NEWS – Hyderabad based Indian National Centre for Ocean Information Services (INCOIS) has launched New Product called Bayesian Convolutional Neural Network (BCNN)  to Forecast El Niño and La Niña Conditions

HIGHLIGHTS

Understanding ENSO

• Definition: ENSO (El Niño Southern Oscillation) is a climate phenomenon involving temperature changes in the central and eastern tropical Pacific Ocean and atmospheric fluctuations.
• Phases:
  • El Niño: Warmer than usual temperatures in the eastern Pacific.
  • La Niña: Cooler than usual temperatures in the eastern Pacific.
  • Neutral: Cooler waters in the eastern Pacific due to prevailing east-to-west winds.

Impact on India

• El Niño: Leads to weak monsoon and intense heatwaves.
• La Niña: Results in a strong monsoon.

Features of the New Product

• Combines dynamic weather modeling with AI for enhanced accuracy.
• Uses Niño3.4 index value to determine ENSO phases by averaging sea surface temperature (SST) anomalies in the central equatorial Pacific.

Comparison with Existing Models

• Statistical Models:
  • Generate forecasts based on historical data from various regions.
• Dynamic Models:
  • Utilize 3D mathematical simulations of the atmosphere with High Performance Computers (HPC).
  • More accurate than statistical models.
• BCNN Advantage:
  • Integrates AI with dynamic modeling.
  • Provides predictions up to 15 months in advance, compared to 6-9 months by existing models.

Development Challenges

• Data Scarcity:
  • Limited historical weather data, especially for oceans.
  • Historical oceanic temperature records available only since 1871.
• Solution:
  • Augmented data using historical runs from the Coupled Model Intercomparison Project (CMIP) phases 5 and 6.
  • Developed the BCNN model over eight months with several testing phases.