Applied Data Science & Engineering for Local Weather Forecasts

Explore applied data science and engineering techniques for local weather forecasting in this 33-minute conference talk from GOTO Berlin 2016. Discover how dramatic improvements in weather forecasting and climate modeling have led to exponential increases in data generation. Learn about MeteoGroup's ability to generate 10-day weather forecasts on-the-fly from a browser. Delve into topics such as numerical weather prediction models, forecasting systems, model engineering, scaling with Apache Airflow, and forecast verification metrics. Gain insights into the iterative process of model improvement and the results achieved. Presented by Nikhil Podduturi, Senior Data Engineer/Data Scientist at MeteoGroup, this talk provides a comprehensive overview of modern weather forecasting techniques and the data science behind them.

Intro
Agenda
Types of data
Numerical weather prediction models
Outcome
MeteoGroup forecasting system
Next generation forecasting system
Iteration 1
Iteration 2
Baseline model engineering
Scaling with Apache Airflow
Apache Airflow DAG
Airflow and Mesos
Model improvement cycle
Forecast verification
Verification metrics
Results
Improvements
CONFERENCES