‘Climate change making certain aspects of weather forecasting challenging’

Akshay Deoras is a doctoral researcher in the Department of Meteorology, University of Reading, UK. He is currently working on understanding the predictions and characteristics of Indian monsoon low-pressure systems.

He is also an independent meteorologist with expertise in forecasting severe weather events. In an interview with SANJANA BHALERAO he explains various aspects of weather forecasting challenges. Excerpts:

Can you, in layman’s language, explain weather models and the science behind them?

Let us try to understand the science behind them with the help of a very simple example. Imagine that you have an important meeting after three days, but you are not feeling well today. You approach a doctor to know if your health will allow you to attend the meeting. The doctor carefully examines your current health condition, links it with some known aspects of medical sciences, and then comes up with an answer.

So, your doctor is simulating your near-future health based on your current health and known aspects of medical sciences. The science behind weather models is similar to this example. We know that the near-future state of the atmosphere can be predicted based on its current state and this information is embedded in complex computer models. Current weather conditions observed by various observing systems (e.g., weather satellites) are then fed into these computer models for producing weather forecasts. The calculations are extremely tricky; so, supercomputers are needed.

What are the limitations of weather models?

One major limitation of weather models is that their accuracy decreases as the forecast lead time increases. This means that a weather forecast for tomorrow will generally be more accurate than a weather forecast for someday in the next week. Another limitation is that weather models can well predict large-scale weather conditions (e.g., tropical cyclones), but they may not always well predict smaller-scale weather events such as thunderstorms. In some cases, they may falsely predict certain weather events with high confidence such as the rapid intensification of tropical cyclones or heavy rainfall in a city.

There has been an increase in extreme weather events in the country, considered to be an impact of changing climate patterns. Mumbai was under a heatwave warning in March, something unheard of earlier. Last year, extreme rainfall claimed over 200 lives in the state. Is climate change introducing fresh uncertainties in weather forecasting?

In my opinion, climate change is making certain aspects of weather forecasting challenging. Of course, the most visible impact of global warming is on temperature (e.g., increasing frequency, intensity and duration of heatwaves). However, models do a decent job at predicting temperatures than predicting rainfall.

We know that global warming can intensify extreme rainfall events, thereby increasing the risk of hydrological disasters. Forecasting rainfall over a region becomes very challenging if there are mountains in the region. We saw Mahabaleshwar in western Maharashtra receiving over 1500 mm rainfall in three days in July 2021, which was not well predicted by weather models. We are also facing challenges in predicting the rapid intensification of tropical cyclones as well as thunderstorms—the “monster” thunderstorm in Mumbai on 18 July 2021 can be considered to be an example of this.

Is India equipped to meet these climate change-induced challenges?

Not totally at the moment. See there are a few things here. The unpleasant truth about such challenges is that they will continue to occur and cause deaths and damage. However, disaster mitigation is in our hands; we can save lives if such events are accurately predicted well in advance, warnings are swiftly communicated to stakeholders, and people have a clear idea of how to respond. Our current weather models are a lot more accurate than what they were a few decades ago. The current preparedness is also a lot better for certain weather events such as tropical cyclones along the east coast. However, the same is not true for the west coast, which might be threatened again shortly even if the odds will always remain smaller than those for the east coast. Besides, high-impact events such as lightning strikes, flash floods, and landslides remain a matter of concern given their relatively short predictability period.

We must not forget that even if we improve our weather models in the future, we will still need to beef up catastrophe modelling, dissemination of advisories, and weather literacy among people. Let us take western Maharashtra’s example, which witnessed many deadly landslides in July 2021. There are no landslide prediction models or a warning system in place that could be used this monsoon season given the IMD’s long-range forecast of above-normal seasonal rainfall in this region.

Is a localised forecast, say a forecast specific to an administrative ward, possible? And is it required?

Stakeholders have always demanded localised forecasts. In India, we do not need such forecasts for temperatures, but rainfall forecasts specific to an administrative ward will certainly benefit everyone. However, producing such a local-level forecast could be a challenging and costly task. Moreover, such forecasts need to be accurate; otherwise, there will be a lot of inconvenience due to false alarms. As a result, I believe that the need of the hour is to improve the detection and real-time monitoring of weather events and swift dissemination of warnings to stakeholders in a user-friendly language.