HadCM3 turns 18 (ish) and vegetation over 120 ka

HadAM3 was first introduced to the world in about 1999, and in 2017 it’s still going strong. When a model is pretty much old enough to smoke, drink, drive and has multiple offspring – what do you do to celebrate? Well, we gave HadCM3 a model and configuration description paper. (There may also have been cake.) The GMD paper has been over a year in the making, a big group effort by all of the users of HadCM3 in BRIDGE, and as the primary co-ordinator of the paper, I’m really glad . The paper is in review with GMD, so you can take a early peek at it.

As a smaller celebration, HadCM3 is the model used in a paper showing an ensemble of 124 (yep really, not a typo) simulations that make up time snapshots of the last 120 ka, with and without vegetation. What we show is that whereas in the near future the carbon is often the dominant terrestrial biosphere effect on climate, the longer timespans and greater uncertainty of the carbon under icesheets and on now submerged land mean that the biogeophysical (albedo, etc.) effects are very dominant. This paper is also in open review, with Climate of the Past. So if you feel the need to sneak an early look at how vegetation affects climate over the last glacial-interglacial cycle, you can do so.

Interview advice for postdocs


There seems to be a lack of advice on the web for postdoctoral posts – those jobs between a PhD and a permanent academic position. All the advice is about academic lectureship positions. There’s also advice about getting onto a PhD program. But the middle ground is forgotten about. Nearly all of us in science will go through a postdoc and are generally unprepared for how to interview for it.

In two years since finishing my PhD, I’ve done six interviews and got three jobs. So I’m going to share some of what I’ve learnt in those interviews and this process. All the usual provisos apply & YMMV.

General Principles, Harsh Realities

Many postdoc positions are already allocated and there is no point in applying or interviewing unless you are the person who the position is intended for.

Sorry. This is the truth. Most universities force professors to advertise jobs, even if they have a candidate they want for the job. This is great if you are that candidate, but a colossal waste of time if you are not (I have been in both positions).

Indentifying jobs that are a waste of your time is actually easy. Usually, the requirements are very specific and include things that seem a bit unrelated. This is done so it is easy to say that the preferred candidate is the right person for the job, because they are a perfect fit.

So it’s imperative to contact the PI before you apply for the job. There is always a contact on the job description. Phone them, or email and ask for a quick chat on the phone. Why phone? So they can be more candid with you. Don’t be afraid to ask questions like, “do you have a particular candidate in mind for this job?”. Or “what outcomes are key for this job?”

The other thing you definitely want is a copy of the grant proposal that the job is funded from, and which work package or similar the job is designed to fulfil. This will help you understand exactly what deliverables the job will actually have.

The application letter and CV

As PhD students we think often that a PI wants to employ a great researcher, and we try to show we are that. Actually, somewhat unreasonably, most PIs seem to want to employ an expert.

Trying to catch a postdoc job in an interview.

Trying to catch a postdoc job in an interview.

So what you need to do in the application (and interview) is draw direct links between the experience you have and what the project aims to do. E.g. “The project aims to do X, and I have experience doing X though blah.”

Don’t be afraid to talk about their project. In fact, showing that you’ve read about the topic, the project and them, will help you a lot. They want someone with enthusiasm, that they can trust to be engaged and get on with the work.


The interview itself

Most interviews I’ve done have had three components: a short talk, questions about the talk, then generic questions.


The talk

  • Read what the guidance says about this talk carefully. Does it say, ‘your work and your fit to the project’? Then give 50/50 to each aspect.
  • Don’t overrun. Really. They aren’t timing it usually, but everyone knows when you’ve overrun and is grumpy.
  • Don’t talk about yourself too much. Honestly, they’re not interested in you, they’re interested in what you can do for the project. Harsh but true.
  • Don’t be tempted to go into any detail. The job of this talk to convince them you’ve done some interesting work and you’ve published it (or will publish it). The talk I bombed the worst was when I tried to give a mini summary of several bits of work I did. It overran and was a disaster.
  • Do talk about the project, how you would do it, and what aspects you’re particularly interested in.
  • Do take questions and go on tangents. This is their opportunity to get to know you – show them that you’re flexible and can talk about what they are interested in.

Questions after the talk

This is their opportunity to ask for more details about you and your work. If you’ve done your job properly and intrigued them, this bit should be fun.

Generic questions

In the UK it seems that they have to ask the same questions to each candidate. So expect very generic questions about how you would do the project, and questions that check your basic knowledge of the subject area. Expect and prepare the following:

  • Generic questions you would ask to check basic and advanced knowledge of the subject area.
  • Conventional academic interview questions. They’re a chance to really shine. Things like:
    • What are your strengths/weaknesses
    • What work you are most proud of?
    • How would you manage if this critical thing failed?
  • What are the potential pitfalls of the project and how would you manage them?

