NSIDC-0709.002/1992.02.07/IceBoundaries_Antarctica_v02.shp'
ew = gpd.read_file(fname)
df = gpd.GeoDataFrame(df, geometry=gpd.points_from_xy(df['longitude'],df['latitude']), crs="EPSG:4326")
df = df.drop(columns=['latitude','longitude'])
df = df.to_crs('epsg:3031')
e = ew.to_crs('epsg:3031')
idx = ew.sindex.nearest(df['geometry'], return_all=False)
df['Region'] = ''
for dfidx,ewidx in idx.T:
arr = df.iloc[dfidx].copy(deep=True)
arr['Region'] = ew.iloc[ewidx]['Regions']
df.iloc[dfidx] = arr
# df.loc['Total'] = [df['Mass'].sum(), None, 'All']
dd = df[['Mass','Region']].groupby('Region').sum()
dd.loc['Total'] = dd.sum()
dd
#+end_src
#+RESULTS:
| Region | Mass |
|-----------+------------|
| East | 923.794 |
| Islands | 1.28338 |
| Peninsula | 152.536 |
| West | 857.468 |
| Total | 1935.08 |
Total discharge is then
| Region | Rignot (Ground-to-ocean + Islands | Davison (Ground-to-shelf) | Sum |
|-----------+-----------------------------------+---------------------------+------|
| East | 223 | 924 | 1147 |
| West | 46 | 856 | 902 |
| Peninsula | 139 | 153 | 292 |
| Total | | | 2341 |
#+TBLFM: $4=$2+$3::@>$4=vsum(@2..@-1)
*** Antarctic Ice shelves
**** Calving: Greene 2022
#+NAME: load_greene_2022_calving
#+begin_src jupyter-python :exports both :display plain
import pandas as pd
fname = "/home/kdm/data/Greene_2022/data/greene_Supplementary_Table_1.xlsx"
df = pd.read_excel(fname, index_col=1, skiprows=4)
df = df.drop(index=df.index[0])
df = df.drop(index=['Antarctica'])
df = df[df.columns[[1,2,9]]]
df.columns = ['latitude','longitude','Mass']
import geopandas as gpd
fname = '~/data/NSIDC/NSIDC-0709.002/1992.02.07/IceBoundaries_Antarctica_v02.shp'
ew = gpd.read_file(fname)
df = gpd.GeoDataFrame(df, geometry=gpd.points_from_xy(df['longitude'],df['latitude']), crs="EPSG:4326")
df = df.to_crs('epsg:3031')
e = ew.to_crs('epsg:3031')
idx = ew.sindex.nearest(df['geometry'], return_all=False)
df['Region'] = ''
for dfidx,ewidx in idx.T:
arr = df.iloc[dfidx].copy(deep=True)
arr['Region'] = ew.iloc[ewidx]['Regions']
df.iloc[dfidx] = arr
df = df.drop(columns=['latitude','longitude'])
# df.loc['Total'] = [df['Mass'].sum(), None, 'All']
dd = df[['Mass','Region']].groupby('Region').sum()
dd.loc['Total'] = dd.sum(axis='rows')
dd
#+end_src
#+RESULTS: load_greene_2022_calving
: Mass
: Region
: 44.705315
: East 694.100336
: Islands 1.518034
: Peninsula 103.675815
: West 566.997529
: Total 1410.997028
The above is shelf calving
Total calving is shelf calving (Greene) + non-shelf calving (331; Rignot) + islands (77; Rignot)
| Region | Values | Sum | Comment |
|-----------+-----------+-----+--------------------------------------------|
| East | 694 + 223 | 917 | East non-shelf discharge + 60 % of islands |
| West | 567 + 154 | 721 | West non-shelf discharge + 30 % of islands |
| Peninsula | 104 + 31 | 135 | Peninsula + 10 % islands |
#+TBLFM: $3=$2
***** Uncertainty
From p.3 of citet:greene_2022 "Antarctica has experienced a net loss of 5,874 ± 396 Gt of ice owing to calving"
396/5874 % = 6.74157303371
From data K189 & L189 = 1411.0 28.1 or 28/1411% = 1.98440822112
**** Shelf freeze/melt
#+BEGIN_SRC jupyter-python :exports both
import xarray as xr
