Air Quality Forecasting via Dynamic Blending of Global and Regional Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing air quality forecasting models suffer from decreased accuracy over time due to the inability to effectively capture large-scale information, leading to limitations in both short-term and long-term forecasting.
Innovation Solution
A method that blends global and regional data using a synoptic scale correction factor, determined by weather pattern classifications, to improve the accuracy of air quality forecasts by dynamically filtering and combining large-scale and small-scale atmospheric information from global and regional weather models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If known air quality forecasting numerical models are used, then air quality forecasts can be generated, but the accuracy decreases with model integration time due to inability to capture large scale information
Solution Approach 1:
The patent merges global weather model data (capturing large-scale atmospheric patterns) with regional weather model data (capturing local conditions) to create a blended dataset. This combination allows the model to maintain accuracy over longer integration periods by incorporating both synoptic-scale features that evolve slowly and regional features that capture local variability.
Solution Approach 2:
The patent segments the atmospheric information into distinct scales: global large-scale patterns and regional small-scale features. By processing and blending these segmented components separately, the model preserves the accuracy benefits of both scales throughout the forecast integration period.
2Measurement precision
If regional weather model data is used alone, then local air quality conditions can be forecast, but large scale atmospheric information cannot be captured well
Solution Approach 1:
The patent combines regional weather model output with global weather model output, where the global model provides the missing large-scale atmospheric information. The blending process integrates these complementary data sources to produce a comprehensive dataset that contains both local and large-scale features.
Solution Approach 2:
The patent uses an incremental spatial filter as an intermediary to selectively blend global and regional data. This filter acts as a mediator that determines which spatial scales from each model should be combined, ensuring that large-scale information from the global model is properly incorporated while preserving regional details.
3Loss of information
If global weather model data is used alone, then large scale information can be captured, but local regional details are lost
Solution Approach 1:
The patent merges global weather model data with regional weather model data in a blended dataset. This merging ensures that large-scale information from the global model is preserved while simultaneously incorporating local regional details from the regional model, achieving both objectives together.
Solution Approach 2:
The patent applies local quality by allowing different parts of the blended dataset to have different characteristics: global data dominates at large spatial scales while regional data dominates at local scales. This scale-dependent blending ensures appropriate information is used for each spatial resolution.
4Measurement precision
If dynamic blending of global and regional data with synoptic scale correction factor is implemented, then air quality forecasting accuracy is improved for both short-term and long-term forecasts, but model complexity increases
Solution Approach 1:
The patent implements dynamic blending where the contribution of global versus regional data varies based on synoptic conditions. The synoptic scale correction factor dynamically adjusts the blending ratio, allowing the model to adapt to different atmospheric regimes and optimize accuracy for each condition rather than using a fixed blending approach.
Data Source
AI summary
According to one or more embodiments of the present invention, a method of forecasting air quality is provided. The method includes determining weather pattern classifications based on global atmospheric information from a global weather model and determining a synoptic scale correction factor in response to the determination of the weather pattern classifications. The method also includes blending the global atmospheric information the synoptic scale correction factor to produce a data set and blending the data set with regional atmospheric information from a regional weather model to generate weather fields. The method further includes blending chemical information from a global chemical model and the synoptic scale correction factor to produce a second data set and blending the second data set into a regional chemical model based on the weather fields to forecast the air quality.