Bob Plant: Evaluating and improving high-resolution high-impact weather numerical weather forecasts

My name is Bob Plant. I’m an academic at
the University of Reading and the project is to do with high resolution,
numerical weather forecasting, especially with a focus on extreme or high impact
events. So to give a little bit of context with that, what we mean by high
resolution weather forecasts is one where the grid spacing in the model
might be in the order of about a kilometre or so. That’s quite an important scale
actually because we’re especially interested in convective storms that may
have a kind of physical scale in the real world of a few kilometres, so one
kilometre is kind of what you need to be able to represent these things properly.
They’re important because the convective storms are often what drives the really
high impact weather, so events such as flash flooding or extreme wind gusts, are
very often associated with these storms on those kind of few kilometre
scales. It’s relatively new to be able to do this, to be able to run the models at
these scales, so it’s perhaps a decade or so that operational centres have been
trying to do them and they’ve been sort of slowly coming on stream over that
time but the field is starting to get a little bit more mature now. So, not very
long ago, everyone was just still very excited that you could run these things
at all and start to see these storms and then, you know, just really thrilled
to look at the outputs. But we’re starting to get to a point now
where, you know, we’re less excited about the whole thing and
much more looking into the details and much more aware that these things are
very useful but far from perfect in some ways. We started to look more into what other deficiencies are there. So that’s the kind of situation that you
would be walking into and what we’d like to do is to kind of really start to
address those deficiencies properly now, to understand scientifically where
they’re coming from and and how we can improve those. So there are two in particular
that we were interested in looking at for this project. One is to do with the
spatial structure of the storms, so one extreme the storms may be just dotted higgledy-piggledy around the country. Another extreme they can have
much more organised structure, maybe they form some
line or some other feature. And what we tend to find if we use different models
to simulate these same events, is some of the models are very prone to generate
organised behaviour, others are very reluctant to do that and reality
actually lives, seems to live somewhere in between those two extremes. So what we
need to do is to kind of really unpack and understand that property. That’s
much more or less anecdotal impression from the forecast users, rather than something we really understand properly at the moment. The other kind of
slightly troubling aspects of these models is that the people who are
running the models in tropical locations, let’s say West Africa, are
typically going for quite different sort of settings and model choices, compared
to people running in a more mid-latitude situation, which is over the UK. Now in
reality, the fundamental physics and dynamics of the atmosphere is not
completely different in those different parts of the world of course, really
that’s pointing to some deficiencies in the model that have been sort of covered
up in slightly different ways, depending which processes are slightly more
important and in different places in the world.
So that’s unsatisfactory in some ways and what we’d really like to do we feel,
is if we understood the science better, we ought to be able to bring the
settings together and get a good setup that works well right away across the
world. So it’s for those reasons that actually we’re choosing to focus on
the weather forecasting over South Africa for this project. The reason there
is that South Africa is not, you know, sort of mid-latitude location, it’s not
your tropical location. It has behaviour that looks sometimes like one,
sometimes like the other, sometimes somewhere in between, depending on the
time of year and where you are in the country and so on. So really they
can’t sort of budge this situation by choosing one or the other, you have to
really know, to get a good forecast there, you really
do have to address this problem and come to some setting that will work well in a
very much more general sense. So the people involved in this project, apart
from myself as an academic here, there’s another academic, Thorwald Stein,
at the University, who’s an expert of conflicting the models with the
observations particularly but we’ve also got involvement from the forecasting
services, so we’re interested in developing and understanding the Met
Office forecast models, so we have a Met Office supervisor involved but also an
involvement from the South African weather services, so we can really
understand the problems that they face and what what improvements they need to
see. So there’s a brief description of the project, very happy to be contacted by email or please take a look at the the advert online, which will give
you a bit more information again.

Glenn Chapman

Leave a Reply

Your email address will not be published. Required fields are marked *