Aslak Grinsted‎ > ‎

Misc. Debris

A list of debris posts can be found here.

Thoughts on the Pachauri harassment case

posted Feb 23, 2015, 10:35 AM by Aslak Grinsted   [ updated Feb 23, 2015, 11:33 AM ]

Pachauri has been accused of sexual harassment. You can read about it here:
Pachauri's initial defense was to claim that he didn't do it, but that his phone was hacked. That is not entirely implausible considering the dirty lengths that climate obstructionists would go to (e.g. the CRU hack). Hence I initially withheld any commentary. The climate community I follow on twitter has also mostly been quiet on this. 

Since then more news has surfaced. We have seen the content of some of the messages in the case, and more women have spoken out against Pachauri:
I don't know if he is guilty or not, but these additional reports push me towards the womens account.  

I don't usually comment on gender issues, but I think it is important to publicly make it clear that sexual harassment is not OK regardless of who you are, or how much bad publicity it might give for climate science. 

A better windows experience

posted Jan 22, 2015, 9:21 AM by Aslak Grinsted   [ updated Mar 9, 2015, 4:11 AM ]


Chocolatey is a package manager sort of like apt-get for windows, or a command-line version of ninite. You can do stuff like:

choco install <packagename> 
By Luisa Contreras (Flickr: bizcocho de chocolate y nata) [CC BY 2.0 (], via Wikimedia Commons
Usually you have to run choco as an administrator. That means starting the command prompt as an admin. That can get annoying pretty fast, so I recommend installing the sudo package as the next step.

choco install sudo

After this you can install a bunch of other stuff using "sudo choco install xxxx" install, without starting an administrative shell.  

A better terminal, and putty
sudo cinst conemu 

I installed putty manually instead of using chocolatey because that makes it easier to get conemu to cooperate with putty.  see also comments here

SSHFS adds the ability to mount drives over ssh

sudo cinst win-sshfs

I recommend installing FiSSH instead (a fork of win-sshfs) because it is compatible with pageant. I have not tested if it is better than this fork.   


sudo cinst SublimeText3
sudo cinst SublimeText3.powershellalias

Python2.xx (& essentials)

For scientific python you might want to try Anaconda instead of installing python yourself. That is not in the chocolatey.

Note this python section is from memory and may not be optimal.  
sudo cinst python2
pip install --upgrade pip
pip install virtualenv
pip install virtualenvwrapper-win
If you already have python2, then you might want to remove it first using the control panel. 

If PIP is not found, then it is most likely a problem with your path. 

Bayesian exercises for Inverse problems

posted Oct 22, 2014, 2:17 AM by Aslak Grinsted   [ updated Oct 23, 2014, 4:18 AM ]

Below is a pdf with some exercises for that i use in my course on inverse problems.

Problem 0: Goal understand why it is attractive to use loglikelihoods over likelihoods.
Problem 1: An exact model with a single model parameter. This makes easy to plot.
Problem 2: A variant of problem 1 with an inexact model.
Problem 3: Goal: Show that it is easier to calculate derived quantities such as "expected value" of the posterior when you use mcmc.
Problem 4: A problem where it is easy to understand why is the likelihood called a likelihood. 
Problem 5: A problem where you have to grasp many concepts. 

Exercises 4 & 5 are probably the most fun. 

Tracking ice motion in Landsat 8

posted Jun 11, 2014, 1:54 AM by Aslak Grinsted   [ updated Jun 11, 2014, 1:57 AM ]

I have been attempting to track ice motion in landsat 8 files. I have only been able to find the L1T files. These are terrain corrected files. Unfortunately for my study region (the Renland ice cap, East Greenland) the DEM used in this processing is of too poor quality and which introduces large geodetic errors. DEM errors can be seen as residual perspective effects [Click image on the right to expand to an animated GIF showing flipping between two images]. 

If you know where to get L1G, then please email me! 

I would therefore really like to have the L1G product and account for the terrain correction my self (using gimpdem). 

Another advantage of L1G over L1T is that there is less resampling which degrades feature tracking performance. 

(This work is partly motivated by testing a new open source feature tracking & georeferencing toolbox for matlab we have been working on. I am looking forwards to putting it on github.)

Descriptions of the levels of processing 

  • Level 1 Systematically Corrected (L1G)
    • The Level 1G (L1G) data product provides systematic radiometric and geometric accuracy, which is derived from data collected by the sensor and spacecraft. The scene will be rotated, aligned, and georeferenced to a user-specified map projection. Geometric accuracy of the systematically corrected product should be within 250 meters (1 sigma) for low-relief areas at sea level.
  • Level 1 Terrain Corrected (L1T)
    • The Level 1T (L1T) data product provides systematic radiometric accuracy, geometric accuracy by incorporating ground control points, while also employing a Digital Elevation Model (DEM) for topographic accuracy. Geodetic accuracy of the product depends on the accuracy of the ground control points and the resolution of the DEM used.