ds = xr.open_mfdataset("~/data/Paolo_2023/ANT_G1920V01_IceShelfMelt.nc")
ds = ds[['melt','melt_err']].sel({'time':slice('2010-01-01','2017-12-31')}).mean(dim='time')
delayed_obj = ds.to_netcdf('tmp/shelf_melt.nc', compute=False)
from dask.diagnostics import ProgressBar
with ProgressBar():
results = delayed_obj.compute()
print(ds)
#+END_SRC
#+RESULTS:
: [########################################] | 100% Completed | 5.35 s
: Size: 68MB
: Dimensions: (y: 2916, x: 2916)
: Coordinates:
: * x (x) float64 23kB -2.798e+06 -2.796e+06 ... 2.796e+06 2.798e+06
: * y (y) float64 23kB 2.798e+06 2.796e+06 ... -2.796e+06 -2.798e+06
: Data variables:
: melt (y, x) float32 34MB dask.array
: melt_err (y, x) float32 34MB dask.array
#+BEGIN_SRC bash :exports both :results verbatim
g.mapset -c Paolo_2023
ncdump -v x tmp/shelf_melt.nc # 2916x2916
ncdump -v y tmp/shelf_melt.nc
x0=-2798407.5
x1=2798392.5
y0=-2798392.5
y1=2798407.5
g.region w=$(( -2798407 - 960 )) e=$(( 2798392 + 960 )) s=$(( -2798392 - 960 )) n=$(( 2798407 + 960 )) res=1920 -p
r.mapcalc "area = area()"
r.in.gdal -o input=NetCDF:tmp/shelf_melt.nc:melt output=melt
r.in.gdal -o input=NetCDF:tmp/shelf_melt.nc:melt_err output=err
r.region -c map=melt
r.region -c map=err
## + kg/m^2 -> Gt: / 1E12
r.mapcalc "melt = melt * 1000 * area / exp(10,12)" --o
r.mapcalc "err = err * 1000 * area / exp(10,12)" --o
r.mapcalc "melt_on = if(melt > 0, melt, null())"
r.mapcalc "err_on = if(melt > 0, err, null())"
r.mapcalc "melt_off = if(melt < 0, melt, null())"
r.mapcalc "err_off = if(melt < 0, err, null())"
r.colors -ae map=melt color=difference
r.colors -ge map=melt_on color=viridis
r.colors -ge map=melt_off color=viridis
# d.rast melt
# d.rast melt_on
# d.rast melt_off
r.mapcalc "basins = if((basins@PERMANENT == 1) | (basins@PERMANENT == 11), 1, 0)"
r.mapcalc "basins = if((basins@PERMANENT == 2) | (basins@PERMANENT == 12), 2, basins)"
r.mapcalc "basins = if((basins@PERMANENT == 3) | (basins@PERMANENT == 13), 3, basins)"
r.colors map=basins color=viridis
r.category basins separator=":" rules=- << EOF
1:East
2:West
3:Peninsula
EOF
#+END_SRC
***** Stats
#+begin_src bash :exports both :results verbatim :session *projects/sankey-shell*
r.grow.distance input=basins value=basins_grow distance=10 --q
r.mapcalc "basins_grow = int(basins_grow)" --q
r.category map=basins | r.category map=basins_grow rules=- --q
#+end_src
#+begin_src bash :exports both :results verbatim :session *projects/sankey-shell*
echo "NET"
r.univar -gt map=melt zones=basins_grow | cut -d"|" -f2,13 | column -s"|" -t | sed 's/label.*//'
# r.univar -gt map=err zones=basins | cut -d"|" -f2,13 | column -s"|" -t | sed 's/label.*//'
r.univar -g melt | grep sum
r.univar -g err | grep sum
echo ""
echo "FREEZE_ON"
r.univar -gt map=melt_on zones=basins_grow | cut -d"|" -f2,13 | column -s"|" -t | sed 's/label.*//'
# r.univar -gt map=err_on zones=basins | cut -d"|" -f2,13 | column -s"|" -t | sed 's/label.*//'
r.univar -g melt_on | grep sum
r.univar -g err_on | grep sum
echo ""
echo "MELT_OFF"
r.univar -gt map=melt_off zones=basins_grow | cut -d"|" -f2,13 | column -s"|" -t | sed 's/label.*//'
# r.univar -gt map=err_off zones=basins | cut -d"|" -f2,13 | column -s"|" -t | sed 's/label.*//'
r.univar -g melt_off | grep sum