At the end and after the interview

If things have gone very well, they might ask you something like, “Would you take the job if we offered it?” or “When could you start?”. It is absolutely fine to be evasive and put in provisos here. Things like, “I’d be really happy to take the job, subject to discussions about pay and conditions” or “If you were to formally offer me the job, in principle I’d be delighted to work with you.” You’ll note that what these phrases do is sound very enthusiastic and give no commitment at all.

Don’t read too much into how long they keep you hanging on the result. Academics are busy and aren’t always considerate.

If you interviewed, it’s always worth asking for some feedback. Even if you got the job – ask what they liked or didn’t like. You won’t always get a straight answer, but it might be something useful, you never know.

A final word

Good luck with your post-doc interviews! And please give me a shout on twitter if there are other things I should add to this.

CO2 removal – feasibility questions

Does BECCS really cut it when you’re aiming to limit climate change to 2 degrees, or even just 1.5 degrees? In Nature, Phil Williamson argues the answer is “no”, and uses work by Andy Wiltshire and myself on the Planetary limits to BECCS Negative Emissions to show why.

The future is unknown for the effect of BECCS

BECCS is BioEnergy with Carbon Capture and Storage, and is included in many projections of future climate change that stay relatively low, not just those which aim to remain under 2 degrees warming.

Williamson highlights work Andy and I did, which is freely available to download from the AVOID project website, which shows that when you include the CO2 released by the land-use change, the net CO2 impact of BECCS begins to look rather different. In fact, in its worst case scenario, it can even result in more carbon being released into the atmosphere than is sequestered.

It all depends on the details of when and where the deforestation for cropland occurs, and how long the land is used for bioenergy crops. In essence, the longer land is used for bioenergy crops, the more likely it is the ‘carbon debt’ from the deforestation can be ‘paid back’. In the case of ‘abandoned agricultural land’, there is no debt to be paid – it was already cropland. But in a scenario where forest is removed to make way for BECCS, there is a considerable carbon debt to be paid back, before the BECCS actually makes a positive contribution to mitigating climate change.

However, recent work published in Science shows the real-world effect of what models have always shown: the non-carbon effects also matter. So where the figure above shows carbon emissions from large areas of forest converted to cropland and no agricultural productivity growth, adding in the non-carbon effects changes the story slightly. Because of more than half a degree of cooling from albedo effects, that scenario actually gives a small net cooling, if you’re only considering those land conversions as food crops. If they were BECCS, then probably the cooling would be even larger.

Now, I’m certainly not advocating chopping all the trees in the world down to grow BECCS and limit climate change. The biodiversity impacts would, in my opinion, be utterly unacceptable. But neither am I keen on ignoring biogeophysical effects, just because they don’t fit neatly into our carbon accounting.

Partly though, this is a failing of our science. We need clear metrics that can be used in climate change scenarios. There are ways that the biogeophysics could be fully enough represented that it could go into these sorts of policy decisions. Hopefully it won’t be long before I will be able to update you on how that aim is progressing.

Beyond Land Use Change

There is a growing need to address climate forced changes to the land cover, as well as looking at land use change.

Our paper comparing Climate Induced Land Cover Change (CILCC) to LULCC (Land Use Land Cover Change) in the Land Use special issue in Global Biogeochemical Cycles compares the changes in land cover from the changing climate (as carbon dioxide increases) and from land use change (human land requirements). We found that despite LULCC being a main research topic for terrestrial climate scientists, actually the neglected topic of CILCC was much more important in terms of land area affected and carbon emitted. We suggested that if how much research is done on a topic is influenced by how much affect it has on the climate, that we should look at CILCC more than we do LULCC. At present, LULCC is a much more popular topic than CILCC, so this is a call to arms.

That’s just one paper, but there is a gathering zeitgeist about this subject. In the same special issue, Schneck et al. look at how anthropogenic LCC (LULCC) can be offset by natural land cover change (aka CILCC) on the millennial scale. Again, they show that climate forced changes to vegetation are important.

Earlier in the year, Ahlstrom et al published a paper in Environmental Research Letters showing that dynamic vegetation (aka CILCC) is a significant part of the terrestrial carbon cycle.

Then a couple of weeks ago, Almut Arneth had an opinion piece in Nature highlighting the complexities of the patterns of land use that emerge from human and climate drivers and calling for better accounting for land cover change in modelling.

