Forecasting method with machine learning

1. A computer-automated method for generating a weather forecast from current weather data and an intensity value associated with an historical storm, comprising:

obtaining or deriving current weather data including weather data stored in a database associated with one or more weather parameter tiles, the current weather data derived at least in part from at least one sensor, and

performing, with at least one processor in response to instructions stored in at least one non-transitory memory, operations comprising:

determining when the obtained current weather data includes characteristics indicative of a storm, and detecting storm presence when the obtained current weather data is determined to include characteristics indicative of a storm,

retrieving one or more tracked storms from at least one database,

comparing characteristics of the detected storm presence with characteristics of the retrieved one or more tracked storms including retrieving a tracked storm intensity profile associated with the one or more tracked storms from the at least one database,

updating the tracked storm intensity profile based, at least in part, on matching of the characteristics of the detected storm presence with stored tracked storm characteristics, and saving the updated tracked storm intensity profile to the at least one database,

obtaining information corresponding to one or more historical storms from at least one historical storm database, wherein the obtained information includes characteristics of a historical storm and a historical storm intensity profile associated with the historical storm,

comparing reflectivity and/or storm path characteristics of the one or more tracked storms with reflectivity and/or storm path characteristics of the one or more historical storms,

selecting the historical storm in response to the comparison of the reflectivity and/or storm path characteristics of the one or more tracked storms with characteristics of the one or more historical storms,

predicting a future value of a weather parameter associated with the one or more tracked storms,

using the historical storm intensity profile associated with the selected historical storm to adjust the predicted future value of the weather parameter,

assigning the adjusted predicted future value of the weather parameter to one or more tile layers, and

storing the assigned predicted future value of the weather parameter in a data structure associated with the one or more tile layers.

2. The method of claim 1 wherein the tracked storm intensity profile includes a time series of weather parameter intensity values.

3. The method of claim 2 wherein the weather parameter intensity values include precipitation intensity values and/or other weather parameter intensity values.

4. The method of claim 1 wherein the tracked storm intensity profile defines or is defined by a linear equation generated by curve fitting a time series of weather parameter intensity values.

5. The method of claim 1 wherein the historical storm intensity profile includes a blended storm intensity profile that includes information corresponding to each of two or more historical storms.

6. The method of claim 1 wherein the operations further include:

retrieving one or more storm intensity profile rules from a data store; and

selecting a historical storm intensity profile by using a classifier that implements the one or more storm intensity profile rules to classify a tracked storm.

7. The method of claim 1 wherein the operations further include:

determining a metric indicating a degree of similarity between the one or more tracked storms and each of two or more historical storms;

generating an array of plural historical storms ordered and/or indexed and/or ranked according to a degree of similarity (DoS) with the one or more tracked storms, and

selecting a storm intensity profile corresponding to a historical storm based on the ordering and/or indexing and/or ranking.

8. The method of claim 7 wherein the operations further include:

generating a value of a quality metric associated with a predicted future state of the tracked storm;

comparing the value of the quality metric to a threshold value; and

when the value of the quality metric is less than the threshold value, selecting a different storm intensity profile corresponding to a different historical storm.

9. The method of claim 8 wherein selecting the different storm intensity profile includes generating a new or updated array of plural historical storms.

10. The method of claim 1 wherein the operations further include:

generating a forecast error metric value associated with predicted future characteristics of the one or more tracked storms;

comparing the forecast error metric value to a threshold value; and

when the forecast error metric value is greater than the threshold value, adjusting a historical storm intensity profile parameter.

11. The method of claim 10 wherein the operations further include storing the adjusted historical storm intensity profile parameter in the at least one historical storm database.

12. The method of claim 10 wherein adjusting the historical storm intensity profile parameter includes:

obtaining additional current weather parameter data; and

implementing a Bayesian scheme to incorporate additional current weather parameter data into an algorithm for updating the historical storm intensity profile parameter.

13. The method of claim 1 wherein the operations further include:

obtaining additional current weather parameter data;

updating the tracked storm intensity profile based on the additional current weather data; and

creating an entry corresponding to the one or more tracked storms in the at least one historical storm database, the entry including the updated tracked storm intensity profile.

14. The method of claim 13 wherein the created entry includes one or more of the following corresponding to and/or associated with the one or more tracked storms:

a time series of weather parameter data;

geographic features;

storm object and motion data; and

controlling variables.

15. The method of claim 1 , wherein predicting the future value of the weather parameter includes:

retrieving, from the at least one historical storm database, information indicating controlling variables for a forecast associated with the selected historical storm, and using the controlling variables to initiate a forecast to predict a future value of the weather parameter.

16. The method of claim 15 , wherein the operations further include:

selecting two or more sets of forecast initiation parameters from among a set of weather parameters associated with a historical storm,

generating multiple forecasts for the historical storm, including a separate forecast of the historical storm using the sets of forecast initiation parameters,

determining a quality metric of each of the multiple forecasts by comparing forecast results to historical storm weather parameter data,

determining that the controlling variables for the historical storm are the set of forecast initiation parameters associated with the forecast having a skill metric with the highest value, and

saving the controlling factors in the at least one historical storm database.