West Antarctic collapse in the IPCC reports

posted Jun 4, 2014, 2:32 PM by Aslak Grinsted   [ updated Jan 7, 2015, 11:36 PM ]

I have extracted some quotes from the IPCC reports since 1990 showing how West Antarctic Ice Sheet collapse has been discussed over time. It is not an exhaustive list.
  • FAR 1990: 
    • “Antarctica is expected to contribute negatively to sea level due to increased snow accumulation associated with warming. A rapid disintegration of the West Antarctic Ice Sheet due to global warming is unlikely within the next century”. 
  • SAR 1995: 
    • “Nonetheless, the likelihood of a major sea level rise by the year 2100 due to the collapse of the West Antarctic ice sheet is considered low.”
    • "For Antarctica, the sensitivity value is -0.20 ± 0.25 mm/yr/°C (including a term for the possible instability of the West Antarctic ice sheet)," (for a revised IPCC projection in section 7.5.2)
    • "Our ignorance of the specific circumstances under which West Antarctica might collapse limits the ability to quantify the risk of such an event occurring, either in total or in part, in the next 100 to 1000 years."
  • TAR 2001:
    • “The range of projections given above makes no allowance for ice dynamic instability of the WAIS. It is now widely agreed that major loss of grounded ice and accelerated sea level rise are very unlikely during the 21st century.”
  • AR4 2007: 
    • “new evidence of recent rapid changes in the Antarctic Peninsula, West Antarctica and Greenland (see Section has again raised the possibility of larger dynamical changes in the future than are projected by state-of-the-art continental models, such as cited above, because these models do not incorporate all the processes responsible for the rapid marginal thinning currently taking place (Box 4.1; Alley et al., 2005a; Vaughan, 2007).”
    • "Abrupt climate changes, such as the collapse of the West Antarctic Ice Sheet, the rapid loss of the Greenland Ice Sheet or largescale changes of ocean circulation systems, are not considered likely to occur in the 21st century, based on currently available model results."
    • "Recent satellite and in situ observations of ice streams behind disintegrating ice shelves highlight some rapid reactions of ice sheet systems. This raises new concern about the overall stability of the West Antarctic Ice Sheet, the collapse of which would trigger another five to six metres of sea level rise. While these streams appear buttressed by the shelves in front of them, it is currently unknown whether a reduction or failure of this buttressing of relatively limited areas of the ice sheet could actually trigger a widespread discharge of many ice streams and hence a destabilisation of the entire West Antarctic Ice Sheet. Ice sheet models are only beginning to capture such small-scale dynamical processes that involve complicated interactions with the glacier bed and the ocean at the perimeter of the ice sheet. Therefore, no quantitative information is available from the current generation of ice sheet models as to the likelihood or timing of such an event."
    • "We take this as an estimate of the part of the present ice sheet mass imbalance that is due to recent ice flow acceleration (Section, and assume that this contribution will persist unchanged." (note another "scaled-up" discharge scenario was also constructed as an additional term to the main projections of the report.)
  • AR5 2013: 
    • “Only the collapse of the marine-based sectors of the Antarctic ice sheet, if initiated, could cause GMSL to rise substantially above the likely range during the 21st century. This potential additional contribution cannot be precisely quantified but there is medium confidence that it would not exceed several tenths of a meter of sea level rise”.
    • “In summary, ice-dynamics theory, numerical simulations, and paleo records indicate that the existence of a marine-ice sheet instability asso­ciated with abrupt and irreversible ice loss from the Antarctic ice sheet is possible in response to climate forcing. However, theoretical consid­erations, current observations, numerical models, and paleo records cur­rently do not allow a quantification of the timing of the onset of such an instability or of the magnitude of its multi-century contribution.”

See also these posts:

What does "likely" mean?

posted May 9, 2014, 4:48 AM by Aslak Grinsted   [ updated May 14, 2014, 5:34 AM ]

It can be quite difficult to interpret what likely exactly means sometimes. For example in the AR5 sea level chapter they report a likely range of 21-33 cm for thermal expansion (RCP8.5 table 13.5). I have taken that to mean that this was the 66% uncertainty interval based on the IPCC uncertainty guideline note (see table 1). However, I just realized that this range was actually calculated as the 5-95% range from CMIP5. From table 1 I would have called that the very likely range. Can anybody explain to me the motivation for calling it the likely range?

