Context: Recently, the Indian National Centre for Ocean Information Services (INCOIS) has introduced a new tool called the Bayesian Convolutional Neural Network (BCNN) to forecast El Niño and La Niña conditions up to 15 months in advance.

What is BCNN?

• It uses latest technologies, such as Artificial Intelligence (AI), deep learning, and machine learning (ML), to improve forecasts related to the ENSO phases.

• It observes slow changes in the ocean and how they interact with the atmosphere.

• It is a combination of dynamic models with AI. This helps it forecast the emergence of El Niño and La Niña conditions with a 15-month lead time — unlike other models, which can give a prediction up to six to nine months in advance.

About ENSO

• The El Nino Southern Oscillation (ENSO) is a climatic phenomenon characterized by variations in the temperature of the central and eastern tropical Pacific Ocean, along with atmospheric changes.

• ENSO significantly impacts global weather patterns and occurs in cycles every 2-7 years.

Three phases of ENSO

• El Niño: Warm phase, where the eastern Pacific waters are warmer than usual due to weakened wind systems.

• La Niña: Cool phase, where the eastern Pacific waters are cooler than usual due to stronger wind systems.

• Neutral: The eastern Pacific (near the northwestern coast of South America) is cooler compared to the western Pacific (between the Philippines and Indonesia) due to prevailing east-to-west wind systems.

• In India, El Niño typically results in weak monsoons and severe heat waves, while La Niña leads to strong monsoons.

The New BCNN Product

• The BCNN uses Artificial Intelligence (AI), deep learning, and machine learning (ML) to enhance the prediction of ENSO phases by calculating the Niño 3.4 index, which is derived from the sea surface temperature (SST) anomalies in the central equatorial Pacific.

• Niño 3.4 index value is obtained by averaging the sea surface temperature (SST) anomaly in the central equatorial Pacific, extending from 5°N to 5°S, and 170°W to 120°W.

• This model can predict El Niño and La Niña conditions with a lead time of 15 months, compared to the 6-9 month predictions offered by other models.

Comparison with Existing Models

• Current weather forecasting models are either statistical, using data from various sources, or dynamic, involving 3D simulations on High-Performance Computers (HPC).

• The BCNN integrates dynamic modelling with AI, improving the accuracy and extending the lead time of ENSO predictions.

Challenges

• Developing the BCNN involved overcoming significant challenges, particularly the limited historical oceanic data. 

• While terrestrial weather data is abundant, oceanic records are relatively scarce, with comprehensive global ocean temperature records available only since 1871.

• This limitation was addressed by incorporating data from the Coupled Model Intercomparison Project (CMIP) phases 5 and 6, which include historical climate data from 1850 to 2014.

• CMIP is an experimental framework that allows climate modellers to simulate various scenarios to test the past climate and also project future climate situations.

Forecast by BCNN

• The BCNN predicts La Niña conditions will develop between July and September with a probability of 70-90% and continue until February 2025.

Indian National Centre for Ocean Information Services (INCOIS)

• It is a globally recognized autonomous institution under the Ministry of Earth Sciences.

• It provides the best possible ocean information and advisory services to society, industry, government agencies and the scientific community through sustained ocean observations.

• Provides round-the-clock monitoring and warning services for the coastal population on tsunamis, storm surges, high waves, etc.