Browse Prior Art Database

Adaptive learning numerical weather forecasting model Disclosure Number: IPCOM000015255D
Original Publication Date: 2001-Oct-10
Included in the Prior Art Database: 2003-Jun-20
Document File: 8 page(s) / 118K

Publishing Venue



Summary and statement of problem

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 13% of the total text.

Page 1 of 8

Adaptive learning numerical weather forecasting model

Summary and statement of problem

The behavior of the atmosphere is influenced by multifarious complex phenomena with scales ranging from the molecular to planetary and from microseconds to years. The weather varies in response to a myriad of atmospheric perturbations in the form of waves, vortices, circulations etc., having life cycles of minutes, days, months, or even longer. A small eddy of the order of a few centimeters dies down in a few seconds. Similarly a tornado of the order of a few kilometers dies down in a few hours. There are other atmospheric systems like monsoons whose horizontal dimensions may be several thousands of kilometers and prevail for a season. The representation of all these systems in single entity is a complex issue, because it involves simulation of systems of various space and time scales. Further, these systems interact amongst themselves nonlinearly. For example, a tornado may pass on some energy to a thunderstorm which in turn pass on to a cyclone and so on. This energy cascading is subtle phenomenon in weather prediction. Moreover there is a scale interaction in single system as well. In cyclones, the movement is governed by large-scale flows. However, intense vertical motions surrounding the eyewall are of small-scale nature. The representation of these two scales in single entity is a difficult task. In order to encompass all the scales of motion, a very high-resolution models concomitant with very high computing power are required.

Furthermore, the behavior of the atmosphere is governed by a set of physical laws, which can be expressed as mathematical equations. These take into account how atmospheric quantities or fields (such as temperature, humidity, wind speed and direction for example) will change their values at the present time. If we can solve these equations, we will have a description of the future state, a forecast of the atmosphere, derived from a current state (initial value), which we can interpret in terms of "weather"- rain, temperature, sunshine and wind. However, these nonlinear partial differential equations have no exact solutions that can give us the future values. The approximate solutions of these equations are provided through numerical means. Representing fields with approximate numerical values is called discretization, which emphasizes the limits of the classical numerical approach. The smaller the set of numbers, the coarser the discretization and the less detail we will have about the future state of the atmosphere. The dynamical numerical weather prediction is an initial value problem, which requires a closed set of appropriate physical laws expressed in mathematical form, suitable initial and boundary conditions, and an accurate numerical method of integrating the system of equations forward in time. The actual numerical prediction is the collection and checking of surface and upper air meteorological variables...