[EDIT2: I asked Jonathan Gregory, and he has been very helpful in explaining the motivation to me. He says: "[Sect explains that] the 5-95% range of CMIP5 models coincides with the assessed likely range of TCR. That's the main reason for proceeding this way. It implies, as you say, that the CMIP5 models do not cover the entire range which is considered "very likely" but we do not have sufficient confidence to quantify a "very likely" range for projections."]

This also affects other parts of the sea level chapter. For example they say that only marine instabilities could cause a rise greater than the likely range. But does that mean that there is a 33% or 10% or chance of that? I have no idea... (or indeed 5% as they appear to use 5-95% for the likely range in this chapter).


The explanation is that the true uncertainty is considered to be greater than the model spread.

The supplementary material to the sea level chapter says that they follow section when they use 5-95% as the likely range. 

[EDIT: I just found an explanation in an SPM footnote: "Calculated from projections as 5−95% model ranges. These ranges are then assessed to be likely ranges after accounting for additional uncertainties or different levels of confidence in models. For projections of global mean sea level rise confidence is medium for both time horizons."]  

If you have any comments then please tweet me at @agrinsted or mail me. I'd greatly appreciate it. 

Vulnerability of marine based sectors of Antarctica

posted May 7, 2014, 2:44 AM by Aslak Grinsted   [ updated Nov 3, 2014, 11:38 PM ]

Should we take the risk of Antarctic collapse seriously?
The IPCC AR5 sea level chapter considered instability of marine-based sectors of the ice sheets to be unlikely (Church et al., 2013). However, post-AR5 modelling indicates that Pine Island Glacier in Antarctica is already engaged in an unstable retreat (Favier et al., 2014), a situation that is projected to extend to neighboring Thwaites glacier (Favier et al., 2014), and even to East Antarctica (Sun et al., 2014).
A recent observational study found an observed sustained increase in ice discharge from the Amundsen Sea Embayment, West Antarctica, from 1973 to 2013 (Mouginot et al. 2014). This prompted this reaction from Eric Steig

Models from Mengel & Levermann (2014) also show that "We have probably overestimated the stability of East Antarctica so far". This is early work and inconclusive, but it highlights a need for further study. [Remark: At EGU2014 Vermeersen and Pollard gave presentations showing WAIS collapse in a few centuries, and a large EAIS response on longer time scales. I assume that this work will be out really soon, so i will update this post when I know more. ]

Crumbling ice figure courtesy Jan Åström, CSC, Finland

AR5 sea level projection unsuitable for adaptation planning?

For adaptation planning we really need the full unconditional uncertainty distribution. The above studies shows that the AR5 conditionality on no marine instability may be excluding a important part of the pdf. It appears to me to be much more than just a very remote possibility. This is also reflected in a recent expert elicitation from Bamber & Aspinall (2013). I question how useful the AR5 projections can be when used as-is for local adaptation planning. 

  1. Church, J.A., P.U. Clark, A. Cazenave, J.M. Gregory, S. Jevrejeva, A. Levermann, M.A. Merrifield, G.A. Milne, R.S. Nerem, P.D. Nunn, A.J. Payne, W.T. Pfeffer, D. Stammer and A.S. Unnikrishnan, (2013a): Sea Level Change. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA [link]
  2. Favier, L., Durand, G., Cornford, S. L., Gudmundsson, G. H., Gagliardini, O., Gillet-Chaulet, F., T. Zwinger, A. J. Payne, and A.M. Le Brocq. (2014). Retreat of Pine Island Glacier controlled by marine ice-sheet instability, Nature Climate Change, 4, 117–121 doi:10.1038/nclimate2094
  3. Sun, S., Cornford, S. L., Liu, Y., & Moore, J. C. (2014). Dynamic response of Antarctic ice shelves to bedrock uncertainty. The Cryosphere Discuss., 8, 479-508, 2014, [link]
  4. Mouginot, J., Rignot, E., & Scheuchl, B. (2014). Sustained increase in ice discharge from the Amundsen Sea Embayment, West Antarctica, from 1973 to 2013. Geophysical Research Letters, 41(5), 1576-1584. [link]
  5. M. Mengel, A. Levermann. Ice plug prevents irreversible discharge from East Antarctica. Nature Climate Change, 2014; DOI: 10.1038/NCLIMATE2226
  6. Bamber, J. L., and W. P. Aspinall. (2013), An expert judgement assessment of future sea level rise from the ice sheets. Nature Climate Change. doi:10.1038/nclimate1778