r.univar -g err_off | grep sum
#+end_src
#+RESULTS:
#+begin_example
NET
[?2004lEast -319.34788697967
West -537.161194600709
Peninsula -153.245144876904
[?2004l[01;31m[Ksum[m[K=-1009.75422645726
[?2004l[01;31m[Ksum[m[K=3041.55065208086
[?2004l
[?2004lFREEZE_ON
[?2004lEast 207.669949989514
West 146.976649882162
Peninsula 10.8694466267689
[?2004l[01;31m[Ksum[m[K=365.516046498447
[?2004l[01;31m[Ksum[m[K=1086.24089094716
[?2004l
[?2004lMELT_OFF
[?2004lEast -527.017836969183
West -684.137844482863
Peninsula -164.114591503673
[?2004l[01;31m[Ksum[m[K=-1375.27027295574
[?2004l[01;31m[Ksum[m[K=1955.30976113372
#+end_example
**** GZ retreat
Email from Davison
| Ice Shelf | Mass change due to grounding line migration from 1997 to 2021 (Gt) | Error (Gt) |
| Pine Island | 220 | 40 |
| Thwaites | 230 | 25 |
| Crosson | 200 | 25 |
| Dotson | 420 | 80 |
(220+230+200+420)/(2021-1997) = 44.5833333333
Uncertainty: p. 3 of citet:davison_2023 "groundling line retreat (1070 ± 170 Gt),"
170/1070 % = 15.8878504673
**** Frontal retreat and advance: Greene 2022
#+NAME: load_greene_2022_adv_ret
#+begin_src jupyter-python :exports both
import pandas as pd
import numpy as np
fname = "/home/kdm/data/Greene_2022/data/greene_Supplementary_Table_1.xlsx"
df = pd.read_excel(fname, index_col=1, skiprows=4)
##############################################################################
###
### cleanup
###
df = df.drop(index=df.index[0])
df = df.drop(index=['Antarctica'])
lon = df['Unnamed: 3']
lat = df['Unnamed: 2']
for c in df.columns: # Drop extra columns
if 'Unnamed' in str(c):
df = df.drop(columns=c)
if 'Gt/yr' in str(c):
df = df.drop(columns=c)
if ('control run' in str(c)) | ('instantaneous' in str(c)):
df = df.drop(columns=c)
for c in df.columns:
if type(c) == str: df = df.drop(columns=c)
# df = df.drop(columns=[2000.75])
# df = df.drop(columns=[1997.75])
# df.columns = df.columns.round().astype(int)
##############################################################################
#+end_src
#+RESULTS: load_greene_2022_adv_ret
#+name: green_2022_mean
#+begin_src jupyter-python :exports both
<>
c = df.columns
diff = df.diff(axis='columns')[c]
diff_gain = diff[diff > 0].sum(axis='columns')
diff_loss = diff[diff < 0].sum(axis='columns')
diff_gain.name = 'Mass'
diff_loss.name = 'Mass'
df_gain = pd.DataFrame(diff_gain)
df_loss = pd.DataFrame(diff_loss)
df_net = df_loss + df_gain
print("Net:")
print('Mass gain', df_net[df_net > 0].sum(axis='rows').values)
print('Mass Loss', df_net[df_net < 0].sum(axis='rows').values)
print('Net mass change', df_net.sum(axis='rows').values)
dt = df.columns[-1] - df.columns[0]
print("\nPer year:")
print('Mass gain', df_net[df_net > 0].sum(axis='rows').values / dt)
print('Mass Loss', df_net[df_net < 0].sum(axis='rows').values / dt)
print('Net mass change', df_net.sum(axis='rows').values / dt)
#+end_src
#+RESULTS: green_2022_mean
: Net:
: Mass gain [4563.264309048649]
: Mass Loss [-9434.897965610035]
: Net mass change [-4871.633656561384]
:
: Per year:
: Mass gain [194.59549292318295]
: Mass Loss [-402.3410646315572]
: Net mass change [-207.74557170837417]
+ Most numbers here match what's in the publication
+ Neither the total nor Ronne match.
