Existing problems in numerical weather forecasting

There are still many problems to be solved in numerical weather prediction:

Introduction of physical processes at sub-grid scale Since the atmosphere is a continuum medium with a continuous motion scale spectrum, regardless of No matter how high the resolution of the model is, there will always be some motion close to or smaller than the grid interval scale (see common calculation methods in numerical weather forecasting), which cannot be accurately reflected in the model. This motion process is called a subgrid process. Turbulence, convection, condensation, and radiation processes all contain subgrid processes. Parametric methods have been used to consider these processes in numerical prediction, that is, large-scale variables are used to describe the statistical effects of subgrid processes on large-scale motions. Although considerable results have been achieved with this approach, there are still many unresolved issues. For example, parameterization cannot consider the impact of large scales on small scales and their feedback effects, there is a lack of objective determination methods for parameter values, and the model is too sensitive to differences in parameterization.

Numerical solutions of nonlinear equations Although under appropriate conditions, it can be proved that the numerical solutions of some stable forms of linear differential equations can approximately represent the true solutions of the corresponding differential equations, but for nonlinear For differential equations, the two solutions may not be completely consistent. There is evidence that sometimes numerical solutions, although computationally stable, bear no resemblance to the true solution (which is a special case where the true solution is known).

Initial value formation problem It includes problems such as initial value processing, application of satellite data and four-dimensional assimilation (see processing and analysis of numerical weather prediction data). These problems have not yet been well solved.

The above problems are all encountered directly when designing patterns. But the most fundamental thing is people's understanding of the laws of weather evolution, especially the understanding of medium-term and long-term weather processes and the occurrence and development of severe storms, which is still insufficient. In addition, although the use of satellites and remote sensing technology to detect the atmosphere has made a certain contribution to providing data in areas with sparse records, the accuracy of meteorological detection and forecasting still needs to be further improved.