I was lucky enough to have two papers published in the last month. The second was the final part of my PhD, so publishing this paper is the real end of my studies. It’s also probably the most challenging paper I wrote, so I thought I would talk about the process a bit.
A long journey
From the beginning, understanding the relationship between ECLs and the Great Dividing Range next to the coast was going to be one of the major points of my PhD. My first model simulations were done three years ago, when I was playing with changing topography for a single month in my regional model, and my first presentation about it was all the way back in June 2015.
So why did it take so long to publish anything? Part of it was getting distracted by other shiny things to study, but it was also due to my growing understanding of what we should and shouldn’t do with regional models - that is, models that are run at high resolutions but over small areas of the globe.
My early simulations were for a single month, which doesn’t really tell you much about changes in the frequency of ECLs. So the next step was running the model over 2 years, just like I did with changes in sea surface temperatures. But after doing all that, I got asked some really important questions by my supervisors.
Simply, what I had been doing for sea surface temperatures was “nudging” the model fields in the upper atmosphere to be more like the global observations. This is good to make sure the broader atmosphere is the same, so we can compare individual events. But unlike oceans, mountains extend well up into the atmosphere, which means that this “nudging” is actually stopping the model from properly incorporating the changes.
So, more model simulations, that had to run for a lot longer. As part of this, I made a mistake setting up one of my model simulations which I didn't realise til it was done, so I had to do it all over again, wasting another month.
Then, finally! My results were analysed and I could write & submit my paper and finish my thesis. Yay!
… about a month before I got my thesis results, the paper was rejected. :(
It was too messy - I had tried to include all my different types of simulations, and the results just weren’t very clear. In the end I had to throw out almost everything - all of those 2-year simulations, the extra runs I did because I had started them wrong, everything except a couple of my long simulations. All those thousands of hours of supercomputer time and countless hours of analysis, for nothing. It was quite depressing, especially since I hadn’t had a rejected paper in half a decade.
But I got over my sadness and rewrote the whole paper from scratch, to create something much clearer. It was hard, but worth it when I resubmitted and the reviewers accepted it with only minor comments. Hooray!
So, what’s the moral of the story?
But what was the paper about?
Climate models tend not to have enough ECLs near the coast, so I wanted to know how important the Great Dividing Range is, since it basically doesn’t exist in a global model that has 200 km between points. So I ran my regional model over Australia & surrounds with the height of all the land set to 0m.
And what did I find? Topography actually isn’t as important as everyone thought. Removing it means we get fewer ECLs coming from the Bass Strait, especially in spring, but the number that form near the coast actually increases! There’s also no change in the number of strong events, although the wind speeds and rain on the coast are lower because they’re not being forced up by the mountains.
A long journey
From the beginning, understanding the relationship between ECLs and the Great Dividing Range next to the coast was going to be one of the major points of my PhD. My first model simulations were done three years ago, when I was playing with changing topography for a single month in my regional model, and my first presentation about it was all the way back in June 2015.
So why did it take so long to publish anything? Part of it was getting distracted by other shiny things to study, but it was also due to my growing understanding of what we should and shouldn’t do with regional models - that is, models that are run at high resolutions but over small areas of the globe.
My early simulations were for a single month, which doesn’t really tell you much about changes in the frequency of ECLs. So the next step was running the model over 2 years, just like I did with changes in sea surface temperatures. But after doing all that, I got asked some really important questions by my supervisors.
Simply, what I had been doing for sea surface temperatures was “nudging” the model fields in the upper atmosphere to be more like the global observations. This is good to make sure the broader atmosphere is the same, so we can compare individual events. But unlike oceans, mountains extend well up into the atmosphere, which means that this “nudging” is actually stopping the model from properly incorporating the changes.
So, more model simulations, that had to run for a lot longer. As part of this, I made a mistake setting up one of my model simulations which I didn't realise til it was done, so I had to do it all over again, wasting another month.
Then, finally! My results were analysed and I could write & submit my paper and finish my thesis. Yay!
… about a month before I got my thesis results, the paper was rejected. :(
It was too messy - I had tried to include all my different types of simulations, and the results just weren’t very clear. In the end I had to throw out almost everything - all of those 2-year simulations, the extra runs I did because I had started them wrong, everything except a couple of my long simulations. All those thousands of hours of supercomputer time and countless hours of analysis, for nothing. It was quite depressing, especially since I hadn’t had a rejected paper in half a decade.
But I got over my sadness and rewrote the whole paper from scratch, to create something much clearer. It was hard, but worth it when I resubmitted and the reviewers accepted it with only minor comments. Hooray!
So, what’s the moral of the story?
- Put more thought in the best way to do your model simulations before you start, so you don’t have to redo them all
- Just because you spent a lot of time on something, doesn’t mean it should be in the paper. Learn to let go
- As always, academia is about persistence. It may feel depressing when a paper gets rejected, but it’s not a reflection on you and there’s always another chance
But what was the paper about?
Climate models tend not to have enough ECLs near the coast, so I wanted to know how important the Great Dividing Range is, since it basically doesn’t exist in a global model that has 200 km between points. So I ran my regional model over Australia & surrounds with the height of all the land set to 0m.
And what did I find? Topography actually isn’t as important as everyone thought. Removing it means we get fewer ECLs coming from the Bass Strait, especially in spring, but the number that form near the coast actually increases! There’s also no change in the number of strong events, although the wind speeds and rain on the coast are lower because they’re not being forced up by the mountains.
That seems like a lot of work for not much result, but it’s actually quite important to know that we shouldn’t focus so much on the Great Dividing Range when it comes to ECLs - it’s much less important than the warm oceans near the coast. So future scientists (or future me) can move on to other questions, like what would happen if we wiped out New Zealand? I think that would be super interesting!