Preface

This User Guide to the ECMWF Products is not like most other "user guides", which provide clear and straightforward instructions how to "plug in", "get started", "execute" and "switch off". Nor is this Guide a handbook in NWP, dynamic meteorology or weather forecasting; the aim of the Guide is to facilitate the use of traditional ECMWF medium range forecast products, and encourage the use of newer, more advanced products such as the wave forecasts and the ensemble forecasts in the medium-range (EPS), monthly and seasonal forecast.

After a short overview of the history of NWP and a background to the creation of the ECMWF, there follows a non-technical description of the forecast system, including the data assimilation. For the same reason that it is possible to drive a car without knowing exactly how the automatic gearing system works, so is it possible to make use of the output from a NWP forecast system without knowing the crucial mechanical details. Strong and weak sides of the forecast system will be addressed in a special chapter.

After presenting the most common forecast fields the guide will address issues related to the interpolation and plotting of meteorological fields. A chapter on forecast verifications discusses the problem of forecast quality. The interpretation of statistics is never trivial: what at first sight appears to be "good" might be "bad", and what looks "bad" might turn out to be "good".

There are some basic principles of interpreting deterministic NWP products, in particular medium-range forecasts. Most important of these is the relation between the atmospheric scale and its predictability. Another important aspect is forecast "jumpiness", which is normally seen as a nuisance. It can, however, be used in a productive way to indicate possible alternative forecast development and thus serve as an introduction to the EPS.

The EPS is sometimes portrayed as a new, revolutionary way of making weather forecasts. Rather it is a logical development of traditional weather forecasting. The aim of weather forecasting is not to predict meteorological parameters for their own sake, but to provide input in decision making processes. Forecasters have always been trying to tell what is most likely to happen, what might happen and what will probably not happen. The optimum way to convey this information is in terms of probabilities.

The EPS provides an overwhelming amount of probability information and offers an almost unlimited combinations of products. It is a challenge for the forecaster to convey the relevant parts of the EPS information to the end-customers or the public, taking the system's shortcomings into account. For this purpose automatic processing of the forecasts products becomes increasingly necessary. One chapter is therefore devoted on discussing some typical problems of statistical calibration of the NWP output.

A fairly new field of great potential value is forecasts beyond ten days. Although detailed day-to-day weather forecasts might not be possible beyond a week, experiments have shown that there at present is some skill in forecasting the large scale air mass patterns up to two weeks. The ECMWF monthly and seasonal forecasts try to go even further by taking advantage of the mutual interaction between the atmosphere and the oceans.

Acknowledgement: This Guide is the fruit of several years of discussions with scientists at EMWF and weather forecasters, both from Europe and elsewhere. It has been the interaction with these two specialized groups, trying to draw the scientific consequences of the forecasters' experiences, trying to see the practical implications of the scientific exploration, which has been the main driving force and inspiration for this publication.