The terminology for discussing this isn’t quite ‘set’, so it’s more challenging to see how these papers are connected. The use of the term ‘natural’ and ‘anthropogenic’ for CILCC and LULCC is problematic since it infers that the climate forced changes are natural, when in fact they are forced by anthropogenic climate change. I got argued down on anthropogenic too, as it infers that the other isn’t anthropogenic. I like the simplicity of natural vs anthropogenic and completely understand why others have used it. But in the end, I created the term CILCC because it fitted so nicely with LULCC (which has become a standard term).

Whatever you’re going to call it (obviously I’m in favour of CILCC), this is topic that we can’t afford to ignore any longer.

Land Cover Change Quiz


Land cover change is usually a forgotten suffix to Land-Use, in LULCC (Land Use Land Cover Change) which refers to how the land cover changes when the land use alters.

But the land cover can be different from climate changes too. Under climate change, the distribution of plant types alters as for instance a region warms, or the water availability decreases. Climate Induced Land Cover Change (CILCC) is projected using a dynamic vegetation model that determines what plant type is likely to grow in a given place.

The Representative Concentration Pathways (RCPs) are a set of projections of future socio-economic scenarios that lead to varying levels of climate change by 2100, used in the latest IPCC report. They range from low levels of overall climate change (RCP2.6), mid-range (RCP4.5), to high (RCP8.5). The RCPs have different LULCC assocaited with them, which have been extensively assessed.

By comparison, CILCC in the RCPs has been more or less ignored.

We have a new paper in Global Biogeochemical Cycles that assesses whether we should be ignoring CILCC and focusing on LULCC. And just for fun, we’ve put some of the results into a quiz.


Take our quick quiz to see what you know about how CILCC compares to LULCC in the RCPs.

Q.1 Between 2005 and 2100 in RCP2.6, 4.5 and 8.5, is the area of land altered larger from LULCC or CILCC?


rcps Land use and climate

Amount of global land cover changed from Land use, and from climate changes, in the RCPs.

In all three RCP scenarios that we looked at, the area of land changed was larger from CILCC than from LULCC. This is despite LULCC having had considerably more attention in recent research than CILCC.

Q.2 Of RCP2.6, 4.5 and 8.5, which has the largest net global loss of trees at 2100 compared to 2005?



The loss of global forest in the RCPs, from Land Use Land Cover Change (LULCC), Climate Induced Land Cover Change (CILCC), and both together (Net forest change).

The largest net loss of forest (including CILCC and LULCC) occurs in RCP2.6. Despite the LULCC in RCP8.5 being very similar to RCP2.6, the CILCC in the higher climate change scenario means that 90% of the deforestation in RCP8.5 is compensated by boreal forest expansion from CILCC.

Q.3 What proportion of overall climate induced terrestrial carbon changes (2005 – 2100) are attributable to CILCC? Is it: a) 6 -12% b) 14 – 22% c) 66 – 80%


b. CILCC accounts for 14 – 22% of total terrestrial carbon changes over the coming century. LULCC accounts for 6 – 12%, and the accumulated carbon (not in areas of land cover change) accounts for the rest – up to 80% of the total carbon change.

Although the climate effects of LULCC has been a popular research area in recent years, CILCC and terrestrial carbon accumulation are arguably more important to the climate.

Q.4 How many of the CMIP5 models includes a) the terrestrial carbon cycle and b) how many of those include dynamic vegetation (to model CILCC)?


Of the 39 CMIP5 models, only 18 include the terrestrial carbon cycle, and we only know of 3 that include dynamic vegetation (and are therefore capable of projecting CILCC). This is a major shortcoming from both a carbon and ecosystem perspective.


How did you do?

0/5 Well, never-mind. Perhaps ocean biogeochemistry – or like, knitting – is more your thing.

1/4 Good start. You may need more work on memorising awkward acronyms (an important skill for this field).

2/4 Half-way to knowing something about land cover change. A little more dedication and you will be arguing about whether secondary land use effects and land management are key uncertainties with the best of them.

3/4 Didn’t you do well? Perhaps you read the paper? Or are you just naturally talented? Well done either way.

4/4 An excellent effort. Clearly you are a dedicated terrestrial carbon scientist, keep it up! (Are you sure you aren’t one of the co-authors? Or did you cheat and look at the answers? Though I suppose that those two aren’t mutually exclusive….)

5/5 Ah, you count both part a and b on Q4? Yeah, I totally agree. And if you got it all right then that is superb skills!


Share the images below to show the world how you did:

land cover quiz passed badge

I did the Land Cover Change Quiz and PASSED.

land cover quiz failed badge

I did the Land Cover Change Quiz and FAILED.


Afforestation gives a net global climate warming


This paper started off with a simple question: is the amount of land use change a good guide to how much net climate effect it will have? But it came up with a different sort of answer: that afforestation is not the panacea that we might hope it would be, and may even be a perverse incentive when it comes to climate.

