Snowfall Probability Distribution Forecasting System

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current snowfall accumulation forecasts provide deterministic predictions without conveying probability distributions or the forecaster's confidence level, failing to account for the range of potential outcomes and uncertainties in snowstorms.

Innovation Solution

A rules-based process leveraging third-party weather forecasts to generate snowfall probability distributions, including the most likely accumulation range and probabilities within and outside that range, ensuring consistency with deterministic forecasts and normalizing data for accurate representation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If deterministic forecasts are used, then forecast simplicity and ease of operation are improved, but information completeness and reliability deteriorate

Engineering Contradiction:
Improveforecast simplicityVSAvoidprobability distribution information
Core Design Contradiction:
Ease of operationVSLoss of information

Solution Approach 1:

The forecast is segmented into multiple discrete outcomes with associated probabilities. Instead of a single deterministic value, the system divides the forecast into distinct snowfall accumulation ranges (e.g., 0-1 inch, 1-2 inches, 2-3 inches) and assigns probability weights to each range, allowing users to understand both the most likely outcome and the full spectrum of possible outcomes.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The forecast transitions from a one-dimensional deterministic value to a two-dimensional probability distribution. By adding the probability dimension to the snowfall accumulation forecast, the system provides both the predicted value and the uncertainty associated with that prediction, enriching the information without complicating the basic forecast structure.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Reliability

If probability distributions are generated from ensemble forecasts, then reliability and information completeness are improved, but device complexity and computational requirements worsen

Engineering Contradiction:
Improveforecast accuracyVSAvoidprocessing complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system extracts and processes only the relevant snowfall accumulation data from the ensemble forecasts, filtering out unnecessary meteorological variables and model details. By focusing solely on the snowfall accumulation values and their probability distributions, the system reduces processing complexity while maintaining forecast reliability.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The system transforms the raw ensemble forecast data into standardized probability distribution parameters (mean, standard deviation, percentile values) that can be directly used for forecasting. This parameter transformation simplifies the complex raw data into actionable statistical summaries, reducing computational requirements while preserving forecast accuracy.

Inventive Principle:
Principle #35Parameter changes

3Productivity

If third-party weather forecasts are leveraged, then productivity and forecasting speed are improved, but measurement precision and data reliability worsen

Engineering Contradiction:
Improveforecasting speedVSAvoidforecast accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The system merges multiple third-party weather forecasts into a unified probability distribution. By combining the predictions from various independent sources and using statistical methods to aggregate their results, the system achieves both speed (by processing multiple forecasts simultaneously) and precision (through statistical aggregation that accounts for individual forecast uncertainties).

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP3607363B1System and method for forecasting snowfall probability distributions
Publication Date: 2023.06.28 ACCUWEATHER INC
  • EP3607363B1 patent drawingFigure 1
  • EP3607363B1 patent drawingFigure 2
  • EP3607363B1 patent drawingFigure 3

AI summary

Currently available weather forecasts, which include a specific snowfall accumulation or range, do not convey the probability that snowfall will be within the forecasted snowfall accumulation range, probabilities of other snowfall accumulation amounts, or a forecaster's level of confidence. A snowfall probability distribution forecasting system is disclosed that uses a rules-based process to leverage third party weather forecasts, including members of ensemble forecasts, to generate snowfall probability distributions forecasting the most likely snowfall accumulation range, the probability that snowfall accumulation will be within the most likely snowfall accumulation range, and probabilities that snowfall accumulation will be outside of the most likely snowfall accumulation range. To ensure consistency with the deterministic forecast, the snowfall probability distribution may be shifted based on a deterministic forecast. Because third party weather forecasts can produce a non-normal distribution of snowfall accumulation forecasts, the snowfall probability distribution may be normalized.