+ Here, total is 4871 Gt net change.
+ Below, Ronne net loss is 1031
+ From the paper (paragraph under Fig. 2)
+ Total should be 5874 (missing 5874-4871 = 1003)
+ Ronne should be 2034 (missing 2034-1031 = 1003)
+ But Thwaites, Larsen C, and Ross West match paper, so it seems like I'm parsing the dataset correctly.
+ Filchner matches mass gain.
+ Miss 1003: https://www.nature.com/articles/s41586-022-05037-w/figures/8
Find the top 10 shelves with net and gross mass gain and loss total (summed) over the period
#+begin_src jupyter-python :exports both
tmp = pd.DataFrame(index=np.arange(10))
tmp['Net gain: Name'] = df_net.sort_values(by='Mass', ascending=False).head(10).index
tmp['Net gain: Mass'] = df_net.sort_values(by='Mass', ascending=False).head(10)['Mass'].values
tmp['Net loss: Name'] = df_net.sort_values(by='Mass', ascending=True).head(10).index
tmp['Net loss: Mass'] = df_net.sort_values(by='Mass', ascending=True).head(10)['Mass'].values
tmp['Gross gain: Name'] = df_gain.sort_values(by='Mass', ascending=False).head(10).index
tmp['Gross gain: Mass'] = df_gain.sort_values(by='Mass', ascending=False).head(10)['Mass'].values
tmp['Gross loss: Name'] = df_loss.sort_values(by='Mass', ascending=True).head(10).index
tmp['Gross loss: Mass'] = df_loss.sort_values(by='Mass', ascending=True).head(10)['Mass'].values
tmp
#+end_src
#+RESULTS:
| | Net gain: Name | Net gain: Mass | Net loss: Name | Net loss: Mass | Gross gain: Name | Gross gain: Mass | Gross loss: Name | Gross loss: Mass |
|----+------------------+------------------+------------------+------------------+--------------------+--------------------+--------------------+--------------------|
| 0 | Filchner | 1796.26 | Thwaites | -1968.41 | Ronne | 2762.7 | Ronne | -3794.31 |
| 1 | Amery | 569.883 | Larsen C | -1166.92 | Ross West | 1968.02 | Ross West | -2897.63 |
| 2 | Cook | 414.571 | Ronne | -1031.61 | Filchner | 1843.53 | Thwaites | -2248.52 |
| 3 | Shackleton | 369.773 | Ross West | -929.617 | Amery | 917.121 | Larsen C | -1619.1 |
| 4 | Brunt Stancomb | 362.787 | Wilkins | -622.156 | Ross East | 857.219 | Pine Island | -1231.67 |
| 5 | West | 278.333 | Larsen B | -530.038 | Pine Island | 780.117 | Ross East | -1135.23 |
| 6 | Jelbart | 169.419 | Pine Island | -451.554 | Brunt Stancomb | 493.663 | Ninnis | -675.229 |
| 7 | Fimbul | 148.221 | Mertz | -381.971 | Shackleton | 493.661 | Wilkins | -642.061 |
| 8 | Riiser-Larsen | 84.376 | Ninnis | -300.936 | Cook | 454.636 | Larsen B | -584.603 |
| 9 | Pourquoi Pas | 64.9883 | Larsen A | -286.377 | Larsen C | 452.177 | Mertz | -571.076 |
Now convert to Gt/year
#+BEGIN_SRC jupyter-python :exports both
for col in tmp.columns:
if 'Mass' in col: tmp[col] = tmp[col] / c.size
tmp
#+END_SRC
#+RESULTS:
| | Net gain: Name | Net gain: Mass | Net loss: Name | Net loss: Mass | Gross gain: Name | Gross gain: Mass | Gross loss: Name | Gross loss: Mass |
|----+------------------+------------------+------------------+------------------+--------------------+--------------------+--------------------+--------------------|
| 0 | Filchner | 74.8441 | Thwaites | -82.0169 | Ronne | 115.113 | Ronne | -158.096 |
