A –Streams of Thought– contribution by Wouter Knoben (WK)
Francesca Pianosi is a senior lecturer at the University of Bristol. She currently holds a prestigious Early Career EPSRC “Living with Environmental Change” Fellowship and was awarded the EGU Arne Richter award for Outstanding Young Scientists in 2015. Her research focuses around uncertainty and sensitivity analysis, and water resources management. She is the lead developer of the SAFE Global Sensitivity Analysis toolbox (Matlab/R/Python: www.safetoolbox.info).
WK: Can you tell us a little about your background, your formal education?
I did an MSc in Environmental Engineering at the Technical University of Milano and stayed there to do a PhD in Informatics Engineering. This is essentially the engineering version of computer science. The department had people working on a wide variety of topics but I was part of a small group inside it that applied mathematical theory to environmental problems (my PhD project focused on water resources modelling). There were three professors there working on atmospheric systems, population dynamics and water systems respectively, so we used to say “Air, Animals and Water are covered” (we missed Earth!). On the one hand it was very nice to be part of such a varied department, because I got exposed to many different topics and that is good for building confidence. On the other hand, I would sometimes end up in seminars about stabilizing space rockets during landing, which was not really directly useful for my work!
WK: What was it like doing an environmental PhD in such a research group?
That was three years and included classes on many different topics: agent-based modelling, Kalman-filters, etc. It was all very heavy on maths and mostly applied to robots and such. It was very interesting and I would have never done this in a Civil Engineering department. There was this one philosophy class that was mostly about human consciousness and the question if and when we can say computers have this too. It was a good place for opening my mind to many different approaches and for building confidence. Now, whenever someone brings up some sort of mathematical approach, I know there’s a good chance I’ll be able to make sense of it. However, when I eventually went to Unesco IHE in Delft for a visiting period, it was a nice change to be part of an environment where everything had to do with hydrology. Every week there would be a seminar on an aspect of hydrology and hydrological modelling and I realized how much being in the “wrong” department can be detrimental to your development of field-specific knowledge.
WK: What research/projects are you currently conducting? What got you started on this current research/project?
I’m currently working on two main topics. The first is quantification and attribution of uncertainty in model outputs, which is a logical result of my post-doc work at Bristol. I use Monte Carlo simulations to propagate uncertainties through simulation models and quantify of the uncertainty in the model output, for instance through a range or a frequency distribution. And then use a set of statistical techniques, called Global Sensitivity Analysis (GSA), to understand which of the input uncertainties mostly contributed to that output uncertainty, so we know where to priorities our efforts for uncertainty reduction and to improve our model. In this context I am interested in developing new methods to perform these tasks more efficiently, but also in discovering new problems to which these methods can be applied. For example at the moment I am very interested in using GSA as a ‘stress test’ tool, to check whether the model ‘behaves’ consistently with our understanding of hydrological systems. This is very good for debugging but also more broadly to evaluate models when we cannot rely on the ‘traditional’ fit-to-data criterion, because data are too sparse or of poor quality. My second research topic is modelling for management of water resources systems, where I combine modelling natural processes with the impacts humans have on these systems – ultimately with the aim of supporting decision-making. This links back to my PhD and earlier postdocs in Milan. After moving to Bristol, it seemed a shame to not do anything with the water resources topic anymore, so I started working on that side-by-side with the uncertainty stuff. Now there is a clear connection between both topics though: how can we use models to support water management decisions when we know modelling results can be highly uncertain? I think the good news here is that decision-makers often only look at few aggregate model outputs to inform their decisions (they need synthesis!) and we know from GSA studies that aggregate outputs are typically controlled by a small number of inputs. So the sources of uncertainty that we should tackle are often quite limited. On the other hand, the bad news is that those uncertainties are often very difficult to reduce, particularly when they relate to assumptions about the evolution of socio-economic processes. So we should really shift our thinking from looking at models as prediction tools towards looking at them as exploratory tools. They can help decision-makers to make sense of possible future scenarios, but we must acknowledge we can hardly say which scenario will ultimately occur.
WK: You’re currently a lecturer. Did you envision yourself in your current position? If so or if not, what is the most important decision you made to help you get there and why?
Honestly, I’m one of those people for whom industry was never a serious option. I went straight from my MSc to PhD to post-doc to the current lecturer position. So it was more or less the plan to become a lecturer but it took me some time to get there. I was relatively old when I decided to leave Italy and try my luck elsewhere. In Italy there was no turn-over at all and no positions were opening up due to budget cuts. This is when I went to the University of Bristol for a post-doc and I got a lectureship two years later. Sometimes I think that the experience of working in industry would be good, but then I meet with the industry partners on my grant and I hear how they have to spend their weeks and suddenly I’m quite happy with where I am! I’m quite suited to being an academic I think and I get enough opportunities for interaction with stakeholders, public bodies, industry partners and such during my research, which I also enjoy a lot. But there are simply a few things in academia that really appeal to me. One thing is that you seem to have more flexibility in setting your own priorities and working hours. I have periods when my mind is very active but there are also times when I struggle to focus. Being able to set your own office hours to accommodate this is very nice, and this also applies to the much longer term: there will be periods in your life when you’re extremely productive and there will be times when the wind goes down a bit. That can be very scary, but it is just something that happens. The other very important aspect to me, is that in academia it’s much easier to justify real long-term thinking. It’s easier to start projects that have no immediate tangible benefit, and might only later start to pay off in your work or that of others. Of course you can’t completely go off on a tangent because you still need to care about grants and publications, but you have a lot of freedom to follow the directions your mind wants to go in.
