Identifying visual storm signatures from satellite images
Satellite images from vast historical archives are analyzed to predict severe storms. We extract and summarize important visual storm evidence from satellite image sequences in a way similar to how meteorologists interpret these images. The method extracts and fits local cloud motions from image sequences to model the storm-related cloud patches. Image data of an entire year are adopted to train the model. The historical storm reports since the year 2000 are used as the ground-truth and statistical priors in the modeling process. Experiments demonstrate the usefulness and potential of the algorithm for producing improved storm forecasts. A preferred method applies cloud motion estimation in image sequences. This aspect of the invention is important because it extracts and models certain patterns of cloud motion, in addition to capturing the cloud displacement.
1. A method of automatic storm detection or prediction based on historical meteorological data and satellite image sequences, for validating and complementing conventional forecasts, the historical meteorological data includes historical storm reports and satellite image archives for historical weather, comprising the steps of:
providing, a digital computer;
receiving, at the digital computer, a sequence of satellite images involving a geographical region;
receiving, at the digital computer, historical meteorological data associated with the geographical region;
executing an algorithm on the digital computer to perform the following operations:
automatically extract visual features indicative of storm signatures from the sequence of satellite images;
learn the correspondences between the visual storm signatures and the occurrences of current and future storms based on the historical meteorological data; and
detect or predict storms in the geographical region using the learned correspondences.
2. The method of claim 1 , wherein the correspondence is a statistical model trained based on the historical meteorological data.
3. The method of claim 1 , including the step of estimating cloud motion in the image sequences to determine dynamic areas having a storm potential.
4. The method of claim 3 , wherein the step of estimating cloud motion in the image sequences includes automatically analyzing the optical flow between adjacent satellite images or among a contiguous sequence of satellite images.
5. The method of claim 4 , including the step of using rotation, divergence, and/or velocity of the optical flow in a local area to model potential storm patterns.
6. The method of claim 1 , including the step of extracting and fitting local cloud motions from image sequences to model storm-related cloud patches.
7. The method of claim 1 , including the step of identifying a dense optical flow field between two adjacent images or within a contiguous sequence of the satellite images.
8. The method of claim 7 , including the step of using the flow field to identify local vortex regions indicative of potential storm systems.
9. The method of claim 8 , including the step of constructing a descriptor for each vortex region based on the visual and optical flow information and the historical storm data.
10. The method of claim 9 , including the step of using the historical storm data to construct a storm statistical density in geographical or temporal coordinates.
11. The method of claim 1 , including the steps of:
using the extracted visual features and the historical meteorological data to identify candidate storm cells;
using machine learning techniques to classify the candidate cells into storm and stormless cells; and
using only candidates classified as storm cells to predict storms.
12. The method of claim 1 , including the step of automatically identifying comma-shaped cloud patterns from the sequence of images.
13. The method of claim 12 , including the step of automatically extracting vortexes in the comma-shaped cloud patterns.
14. The method of claim 1 , including the step of integrating other meteorological data and/or prediction models to improve the accuracy of the prediction.
15. The method of claim 1 , including the step of integrating expert opinions to improve the accuracy of the prediction.