EGU 2014

posted Apr 15, 2014, 3:07 AM by Aslak Grinsted   [ updated Apr 16, 2014, 5:23 AM ]

Me at EGU 2014:
  • Convening the sea level session (Friday afternoon – you are welcome!)
  • Surface velocities at Engabreen produced from time-lapse feature tracking (Alex)
  • Sea level projections from FAR to AR5
  • Sea level rise projection for Northern Europe
  • Trends in global and regional sea levels since 1807 (Sveta)
  • Haar Wavelet Analysis of Climatic Time Series (Zhang)
  • Trends in normalized hurricane damages in the US
I put some pdfs of my posters below (as I make them and as i decide whether they should be on the internet). The regional sea level projection will not be uploaded before i have submitted a manuscript.

I decided to make my posters in inkscape after some painful experiences with powerpoint created pdfs that turned out crap regardless of which workarounds I tried. I also used adobe inDesign which is really cool but also has a bit of a learning curve when you only use it once per year. Inkscape is limited, but perhaps that is a good thing. Atleast it has been a very smooth ride thus far. Three posters in 2 days (including analysis for one of them). So I highly recommend it for posters. 

Another possible workflow would be to use powerpoint for all the text layout, export as pdf, and then import it in inkscape to do the final tweaking. (Powerpoint tip: avoid shadows/transparencies/gradients if you want nice prints)

Comparison of sea level projections

posted Oct 2, 2013, 4:18 AM by Aslak Grinsted   [ updated Oct 23, 2014, 5:10 AM ]

I have been making a nice figure which compares the 21stC sea level projections from the AR5, with previous IPCC reports, Semi empirical models, and an expert elicitation ... I hope this may be useful in presentations for many people (Feel free to use them where ever). See also this page for a similar figure for the Ice sheet contribution only.

The IPCC FAR,SAR,TAR,AR4 have all been converted to RCP scenarios using conversion factors (see below). All projections have been regularized to 100 years using plain scaling. 

Evolution of sea level rise projections RCP85


  • Extrap: constant rate of sea level rise at present day trend from (An absolute lower limit of plausibility IMO)
  • FAR: full range of SLR projections from FAR (taken from SAR table 7.8)
  • SAR: full range of SLR projections from SAR (taken from TAR table 11.14). (SARp369: "Excluding the possibility of collapse of the West Antarctic ice sheet").
  • TAR: full range of SLR projections from TAR table 11.14. (TAR p.642: "The range of projections given above makes no allowance for icedynamic instability of the WAIS".)
  • AR4: SLR projection excluding scaled-up ice sheet discharge. (AR4 WG1 Table 10.7). 
  • AR4+: SLR projection including scaled-up ice sheet discharge. (AR4 WG1 Table 10.7). Context for "larger values cannot be excluded" can be found in the AR4 SPM.
  • SEM: full range of semi-empirical projections in AR5 (from AR5 fig.13.12). 
  • AR5: "process based" ice sheet projections from AR5 table 13.5. These do not account for a potential collapse of Antarctic marine based sectors which may contribute up to several decimetres (indicated with thin shaded line).
  • Ice sheet experts*. refers to Bamber and Aspinall (2013) table S1 5-95% plus non ice sheet contributions from AR5 table 13.5. Note BA13 does not refer to a specific scenario (hence the asterisk)
  • SLR experts refers to the expert elicitation of Horton et al. 2013 (table 1). They do not provide RCP45 but only RCP85 and RCP3PD. However both SEMs and AR5 agree that the projection for RCP45 lies at about a third of the way between. So I have used this weighing. Some assumptions on normality and covariance structure were necessary to derive 5-95% confidence intervals from the likely ranges reported in AR5 table 13.5. 

"Antarctic collapse" does not literally mean a full collapse, but refers to a marine ice sheet instability. Read AR5 text for more precise meaning. We have also published an estimate of worst case sea level rise which you may find here.
Scenario conversion factors that I have used:
The aim of the conversion factors is to predict what the old models would give if forced with new scenarios.
RCP45/A1B=0.90 & RCP85/A1B=1.20 from AR5 fig 13.10
A1B/IS92A=1.20 calculated by forcing a Jevrejeva model with TAR table II.3.11 
IS92A/SA90=0.87 calculated by forcing a Jevrejeva model with SAR fig Ax.9.

Compare TAR II.3.11 with SAR Ax.9. I'd greatly appreciate any comments on how to improve these conversion factors.

