Just one thing…
Ten years ago in Greenland, I did not imagine I’d still be doing this job-I found myself very much nodding along with the sentiments of the piece, the conceit that
This part also made me laugh in recognition about what makes people persevere in research:
Lesson 1. Every day has to count for something
exhausted you are by non-deep junk hours, the less you progress.
This was a curiously freeing part to read and I absolutely agree with it. Too often what John Kennedy calls Natureorscience papers are seen as the gold standard. And yet as Daniel Nettle eloquently points out:
order to defend its exclusive market position. If we all send everything there, the rejection rate goes up. If we all increase the quality of our science, it still nearly all gets rejected, by the very design of the institution. The idea that all good papers can be in Nature or Science is as ludicrous as the idea that all Olympic athletes can get gold medals, but without the strong link between actual ability and finishing position that obtains in the Olympics.
For what it’s worth though, I do believe this:
Or so I tell myself, admittedly through clenched teeth at times.
Is back to just one thing in a way.
This is why I do field work. But it’s also why I’ve embraced the opportunity to learn more about deep learning and AI/ML methods. Learning new stuff is exciting, it keeps you fresh and helps make new connections. It’s when disciplines cross-connect that the exciting stuff happens and the sparks fly in the brain.
Learning to fly a drone and to process the data is something I’ve been working on the last few years. I have a really exciting dataset now but little time to work on it. Ifyou’re looking for an interesting MSC thesis project get in touch!A note of caution though, it’s always easier to start something new than finish an old project. The best colleagues will help you stay on track and make sure you finish what you started!
I’m going to add one more point, which isn’t expressly mentioned in the original piece that started this ramble:
Lesson 5: Cultivate outside interests.
Far too many of us put families, friends, sports, hobbies and anything else that doesn’t taste of work to one side, in pursuit of the all-consuming. It’s not only not healthy, it’s also limiting. The brain needs time off to churn away by itself. You can’t force that unconscious process. Better to take a long walk to admire the flowers than try to twist your brain in knots when you hit a wall. A good night’s sleep is an amazingly effective part of the research process too.
So there we have it, some thoughts on being a (mid-career) scientist and how I have managed to stay in the game. YMMV as the Americans say.
Finally, all that I have said relies on having a supportive employer and good colleagues. The sometimes horrifying stories (take for example this one) of people being pushed out by bullying colleagues, or structural discrimination is a whole other story. And not one I’m going to take on here, but I would point out that without organisation, labour inevitably gets crushed by capital, so organise, join a union, find out what your rights are and make sure that you have a supportive hinterland to help you get through the bad times.
And everyday, do just one thing to help you advance.
[#]30DayMapChallenge #AcWriMo #blogging #job #Jobs #People #Science
=> View attached media | View attached media | View attached media
=> More informations about this toot | View the thread
Looking backwards…
This is the first in a two-parter. At this time of year, posts making bold statements about what happened last year and what we plan to do this year start to become prominent. The last few years I have spent a few hours in the first week of January reviewing what worked, what was fun and what was cool, what was awful and what definitely was a waste of time. I’m not honestly sure that any of this is of interest to anyone except me, so read on, but you have been warned..
2024: Themes of this year: Greenland, Machine Learning, people, and big data…
I visited the world’s largest island 3 times this year – a rather unprecedented number of times for me, with fieldwork in April (it was very cold and there was a lot of snow) to continue a soon to be submitted for publication set of observations in the melange zone and then to establish a new snow observation site.
View from Qaanaaq at evening in early April 2024. In late May and early June, after a slightly longer than expected stop in Ilulissat, we made it to bring in the instruments before the sea ice break-up and happily my new snow observations seem to be working. Now I just need to do set-up the data processing chain, which will be 2025’s paying myself first.
Working with scientists from the Greenland natural resources institute and local hunters on the sea ice. The final trip was in October for a workshop with scientists in Greenland about climate change impacts in Greenland, the subpolar gyre and AMOC for the UN Ocean decade. It was a memorable meeting for the sheer range and quality of science presented as well as for being stranded in Nuuk by a broken aeroplane in quite ridiculously beautiful weather (I mostly stayed in my hotel room to write the aforementioned paper, sadly. In 2025 I will work on my priorities) .