WK: Where do you find inspiration or motivation for your work?
Walking and conferences. I live half an hour away from my office. In the morning I use this time to plan my day and in the evening it is a moment to unload, process what has happened during the day and return to things that deserve more attention than I could give it at the time. Conferences happen less regularly of course, but they are very inspiring. Learning about new research can be very good but equally I sometimes zone out during talks, stop listening and go over a variety of new ideas in my head. The amount of new information you get during conferences can trigger all sorts of new ideas.
WK: You’ve developed an open-source sensitivity analysis toolbox called SAFE, now available in MATLAB, R and Python. What inspired you to do this?
It was mostly a reaction to frustrations I had from using other code. I had collected certain expectations and opinions about how a good coding product should look, so I wanted to see what would happen if I were to make something. I especially like that we added workflows scripts that each contain example code that performs some form of sensitivity analysis by using the toolbox. They give a very good intro to what the toolbox can do and are easy to change by others. I think they’re more flexible and transparent than a graphical user interface would be. Some of the choices we made turned out to be very effective fortunately and some didn’t. That’s normal. Overall I’m very happy with how it turned out. Making this toolbox also returns to the ability to take a long-term view in academia. There were no tangible benefits to making this toolbox in the short term, but in the longer term it has facilitated many studies by others. It’s very nice to see it being used by others.
WK: Any advice for others thinking about publishing a model, toolbox or coding framework?
I’d say that two things are important. The first is to heavily invest in documentation. Both comments in the code and good documentation are essential. Second, and this comes as a consequence of the first, ask yourself if you have the time to do this project properly. Are you filling an important gap and is it worth spending your time on it?
WK: You don’t have a crazy number of publications, but your work is very well-cited and you have received various prestigious awards and fellowships. What’s the key to success?
I’d say my number of publications is OK but not excessive. Some recent ones are well-cited yes but I got my job in Bristol [as a lecturer] and the EGU award when they didn’t have as many citations yet. I grew up with the whole ‘publish-or-perish’ idea of course, but I think you can build up a reputation that goes beyond your publication record. I think this is often a bit overlooked. Of course, when you finish a PhD it’s good to have a few publications, and you should try to take one or two out of any post-doc positions you hold otherwise people might wonder if anything went wrong. But if you get one or two well-written and well-cited publications, those can matter as much as four or five rushed ones. There’s a trade-off between quality and quantity. Hydrology is a fairly small world and other people will be able to keep an eye on what you’re doing. For example, for every presentation I make my slides from scratch and I think that makes the quality of the presentation higher and people will notice this. At some point I realized I had somehow built this reputation of being thorough and well-prepared. So take your work seriously, be prepared and generous and do the best you can. It’s really a small world, and if you aim for consistent quality in your work people will notice.
WK: In 2015, you received the Arne Richter Award for Outstanding Young Scientists from the EGU? What was that like?
Of course I was surprised, I didn’t expect it and I was happy of course. This was one of the occasions where I realized that your reputation is important, because this was before any of my currently top papers had attracted much attention (or been written). Still, various people had apparently thought I’d done my job well and provided the initial support letters for this award. I really enjoyed giving the lecture. It was one of the first opportunities to give my opinions and feel confident about having the right to speak them. It wasn’t like I was a PhD student reporting on some assignment that I had done well, but I felt entitled to speak my opinions and outline where I thought we could do better and how. That was very empowering, I liked that. You know, any award is a mix of deserving it and luck. I’m sure many other people that year were worth the award as much as I was. It’s important to do your job at the best of your capabilities but it also important to have people who help and support you. In order to get an award, you need someone to write a nomination letter for you in the first place! When I was giving my lecture I was wondering at some point how many people in this audience would be worth standing up here but nobody had bothered to write such a letter for them.
WK: What advice can you offer to individuals just starting their (research) careers?
Think of the long term, even if the system pushes you to go for quick and quantity. Quality will be apparent even if it takes longer to get there. And don’t think you can or should do everything alone. Look for supportive environments, where you can find the assistance you need. And finally, at some point try to understand what you’re good at and what you’re not so good at. Of course try to improve the things you can, but focus on your strengths and find people to collaborate with, who can compensate your weaknesses.
About the author
Wouter Knoben is a postdoctoral fellow at the University of Saskatchewan’s Coldwater Laboratory in Canmore, Canada. He is currently active as web editor for the Young Hydrologic Society. Correspondence to firstname.lastname@example.org