My interpretation:
Interestingly sea level projections was coming down (and narrowing) until we started getting worrying records from the ice sheets. By AR4 it became evident that the ice sheets had a far more dynamic behavior than previously thought (Larsen-B, JakobshavnHelheimKangerdlugssuaq). It became clear that the representation of ice physics and marine ice sheet interaction needed to be improved (see SeaRISE & ice2sea). Since then the evidence for an important dynamic ice sheet contribution has only been strengthening with e.g. Thwaites in the Antarctic, and Petermann in Greenland

Note: I might update figs with a better representation of the SEM uncertainties. 

If you have comments then please email or tweet me. 

Optimistic & over-confident ice sheet projections in AR5

posted Sep 30, 2013, 7:47 AM by Aslak Grinsted   [ updated May 7, 2014, 3:15 AM ]

Take away message: The graph on the right shows that AR5 process based ice sheet projections are optimistic and over confident when compared to views of ice sheet experts. To be fair they do mention a possible collapse scenario which could close the gap.


Arguably the most uncertain component of sea level rise projections is the rate of future ice sheet mass loss. In AR4 ice sheet models were unable to simulate key processes and the AR4 sea level projections were hugely criticized for being too conservative.

Since the AR4 there has been great progress in ice sheet modelling but ice-sheet ocean interaction is still a major challenge. E.g. IPCC AR5 is still unable to give scenario dependent projections of the dynamic ice loss (see AR5 table 13.5) and it is unable to assess the probability of an Antarctic collapse. This lead them to exclude this possibility from the process based sea level projections in table 13.5, and only report the 'likely' range.

The "best" picture of the full uncertainty in ice sheet mass loss projections is from the expert elicitation by Bamber & Aspinall (2013). Figure 1 compares how the AR5 "process based" ice sheet projections (table 13.5) compare to the views from this ice sheet expert elicitation.

Conservative & Overconfident

AR5 process based model projections are much more conservative/ optimistic and has much more narrow uncertainties than the ice sheet experts (Fig.1). There can be no good reason for why the AR5 authors have much greater confidence in their ability to project ice sheet loss than ice sheet experts themselves. Notably the best guess view of ice sheet experts nearly falls outside the AR5 process based range. The worst case scenario from ice sheet experts is more than 60 cm higher than the worst case from the AR5 process models.

Clearly the process based SLR projections from AR5 are over-confident and too conservative by themselves. You have to invoke a significant probability of a collapse of Antarctic marine based sectors before it can be reconciled with Bamber & Aspinall (2013). This is particularly important for the worst case, but it is also evident that even the central estimates from AR5 process based models are practically inconsistent with the views of ice sheet experts (fig.1).

Another way to put it: AR5 ice sheet projections are incompatible with the views held by about half of ice sheet experts.

Footnote: other comparisons

  • Uncertainties from semi-empirical models show much better correspondence with ice sheet experts. (fig.1)
  • I consider a constant mass loss at present day rates to be the absolute lower limit of plausibility in a warming world. This is shown as "Extrap" in figure 1. Notice how this lower limit excludes much of the lower tail of the AR5, AR4, and AR4+ sea level projections.
  • The AR5 projects that there is 21% chance that the 21stC ice sheet mass loss will be slower than the present rate under RCP4.5. IMO this is simply implausible. (Assuming normality of the extrap and AR5 numbers)

Figure 1: Projections of ice sheet mass loss over the 21st century under RCP4.5. The AR5 process based projections appear optimistic and over confident when compared with views of ice sheet experts.


  • Extrap: Fixed rate of mass loss rate based on Shepherd et al. (2012). (An absolute lower limit of plausibility IMO)
  • AR4: Ice sheet mass loss excluding scaled-up ice sheet discharge. (AR4 WG1 Table 10.7). 
  • AR4+: Ice sheet mass loss including scaled-up ice sheet discharge. (AR4 WG1 Table 10.7). Context for "larger values cannot be excluded" can be found in the AR4 SPM.
  • SEM*: full range of semi-empirical projections for RCP4.5 subtracted a central estimate of the non-ice sheet contributions to SLR. (Calculated from AR5 table 13.6 minus central values from AR5 table 13.5). 
  • AR5: "process based" ice sheet projections from table 13.5. These do not account for a potential collapse of Antarctic marine based sectors which may contribute up to several decimetres (shown as thin shaded line).
  • Ice sheet experts. refers to Bamber and Aspinall (2013) table S1 5-95%. Notice: not specifically RCP4.5.
All projections have been scaled to 100 years. AR4 estimates are based on A1B but scaled with the RCP45/A1B ratio (=90%) from AR5 figure 13.10. Some assumptions on normality and covariance structure were necessary to derive 5-95% confidence intervals from the likely ranges reported in AR5 table 13.5. 


1-10 of 67