Weather Scenario Sampling With Upscaling and Downscaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for generating computerized weather simulations over large regions result in data matrices that are too large, leading to high computational costs and inefficiencies.
Innovation Solution
A method and system using an optimal multiple-point geostatistics technique to reduce and then restore the resolution of input data, employing upscaling and downscaling techniques to manage large weather simulation data without requiring hardware upgrades, utilizing methods like bicubic resampling and neural networks when available.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If multiple-point geostatistics method is applied to generate weather scenarios over large regions, then simulation coverage area is improved, but computational cost and data matrix size increase significantly
Solution Approach 1:
The patent divides the large weather simulation domain into multiple smaller sub-regions or grids. By segmenting the large data matrix into smaller manageable pieces, the multiple-point geostatistics method can be applied to each segment independently, reducing the computational burden while still covering the entire large region through aggregation of results from segments.
Solution Approach 2:
The patent introduces an intermediary downscaling step that converts high-resolution simulation data into lower-resolution representative data. This intermediary process acts as a mediator between the detailed simulation results and the final output, reducing the data dimensionality and computational complexity while preserving essential weather pattern information.
2Area of stationary object
If multiple-point geostatistics method is applied to generate weather scenarios over large regions, then simulation coverage area is improved, but data matrix size increases to very large dimensions
Solution Approach 1:
The patent segments the large spatial domain into smaller sub-regions, reducing the data matrix size for each segment. This segmentation allows the system to handle smaller data matrices that can be processed efficiently by the multiple-point geostatistics method while still achieving comprehensive coverage through aggregation.
Solution Approach 2:
The patent extracts and removes redundant information from the simulation data through downsampling and selection of representative points. By taking out unnecessary data points and retaining only the most informative ones, the system reduces the overall data matrix size while maintaining the essential weather scenario characteristics.
3Productivity
If input data resolution is reduced to improve computational performance, then computational cost is reduced, but data quality and simulation accuracy deteriorate
Solution Approach 1:
The patent applies preliminary upscaling to the input data before processing, converting low-resolution input data into higher-resolution intermediate data. This preliminary action ensures that the subsequent multiple-point geostatistics processing works with sufficiently detailed data to maintain accuracy, while the overall system can still handle computational loads efficiently.
Solution Approach 2:
The patent introduces an intermediary upscaling step that converts low-resolution input data into higher-resolution intermediate representations. This intermediary process acts as a bridge between the computational efficiency requirement and the data quality requirement, allowing the system to process data at a lower resolution while maintaining output quality through the upscaling transformation.
Data Source
AI summary
Improved performance of a computer simulation using sampling can include using a computer to process a resolution reduction of input data by an upscaling technique providing upscaling of the input data. The input data including a size and a resolution, a location, a date, conditioning points, and weather variable data. The upscaling of the input data generating reduced resolution data with less resolution and less data size than the input data. A multiple-point geostatistics technique is applied to the reduced resolution data for generating a simulation of a new weather scenario, without changes in size or resolution to the reduced resolution data. The computer is used to process a resolution increase of the reduced resolution data from the simulation using a downscaling technique to return the reduced resolution data to the size and the resolution of the input data.