An extensive range of diagnostics are needed to help monitor and improve numerical models. Our vision was mostly inspired by work carried out in Numerical Weather Prediction centres, but it still is valid in the fields of Environment and Climate.
A recurring issue
The need for diagnostics and verification evolves rapidly because many of the characteristics of the models change at high frequency:
- The delivery of new types of observations from satellites and aircrafts.
- The surface observation network increases in density.
- The creation of new types of forecasts, seasonal, climate and probabilistic forecasts. Commercial applications of weather forecasts require specific user-oriented verification.
- Horizontal and vertical model resolutions increase rapidly with computer power and storage capacity.
- Changes in the computer and software environments.
Because of this ever changing context, researchers and developers spend a significant amount of energy developing new strategies or adapting their software by solving new technical problems. The same technical problems are solved many times over, using different techniques of varying quality. This duplication of effort takes place in different locations, sometimes even in the adjacent office. Many attempts have been made to provide scientists with tools requiring no programming, but these always fail to provide all the flexibility needed. In this rapidly evolving scientific world, this approach to information technology from the 1990’s is bound to fail, because it cannot adapt quickly enough to change.
A working solution
Research scientists have to program at some point in their studies, so why not make it easier for them by providing them with versatile high-level software tools hiding technical details? This enables them to react to change but still focus on their field of expertise. This is an important step, because it helps with two difficult problems:
- the model versus diagnostics race
- the handover of software from research to operations
Differences in computer and software environments across different sites can be made irrelevant by the high-level software tools and even the implementation of new scientific methods could be shared. In the past many good tools have been impossible to share as they were too specific to their producers' environment.
Some scientists may have extended technical knowledge whilst someone with a technical background may have significant insight into a scientific discipline. Communication and productivity improve if these two groups of people collaborate on the same team. This aspect is far too often overlooked, every organisation needs people who can liaise between groups for inter-disciplinary projects because they understand different languages.
We aspire to take this role in a collaboration between different geoscience organisations to share technical and scientific expertise. We aim to liaise between different contributors to obtain a programming framework with its applications to be centralised and shared by the community. This will allow NWP centres to share their workmanship and financial resources.
A shared set of tools dealing with standardised data formats also facilitates the spread and the strength of those standards. In the field of Meteorology, shared application would enforce WMO (World Meteorological Organisation) data exchange standards. Moreover, freely available applications are very efficient vehicles for expert training and technology transfers between organisations.