| 1 | Amery | 23.7451 | Larsen C | -48.6219 | Ross West | 82.0007 | Ross West | -120.735 |
| 2 | Cook | 17.2738 | Ronne | -42.9837 | Filchner | 76.8138 | Thwaites | -93.6885 |
| 3 | Shackleton | 15.4072 | Ross West | -38.734 | Amery | 38.2134 | Larsen C | -67.4626 |
| 4 | Brunt Stancomb | 15.1161 | Wilkins | -25.9232 | Ross East | 35.7174 | Pine Island | -51.3196 |
| 5 | West | 11.5972 | Larsen B | -22.0849 | Pine Island | 32.5049 | Ross East | -47.3011 |
| 6 | Jelbart | 7.05912 | Pine Island | -18.8147 | Brunt Stancomb | 20.5693 | Ninnis | -28.1346 |
| 7 | Fimbul | 6.17586 | Mertz | -15.9155 | Shackleton | 20.5692 | Wilkins | -26.7525 |
| 8 | Riiser-Larsen | 3.51567 | Ninnis | -12.539 | Cook | 18.9432 | Larsen B | -24.3585 |
| 9 | Pourquoi Pas | 2.70785 | Larsen A | -11.9324 | Larsen C | 18.8407 | Mertz | -23.7948 |
#+begin_src jupyter-python :exports both
<> # provides df_net
# df = df_net
df['longitude'] = lon
df['latitude'] = lat
import geopandas as gpd
fname = '~/data/NSIDC/NSIDC-0709.002/1992.02.07/IceBoundaries_Antarctica_v02.shp'
ew = gpd.read_file(fname)
df = gpd.GeoDataFrame(df, geometry=gpd.points_from_xy(df['longitude'],df['latitude']), crs="EPSG:4326")
df = df.to_crs('epsg:3031')
e = ew.to_crs('epsg:3031')
idx = ew.sindex.nearest(df['geometry'], return_all=False)
df['Region'] = ''
for dfidx,ewidx in idx.T:
arr = df.iloc[dfidx].copy(deep=True)
arr['Region'] = ew.iloc[ewidx]['Regions']
df.iloc[dfidx] = arr
df = df.drop(columns=['latitude','longitude','geometry'])
# df.loc['Total'] = [df['Mass'].sum(), None, 'All']
# df.groupby('Region').sum().round()
diff = df[c].diff(axis='columns')
diff_gain = diff[diff > 0].sum(axis='columns')
diff_loss = diff[diff < 0].sum(axis='columns')
diff_gain.name = 'Mass'
diff_loss.name = 'Mass'
df_gain = pd.DataFrame(diff_gain)
df_loss = pd.DataFrame(diff_loss)
df_net = df_loss + df_gain
df_gain['Region'] = df['Region']
df_loss['Region'] = df['Region']
df_net['Region'] = df['Region']
#+end_src
#+RESULTS:
: Net:
: Mass gain [4563.264309048649]
: Mass Loss [-9434.897965610035]
: Net mass change [-4871.633656561384]
:
: Per year:
: Mass gain [194.59549292318295]
: Mass Loss [-402.3410646315572]
: Net mass change [-207.74557170837417]
#+begin_src jupyter-python :exports both
for loc in ['East','West','Peninsula']:
print("\n", loc)
sub = (df_net['Mass'] > 0) & (df_net['Region'] == loc); print('Mass gain', df_net[sub].drop(columns='Region').sum().values/dt)
sub = (df_net['Mass'] < 0) & (df_net['Region'] == loc); print('Mass loss', df_net[sub].drop(columns='Region').sum().values/dt)
#+end_src
#+RESULTS:
#+begin_example
East
Mass gain [192.48919217062863]
Mass loss [-69.4663382466161]
West
Mass gain [1.923849262408337]
Mass loss [-206.1686424710849]
Peninsula
Mass gain [0.17175689689132026]
Mass loss [-124.70409821345615]
#+end_example
*** GRACE
**** ESA CCI
+ https://data1.geo.tu-dresden.de/ais_gmb/
Results processed by Thorben Döhne are:
| Region | MB | Err | Err % |
|-----------+--------+------+------------|
| Peninsula | -21.4 | 7.2 | -33.644860 |
| East | 35.0 | 40.0 | 114.28571 |
