Mid-range Weather Forecasting And Predicting Disease
Chris - Tell us about your work.
Tim - I think the thing to get over here is that there are three time scales where people try to predict weather in the future. There's the sort of daily weather forecast where we look at what's happening tomorrow or the next day. Then there are these problems of climate change where we might be talking about what might happen to the weather a hundred years from now as a result of the emissions that we're putting into the atmosphere. We're changing atmospheric composition, which is changing weather. Then there's a third time scale, which lies in between the weather time scale and the climate change time scale, which is what's going to happen maybe a few months into the future. That's an area that I'm very much involved with along with a number of colleagues around the world. The key to this time scale of a season or maybe two seasons lies in the oceans. When we make a weather forecast for tomorrow we assume that the ocean temperatures and the ocean currents are pretty constant. They don't change from one day to the next very much. But on time scale'sof a few months ahead then ocean temperatures and ocean currents really do change a lot. A classic example is the El Nino phenomenon in the Pacific ocean, where in the space of a few months temperatures can warm by five or six degrees. When this happens it throws the weather patterns around the world into complete turmoil. So by trying to predict how ocean temperatures and ocean currents evolve in the coming months, we can make useful predictions about weather on that time scale as well. That's something that I'm very much involved in.
Chris - Now when you come to draw up a model like this, how do you actual start? It's very easy to say that we're going to make a few measurements and work out what that does to the weather, but how do you actually marshal that amount of information?
Tim - Essentially the problem isn't much different to that which Alex spoke about for weather forecasting. So there are two essential ingredients. One is observations. Here the key is not just only observations of the atmosphere but also observations of the oceans. So for example we have all sorts of different types of ocean measuring systems: buoys and instruments which go down into the deep and then come up again to beam the information to a satellite. So that gives us information about what both the atmosphere and the oceans are doing right now. The second ingredient is the models that Alex spoke about. These are basically models based on the laws of physics based on things like Newton's laws of motion, which you learn about in school. These are them main models which integrate forward in time information from the atmosphere and the oceans. The key difference between a weather forecast model and a seasonal prediction model is that we need to represent not only the atmosphere but the oceans as well, so we have this two phase approach of getting these complex systems into these computer models.
Kat - In a nutshell, what would you say to someone who said that because we can't predict the weather next week, we can't know what global warming is going to do?
In a nutshell, we run the models and what we make are probabilistic forecasts of climate change. So it is overwhelmingly likely that in a hundred years it'll be warmer than it is now. Exactly how much warmer is actually hard to say. It could be anything between a few degrees warmer and maybe up to about 12 degrees warmer, and that uncertainty reflects to some extent the chaotic nature of the atmosphere. Now decision makers have to take that into account. This is the best information the scientists can give them. There is a risk of what I would say is quite catastrophic warming but based on this information the most likely warming is about 3 or 4 degrees above normal. But basically what we give the policy makers are probability forecasts for the future.
Chris - And just to finish off Tim, What about disease here? If we can tell what the weather's going to do, what are the impacts potentially on our ability to handle and marshal diseases which are obviously a major consideration with climate change or weather in general?
Tim - That's right. There are many applications of this type of weather forecast or seasonal forecast I've been speaking about, and one actually is in disease. Now one of the biggest diseases in the world is malaria. This has been increasing over the years. There's a type of malaria called epidemic malaria which tends to occur in semi-arid regions of the world where the incidence of malaria is not the same year in year out. You get sudden pulses or years where you get a very strong incidence of malaria. It's known that this is related to weather variability, and in particular years that are very wet tend to have these epidemics of malaria. Work that we've been doing in conjunction with scientists in the US is to make predictions of the coming rainy seasons months ahead of time, and by linking those to malaria prediction models, we can then give predictions of whether malaria incidence in this particular epidemic prone regions is likely to be above or below average. Then the authorities can target these areas months ahead of time to spray houses with insecticides to provide children with bed nets, and the sort of things that can really save lives.
Chris - Sounds great but is it actually working?
Tim - It is working. We're doing this now in the field. We published our results earlier this year based on a twenty year study of Botswana which has very good malaria epidemiology so we could test our prediction models and assess the skill, and this is now being used in the field on a very routine basis. This is a very nice example of how we can go from weather forecasting to something that really matters to a lot of people. It's literally life or death for a lot for people in the world.