Apart from fieldwork I have really tried hard on publications this year. I have (like many scientists I suspect), far more data sitting around on hard drives than I have published. It’s a waste and it’s also fun to work on actual data instead of endless emails. This is something I intend to continue focusing on the next few years as well. There is gold in them thar computers…
We had a couple of writing retreats were very successful. These I plan to continue also and the PRECISE project grant is happily flexible enough to do this. I probably achieve as much in terms of data processing and paper writing in 3 focused days as I would in 3 months in the office. It paid off too. I managed to co-author 8 papers published this year (including my first 1st-author paper in ages – a workshop report, but nevertheless it counts.). Some of these are still preprints, so will change, and there are a couple more that have been submitted but are not yet available as preprints. I will submit two more papers in the next 3 weeks as well (1 first author), so January 2025 is going to be the 13th month of 2024 in my mind.
Bootcamps have been a theme the last 3 years, I organised the first in 2022 and so far there have been 4 publications from that first effort. There was another this year in June, ( I have attended them in 2023 and 2024 but was not organising) where we really got going on a project for ESA that I have had my eye on for a while – I hope the publication from that will be ready in the Spring this coming year.
Machine Learning: This was the year I really got machine learning. I’ve been following a graduate course online, and learning from my colleagues and students about implementations. I understand a lot more about the architecture and how to in practice apply neural networks and other techniques like random forests now. This is not before time, as we intend to implement these to contribute to CMIP7 and the next IPCC report. We still have a lot of work to do, but the foundation is laid. And it’s been fun to learn something that, if not exactly new, is a new application of something. In fact the biggest barrier has really been learning new terminology. We have also been fortunate that Eumetsat and the ECMWF have been very helpful in providing us with ML-optimised computer resources to test much of these new models out on. We’re actually running out of resources a bit though, so it’s time to start investigating Lumi, Leonardo and the new Danish centre Gefion to see what we can get out of these.
People: This year our research group has grown with another 2 PhD students, and at the end of the year we also employed a new post-doc. I think it’s large enough now. I’m very aware that if I don’t do my job properly, then not only the research but the people will suffer, so developing people management skills is really important. In any case it’s extremely stimulating to work with such talented young people and I’m really excited to see where the science will take us, given the skills in the team. I hope I have been good enough at managing such a large and young team, but I have my doubts. A focus for 2025 for sure.
Data: This has been the year of big data, not necessarily just for ML purposes but also in the PolarRES project the production and management of an enormous set of future climate projections at very high resolution. More on this anon. Suffice to say, it has taken a lot of my time and mental energy and it’s probably not the most exciting thing to talk about, but we now have 800 Tb of climate simulation data to dig into. I suspect that rewards of this will be coming for years. There has also been a lot of digging into satellite datasets and the bringing together of the two has been very rewarding already. It’s a rich seam, to continue the metaphor, that will be producing scientific gold for many years.
Projects: we have gone in the final year of two projects, PROTECT and PolarRES, both of which will finally end in 2025. We also arrived at the half way point of OCEAN:ICE. So rather than being a year of starts, it has been a year where we have started to prepare for endings – actually this is a fun part of many projects where a lot of the grunt work is out the way and we can start to see what we have actually found out. It can also be a slog of confusing data, writing and editing papers and dealing with h co-author comments. I’ve definitely been in that process this year, hopefully with some of the outputs to come next year…
Proposals: I started 2024 writing a proposal. Colleagues were in 3 different consortia for the same call, alas ours didn’t get funded, but 2 of the others did and will start this year. That is a good result for DMI and our group. I wrote another proposal in the Autumn and contributed to a 4th and finally at the end of the year I heard that both will likely be funded (but are currently embargoed and in negotiation, so no more will be said now). It sometimes feels that spending so much time and energy on proposal writing is putting the cart before the horse, but in fact I find proposal writing something akin to brainstorming. It’s essential of course to ensure we can continue to do the science we want, but it can also help us to clarify our ideas and make sure we’re not on the wrong track. It’s also a good way to keep track of what the funders are actually wanting to know and to help us focus on policy relevance.
There was also an incredible number of meetings, reports, milestones and deliverables, but you probably don’t want to hear about that…
Also missing from this summary is personal life, and, well that is not for sharing publically, but suffice to say, I learnt about raising teenagers, I also had some very good times with friends and family, to all of whom I immensely grateful for being a part of my voyages around the sun.
Anyway, reading all that back, I’m not surprised I ended the year exhausted! I am not planning on quite such a slog in future. I should probably pace myself a bit more this year, the plans for which will be the subject of next week’s post.
[#]climateChange #DMI #fieldwork #Greenland #GreenlandIceSheet #Science
=> View attached media | View attached media
=> More informations about this toot | View the thread
Yesterday in 30 Day Map Challenge I rather hurriedly made a map showing the density of street trees in Copenhagen shown as hexagons. However, there is a big gap in the overall map, because the dataset I used only covered Copenhagen Kommune (local authority) area and Frederiksberg is a separate local authority area where I could not find the data. This was, to put it mildly a little irritating.
