Lightning Forecast Using Radar Reflectivity and Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current lightning forecast methods are not accurate enough to predict the intensity of Spark Discharge of the Air (SDA) and fail to provide reliable short-term forecasts, which can lead to grid device failures and safety risks.
Innovation Solution
A method and apparatus that identify Targeted Weather of SDA Carrier (TWLC) using radar reflectivity data, forecast future TWLC states, build or update an SDA model based on observation data, and calculate the probability of SDA occurrence using a gradient-based optical flow algorithm and random probability model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a linear model is used with historical meteorological data and SDA data, then the forecast can be made using available data, but the accuracy is not enough and cannot forecast the SDA intensity
Solution Approach 1:
The patent transforms the forecast approach from using historical meteorological data with a linear model to using real-time radar reflectivity data with a trained neural network model. This parameter change enables the system to capture non-linear relationships and accurately forecast SDA intensity, directly resolving the contradiction between forecast accuracy and reliability.
Solution Approach 2:
The patent replaces the traditional linear statistical model with a neural network-based intelligent forecast system. This substitution allows the system to process radar reflectivity data and predict SDA intensity with higher accuracy, overcoming the limitations of linear models in capturing complex atmospheric phenomena.
2Reliability
If radar reflectivity data is used to indicate probability of producing SDA under current weather condition, then the current state can be assessed, but the probability of producing SDA in future time cannot be forecast
Solution Approach 1:
The patent trains a neural network model in advance using historical radar reflectivity data and SDA occurrence patterns. This preliminary action enables the system to not only assess current conditions but also predict future SDA probability by analyzing the evolution of radar reflectivity patterns over time, thus overcoming the limitation of purely descriptive radar-based methods.
Solution Approach 2:
The system uses trained neural network models that have learned from historical data to continuously predict future SDA probability based on evolving radar reflectivity patterns. This feedback mechanism allows the system to transition from static current condition assessment to dynamic future forecast, resolving the time prediction limitation.
3Productivity
If a high-resolution weather model is used to forecast TWLC and calculate SDA index, then the forecast process can be completed, but there is a great error in forecasting the TWLC
Solution Approach 1:
The patent extracts and utilizes radar reflectivity data, which directly reflects the physical state of precipitation and atmospheric conditions associated with SDA. By focusing on this specific observable parameter rather than relying on complex weather model simulations of TWLC, the system achieves more accurate SDA forecasts while maintaining processing efficiency.
Solution Approach 2:
The system uses observed radar reflectivity patterns as a proxy for the complex atmospheric processes that weather models attempt to simulate. This copying approach bypasses the errors inherent in TWLC forecasting while capturing the essential features needed for SDA prediction, resolving the contradiction between processing efficiency and accuracy.
Data Source
AI summary
The present invention proposes a lightning forecast method, comprising: identifying a Targeted Weather of SDA carrier (TWLC) based on radar reflectivity data; forecasting a future TWLC state based on the identified TWLC; building or updating an SDA model based on SDA observation data and detected TWLC state-related data; and calculating the probability of producing SDA in the forecasted future TWLC according to the SDA model.