17. A computer-automated method of generating a weather forecast from sensed current weather data derived from at least one sensor and an intensity value associated with an historical storm, comprising performing, with at least processor based on instructions stored in at least one non-transitory memory, operations comprising:

acquiring weather data associated with one or more weather parameter tiles,

determining when the acquired weather data includes characteristics indicative of a storm and detecting presence of the storm when the acquired weather data is determined to include characteristics indicative of the storm,

generating an entry for the acquired weather data in a storm database,

generating a storm intensity profile associated with the acquired weather data,

predicting a future value of a weather parameter associated with the acquired weather data based at least in part on the generated storm intensity profile,

obtaining historical storm characteristics including reflectivity and/or storm path characteristics corresponding to one or more historical storms from a historical storm database,

obtaining historical storm intensity profiles corresponding to one or more historical storms from the historical storm database,

comparing reflectivity and/or storm path characteristics of the storm with reflectivity and/or storm path characteristics corresponding to one or more historical storms,

based on the obtained historical storm characteristics and historical storm intensity profiles and results of the comparing, selecting a historical storm,

using a historical storm intensity profile associated with the selected historical storm to adjust the predicted future value of the weather parameter,

assigning the adjusted predicted future value of the weather parameter to one or more tile layers, and

storing the assigned predicted future value of the weather parameter in a data structure associated with the one or more tile layers.

18. The method of claim 17 wherein the generated storm intensity profile includes a time series of weather parameter intensity values.

19. The method of claim 18 wherein the weather parameter intensity values include precipitation intensity values and other weather parameter intensity values.

20. The method of claim 17 wherein the generated storm intensity profile defines or is defined by a linear equation generated by curve fitting a time series of weather parameter intensity values.

21. The method of claim 17 wherein the historical storm profile associated with the selected historical storm includes a blended storm intensity profile that includes information corresponding to each of two or more historical storms.

22. The method of claim 17 wherein the operations further include:

retrieving one or more storm intensity profile rules from a data store; and

selecting a historical storm intensity profile by using a classifier that implements the one or more storm intensity profile rules.

23. The method of claim 17 wherein selecting a historical storm includes:

determining a metric indicating a degree of similarity between the acquired weather data and each of two or more historical storms; and

generating an array of plural historical storms ordered and/or indexed and/or ranked according to a degree of similarity (DoS) with the acquired weather data, and

selecting a storm intensity profile corresponding to a historical storm based on the ordering and/or indexing and/or ranking.

24. The method of claim 23 wherein the operations further include:

generating a value of a quality metric associated with a predicted future state of the storm;

comparing the value of the quality metric to a threshold value; and

when the value of the quality metric is less than the threshold value, selecting a different matched storm intensity profile corresponding to a different historical storm.

25. The method of claim 24 wherein selecting the different matched storm intensity profile includes generating a new or updated array of plural historical storms.

26. The method of claim 17 wherein the operations further include:

generating a forecast error metric value associated with predicted future characteristics of the storm;

comparing the forecast error metric value to a threshold value; and

when the forecast error metric value is greater than the threshold value, adjusting a historical storm intensity profile parameter.

27. The method of claim 26 wherein the operations further include storing the adjusted historical storm intensity profile parameter in the historical storm database.

28. The method of claim 26 wherein adjusting the historical storm intensity profile parameter includes:

obtaining additional current weather parameter data; and

implementing a Bayesian scheme to incorporate additional current weather parameter data into an algorithm for updating the historical storm intensity profile parameter.

29. The method of claim 17 wherein the operations further include:

obtaining additional current weather parameter data;

updating the generated storm intensity profile based on the additional current weather data; and

creating an entry corresponding to the acquired weather data in the historical storm database, the entry including the updated generated storm intensity profile.

30. The method of claim 29 wherein the entry includes one or more of the following corresponding to and/or associated with the determined storm presence:

a time series of weather parameter data;

geographic features;

storm object and motion data; and

controlling variables.

31. The method of claim 17 , wherein predicting the future value of the weather parameter includes:

retrieving, from the historical storm database, information indicating controlling variables for a forecast associated with the selected historical storm, and

using the controlling variables to initiate a forecast to predict a future value of the weather parameter.

32. The method of claim 31 wherein the operations further include:

selecting two or more sets of forecast initiation parameters from among a set of weather parameters associated with a historical storm,

generating multiple forecasts for the historical storm, including a separate forecast of the historical storm using the sets of forecast initiation parameters,

determining a quality metric of each of the multiple forecasts by comparing forecast results to historical storm weather parameter data,

determining that the controlling variables for the historical storm are the set of forecast initiation parameters associated with the forecast having a skill metric with the highest value, and

saving the controlling factors in the historical storm database.

33. The method of claim 17 wherein the operations further include:

retrieving characteristics of a tracked storm from a database,

comparing characteristics of detected storm presence with the retrieved characteristics of the tracked storm to determine whether the detected storm presence corresponds to the tracked storm, and

when the detected storm presence does not correspond to the tracked storm, generating the entry for the acquired weather data in a tracked storm database, and saving an acquired storm intensity profile to a storm database.