Developed over 22 years with 200,000+ man hours and 100,000+ hours of computer time, and generating forecasts for 6.4 million locations in all 195 countries and islands, a forecast Weather Trends issued one year ago is more accurate than every other weather company’s 5-14 day forecast.
- 75% of the automated forecasting process is math-based using a combination of Gaussian Theory, Periodicity, Climate Cycles and Statistics that automatically cycle through a multi-variable 300-step decision tree of if/then scenarios to generate year-ahead by day or week with high/low temperature and weekly rain/snow forecasts by location everywhere in the world. 100% automated algorithms perform quadrillions of calculations on a Rackspace cloud-dedicated infrastructure using 5.3 million lines of code.
- 20% of the process uses a blend of 8 traditional meteorological atmospheric and oceanic time series (PDO, SOI, QBO, AMO, etc.) to refine the initial year-ahead numerical forecast.
- 4% is an automated location specific correction process to adjust for trend volatility biases, topographic factors and 10 to 30 year climate trend adjustments.
- 1% is human QC from Weathertrends’ degreed meteorologists/climatologists and PhD from Rutgers University, Millersville, Penn State, Plymouth State University, and the University of Miami.
1-14 Day Forecasting