| West | -164.9 | 15.1 | -9.1570649 |
| All | -151.3 | 44.3 | -29.279577 |
#+TBLFM: $4=($3/$2)*100
**** JPL
#+BEGIN_SRC jupyter-python :exports both
import numpy as np
import pandas as pd
from datetime import datetime, timedelta
from uncertainties import unumpy
df = pd.read_csv("~/data/GRACE/antarctica_mass_200204_202410.txt",
comment="H", parse_dates=True, sep="\\s+", header=None,
names=['year','mass','err'])
# Function to convert year.frac to ISO format (YYYY-MM-DD)
def year_frac_to_iso(year_frac):
year = int(year_frac)
frac = year_frac - year
start_of_year = datetime(year, 1, 1)
days_in_year = (datetime(year + 1, 1, 1) - start_of_year).days
date = start_of_year + timedelta(days=frac * days_in_year)
return pd.to_datetime(date.strftime('%Y-%m-%d'))
# Apply the conversion to the 'Year' column
df['date'] = df['year'].apply(year_frac_to_iso)
df = df.drop(columns=['year'])
df = df.set_index('date')
df = df[['mass','err']]
# df.resample('D').mean().interpolate()
# df = df['2010-01-01':'2019-12-31']
df = df['2003-01-01':'2023-12-31']
# df['mass'] = df['mass'] - df['mass'].max()
# arr = unumpy.uarray(df['mass'].values, df['err'].values)
_ = df['mass'].plot() # <-- traditional plot
# df.resample('YS').mean().diff().plot()
# df = df.resample('1D').interpolate().resample('YS').mean().diff().mean()
# df = df.resample('YS').mean().diff().mean()
# print(df['mass'], '+-', df['err'])
print(df['mass'].resample('YS').mean().diff().mean(), '+-', df['err'].resample('YS').mean().mean())
#+END_SRC
#+RESULTS:
:RESULTS:
: -112.11872727272728 +- 38.18804256854257
[[file:./figs_tmp/08a08b6d0213c72ff1030c88955c5d31f26601b4.png]]
:END:
* Misc
** Export tables to CSVs
#+NAME: orgtbl2csv
#+BEGIN_SRC emacs-lisp :var tbl="" :colnames no
(save-excursion
(goto-char (point-min))
(re-search-forward (concat "^#\\+name: " tbl) nil t)
(next-line)
(org-table-export (concat "./dat/" tbl ".csv") "orgtbl-to-csv")
;;(shell-command-to-string (concat "head " tbl ".csv"))
(message (concat "Exported " tbl " to " (concat "./dat/" tbl ".csv")))
)
#+END_SRC
** Convert PDFs to PNG
#+NAME: pdfs2png
#+BEGIN_SRC bash :exports results :results verbatim :results none
convert -density 300 -background white -alpha remove -trim -gravity center -annotate -200+50 'Greenland' gl_baseline.pdf tmp/gl.png
convert -density 300 -background white -alpha remove -trim -gravity center -annotate -200+50 'Antarctica' aq_All.pdf tmp/aq.png
convert -density 300 -background white -alpha remove -trim -gravity center -annotate -100+50 'East' aq_E.pdf ./tmp/aqe.png
convert -density 300 -background white -alpha remove -trim -gravity center -annotate -100+75 'West' aq_W.pdf ./tmp/aqw.png
convert -density 300 -background white -alpha remove -trim -gravity center -annotate -100+60 'Peninsula' aq_P.pdf ./tmp/aqp.png
convert -density 300 -background transparent -alpha remove legend.svg ./tmp/legend.png
composite -gravity center -geometry '100%x75%+200-150' tmp/legend.png tmp/aq.png tmp/aq_legend.png
convert -gravity center -append tmp/{gl,aq_legend}.png ./fig_aq_gl.png
convert -gravity center -append tmp/{aqe,aqw,aqp}.png ./fig_aq_parts.png
#+END_SRC