You might expect that the amount of land use change would be a good guide as to how much it affect the climate, i.e. more land use change, more change to climate, less land use change, less change to climate. But this isn’t the case. The carbon emissions are approximately proportional to the land use change, but there are other factors – changes to the surface reflectance (albedo) and the evapotranspiration – that are not so linear. The result of these other effects are that in one of the IPCC projection of future climate change, the afforestation policy implimented to help mitigate climate actually causes warming rather than cooling. The warming isn’t very much, only about 0.1 K in 2070 to 2100, but it is enough that the afforestation isn’t doing what it was intended to do. However, extending and preserving forest would have a host of other benefits to biodiversity and ecosystem services. Before we chuck out afforestation as an idea all together because it may not benefit the climate, it’s important to think about other things that forest can provide.

If you’re interested in reading more, wander over to the full paper, published in Environmental Research Letters: Full effects of land use change in the representative concentration pathways. (It’s open access, so everyone can read it. Yay!) You can also engage me in discussion on twitter: @tdaviesbarnard

Ensemble members

Initial condition model ensemble members are rather like leaves. If you only have one, you know exactly what a leaf looks like. But if you have more, you’re not quite so sure what a leaf looks like, but maybe you can begin to understand the tree…

Children’s Guide to a Kindle

I made this guide to the Kindle e-reader for a girl I gave a Kindle to as Christmas present. I could’t find a good, simple guide to using a Kindle anywhere on the web, so I made one. The parents really liked the guide, so I thought I’d post it up here and if you’re giving a kid a Kindle, you might like to print it out and use it.

There are two versions of the Children’s guide to Kindle. A docx version (editable) and a pdf version (not editable).

It’s A5 size, so it’s nice printed on A4 as a booklet. To do this is a bit of a hassle, but the steps are basically:

  1. screenshot_90Amend the docx guide as desired.
  2. Save the docx as pdf, either using ‘save as’ or print -> save to pdf
  3. Open the pdf in Adobe Reader or Adobe Acrobat. (Adobe Reader can be downloaded for free).
  4. Press print, and then find the booklet option (it isn’t particularly easy to find. Here is some help.)
  5. It should have printed all the pages in a weird order. There are page numbers to help you check it’s correct. Put them in the right order, fold, staple, and voila, a beautiful booklet to go with your gift.

A couple of things to bear in mind.

  • I registered an email address and an Amazon account for the child, in my name. Accounts shouldn’t be in the name of the child if they are younger than 13. Child protection, etc.
  • If I’d been more efficient, I would have registered the account, then bought the Kindle from that account, so that it was pre-registered.
  • As it was, I had to open the Kindle box, register the Kindle, hook it up to my wifi and download all the (free) books I had downloaded on the child’s behalf. This meant that the Kindle box was open when it arrived. This particular child wasn’t worried by this, but others might be.
  • All this worked well, but was quite a bit of work for me.
  • This Kindle guide was written for a young child (about 8). It might be a bit patronising for adults or teens.
  • The guide uses a couple of images of a Kindle. I don’t own these images. They probably came from Amazon, but if I’ve infringed your copyright and you’re upset, contact me and I will sort it asap.

IPCC colour scheme for the RCPs


For those who work with the RCPs (the Representative Concentration Pathways), it seems like colour schemes should be simple. Same set of simulations – same colours. Easy. But it’s not – almost every paper I’ve read uses a different set of colours.

So… the IPCC report (WG1) should have sorted this out. It did, mostly. *Most* of the plots use this scheme:


RCP2.6 RCP4.5 RCP6.0 RCP8.5
Hex #0000FF #79BCFF #FF822D #FF0000
RGB 0 0 255 121 188 255 255 130 45 255 0 0
Looks like: Royal Blue Light Blue Orange Red


It’s provided here for your convenience in hexadecimal and RGB.

You might not like the colours they used. Like me, you might think that the use of the same diverging colour scheme for maps and RCPs alike is a bit controversial (given the non-linearity of almost anything that’s not watts per meter squared in the RCPs). But maybe, hopefully, we can all accept and use this colour scheme for plots and publications using the RCPs. It makes it so much easier for the reader. And we can do something more productive with our time than puzzle over which colour scheme to use. (I am a big fan of colorbrewer for this purpose.)

Hat tip to Ed Hawkins for helping me with this.

AGU 2013 canopy interception capacity


If you missed seeing me at AGU, or just would like closer look at the work I’m doing at the moment, you can download a copy of the poster.

Normal provisos apply – this work is submitted to a journal, it is yet to be subject to peer review, etc. etc.

If you’d like to know more about this work, please do contact me.