A fellow mastodon user (@tlohde) suggested using the outputs from openstreetmap to fill out the gaps. (And even helpfully provided some code to do so, which should tell you a lot about why I like mastodon so much). A very hurried 10 minutes reprocessing gives the revised map on the below, which has happily filled in much of the Frederiksberg gap. However, a closer comparison with the previous version above shows that, it’s not nearly the same…
The first thing to note is that the maximum number of trees in a polygon from the OSM data is 454, almost twice the 230 from the Copenhagen city council data set. The second thing is that I’m unsure exactly what time periods the Copenhagen data is from. It’s possible there has been a wholesale planting since the original data was collected, but there is no date on the opendata.dk page to indicate when it was sampled, so I can’t know how up to date it is. Openstreetmap may also be missing data of course (and a small remaining gap in northern Frederiksberg suggests it might be). However, the whole central axis of the plot has changed too.
I overlaid the individual trees on the map plot, the two are quite similar, and the long lines suggest tat plantings are following major roads in the city. I wonder however if the main difference is one of definition. Perhaps street trees from the Copenhagen kommune dataset does not include parks and of course those on private property, compared to those in OSM?
Does it really matter? Well maybe. Street trees provide a valuable service in communities: they shade the streets in hot summer days (and can lead to substantial cooling). They also soak up rainwater and their flowers and fruit feed city ecosystems, quite apart from their aesthetic properties. How to protect, conserve and expand the numbers if we don’t know where they are? Or are not for that matter?
I don’t really have time to dig down into this mystery further. 30 Day Map challenge is really about the tools but either way it’s a lesson. No matter how clever the tool, if the underlying data is missing, wrong or otherwise biased in someway, the map will also be wrong.
I’m tempted to add, that all maps are wrong, but some of them are useful..
https://sternaparadisaea.net/2024/11/05/the-curious-case-of-the-moving-trees/
[#]30DayMapChallenge #maps #Nature #Science #wildlife
=> View attached media | View attached media | View attached media
=> More informations about this toot | View the thread
The personal finance community have an important concept of “paying yourself first”*, by which they mean, that when your salary or other form of payment comes in, the first thing you should do is put a given percentage, 10% is commonly used, into a savings account. Only then should you consider spending the rest of your income.
I kind of like this as a concept, and I think it could very usefully be applied to other areas of my life, notably, which is where of course it comes into this blog, science. As I’ve got more senior I’ve found I’m spending more and more time on managerial tasks, meetings, emails, reports, proposals, supervision and less and less on actual science. This is probably fine, it’s the way of the world, but it’s also a pity when part of (most of?) the joy of science is really in the doing. That’s why we put up with paltry wages, high workloads, social media hostility and the rest.
Actually doing science is so much fun.
Admittedly, some of it is more type 2 fun (best enjoyed retrospectively, as anyone who has spent a month CMORising model output or digging snow pits in freezing driving snow conditions can tell you), than type 1 fun (enjoyed in the moment). Nonetheless, I occasionally feel I’m in danger of losing the thread of why I started in this career in the first place.
Type 2 fun: It took us 4 hours to locate and dig that lot out in wind and occasional blizzard conditions. Autumn was absolutely and ridiculously hectic, many project meetings, as well as technical conferences and symposia, proposal deadlines, deliverable deadlines and one-off workshops. I welcome November with open arms. Finally time to do some actual work again! And in the way of paying myself forward, I have started two different but related tracks to get back into the groove this month.
The first, #30DayMapChallenge you can already see some entries for here on a dedicated page. The idea is a new map, according to the prompts from the website 30DayMapChallenge , every day. I’m certainly not going to make all 30. I will be doing well if I manage 10, but already after only 2 days, I can feel my geospatial mojo coming back. There’s nothing like practicing your GIS skills to make you want to do more of them
The second is #AcWriMo, academic writing month. I have 3 papers I’d really like to submit before the end of this year. I’m very close with one, fairly close with the second and to be entirely honest I’m not really sure where I am with the third… Now it may seem unwise to commit to 2 daily activities in November, while recovering from September and October, but in fact they’re pretty complementary. I plan to post maps that are relevant to, or even actually from the papers, and just the process of looking at data is a motivation to get the work done.
So my commitment to #AcWriMo is:
https://sternaparadisaea.net/2024/11/02/paying-yourself-first/
[#]climateChange #Science
=> View attached media | View attached media
=> More informations about this toot | View the thread
=> This profile with reblog | Go to sternaparadisaea.net@sternaparadisaea.net account This content has been proxied by September (3851b).Proxy Information
text/gemini
