AI Being Deployed to Protect Agriculture from El Niño

Izzy Humphreys for Distilled Post

It has been announced by the World Meteorological Organisation (WMO) in the last few weeks that we have officially entered an El Niño weather phase. The UK is often spared the brunt impact of this climate phenomenon, but in some parts of the world this sudden change in temperature and weather patterns can prove detrimental to people’s livelihoods, particularly those in agricultural sectors. 

Current climate models can predict El Niño events to some extent, but new AI technology is being developed that could potentially buy governments and populations valuable months to pre-emptively protect against the worst climate fluctuations.

What is El Niño and how does it impact agriculture? 

El Niño and La Niña are terms for two weather phenomena that create the biggest fluctuation in our climate system. Alternating in irregular cycles every 2 to 7 years, La Niña causes global temperatures to sink to below average whilst El Niño causes them to increase, and together these episodes are described as the El Niño Southern Oscillation (ENSO). 

Both are related to the natural warming and cooling of the sea surface in the Pacific combined with wind trajectories. Due to the close interaction between the ocean and the Earth’s atmosphere, this fluctuation in sea temperature causes widespread changes in the climate which can have a significant impact on infrastructure, agriculture, health, and energy sectors. 

With El Niño creating warmer conditions, drought becomes the main threat to food production, but countries are also more susceptible to heavy rain and flooding. This weather can lead to animal disease outbreaks, plant pests, and forest fires, which have been seen during El Niño weather patterns before. 

Whilst the El Niño heat can sometimes boost food production, particularly with crops like corn and wheat, the extensive dry weather and hotter temperatures are detrimental to lots of ordinarily lucrative produce. Sugar production in India, palm oil production in Indonesia and Malaysia, and cocoa production in West Africa are all expected to plummet during this El Niño period, and prices of these commodities have already risen sharply since the WMO’s announcement. The negative economic effects have been estimated to reach $3 trillion between now and 2029.

Current predictive models

Due to the ENSO cycle’s irregularity, it is difficult to predict and therefore difficult to prepare for. Measures such as reinforcing riverbanks with sandbags, helping governments to prepare food insecurity response plans, and supporting countries in Central America to increase the resilience of households are all being utilised to attempt to combat the potentially devastating impacts of El Niño. But up until recently, climate scientists haven’t been able to predict it more than a year in advance, which stunts many efforts to adequately prepare. 

Existing predictive technologies to anticipate El Niño events rely heavily on small sets of historical statistics and current climate models do not provide detailed enough images to detect the change in our climate system with much accuracy. Satellites, moored buoys, drifting buoys, sea level analysis and expendable buoys make up the “ocean observing systems” which gather data used by large computer models of the ocean and atmosphere as input to predict El Niño. But these systems are only roughly 54% accurate. 

Developing technology 

New AI technology is in the works to attempt to predict ENSO events more in advance and with more accuracy. A convolutional neural network is being developed which can recognise images and has been trained on global images of historic sea surface temperatures and deep ocean temperatures. This allows it to learn how they parallel the future emergence of ENSO events.

In testing this neural network, it managed to detect El Niño events 18 months into the future, and with 74% accuracy. Although this doesn’t sound like much, the extra 6 months' warning of extreme heat, rainfall or drought could be crucial to the precautionary measures taken by countries to protect their agriculture and infrastructure. 

As the new deep learning model continues to be developed, it will hopefully become even more proficient and continue to increase in its predictive power, so farmers can work on breeding climate-resistant crops, companies can prepare for supply chain disruptions, and governments can generate more extensive disaster response systems. The AI technology also possesses the ability to pinpoint which areas of the Pacific will heat up the most, allowing a far more streamlined prediction of climate changes so unnecessary precautions aren’t taken by countries that may not be too negatively impacted.

Even with all the predictive technology in the world, ENSO events can prove cataclysmic for agriculture around the world, particularly those that rely on the exportation of a handful of highly sought-after commodities to maintain their infrastructure and economy. However, with this new AI technology in place, the worst effects of El Niño may be able to be avoided; this next El Niño phase will be pivotal in analysing how adept the new predictive technology is at allowing countries sufficient preparation time.