A simple and pragmatic method utilising the difference between analysed near-surface and Meteosat IR temperatures (Δ T) is presented and applied with the aim of identifying and removing non-precipitation echoes in weather radar composite imagery. Despite inherent deficiencies in these multisource data, such as lower spatial and temporal resolutions relative to the radar data, Δ T is demonstrated to efficiently identify efficiently those areas void of potentially precipitating clouds, and to remove radar echoes in them. A set of 243 manually analysed composites from the summer of 2000 was used to evaluate the method. False alarm rates (FAR), percent correct (PC) and Hanssen-Kuipers skill (HKS) scores were calculated from standard contingency tables for five echo classes: weak, strong, land, sea, and all. FAR was lowered in all classes, PC was generally raised by a few percent to be over 95%, while HKS either remained unchanged or was slightly lowered through the application of Δ T. These results indicate that Δ T successfully removes a significant amount of non-precipitation, sometimes at the expense of a small amount of true precipitation. This penalty is larger over sea, which indicates that the method may need to be tuned differently for land and sea environments. This method may act as a foundation on which improvements to radar data quality control can be made with the introduction of new and improved satellite instrumentation such as that found on board the Meteosat Second Generation platform. However, this type of method should remain complementary to improved signal processing and radar data analysis techniques.

Rainfall events in the United Kingdom during the twentieth century have been surveyed and those identified as extreme by the Flood Studies Report (1975) standards have been examined for common features. Events of duration up to 60 hours were considered in order to investigate those that could cause flash floods. More than half of the 50 events identified were short-period convective storms. The rainfall events were classified by meteorological situation, location and season, allowing the identification of conditions under which extreme rainfall occurred.

Suitable conditions for extreme rainfall were split into three categories: orographic, frontal and convective. The frontal and convective classes were then divided into two sub-classes according to whether significant embedded instability was present in the frontal cases and the nature of the convection in the convective cases. The study revealed a lot of commonality between the cases. For example, all of the orographic events occurred in winter in moist west to southwest airflows, and 80% of the frontal cases involved a slow-moving depression to the south or east and also a slow moving frontal system. A conceptual airflow diagram has been developed for some of the frontal cases. The key result, however, was the discovery that each category of meteorological situation occupied a unique space in a rainfall amount versus duration diagram for each extreme event. This offers exciting opportunities for applying the results of this study and a framework for studying future events.

Estimates of the downward global solar and long-wave radiations are commonly made using simple models. We have tested the estimates produced by a number of these simple models against the values predicted by the radiative transfer model used in a climate model in order to determine their suitability for global applications. For clear sky, two simple models were comparable, but under cloudy conditions a combination of a clear-sky model based on the Angstrom-Prescott equation (which deals with the downwelling solar radiation) with a cloud transmissivity utilising total cloud fraction proved best. The lowest root mean square errors were 27 W m−2 for clear-sky global solar radiation and 90 W m−2 for cloudy conditions. For downward long-wave radiation in clear-sky conditions, the model of Garratt (1992) performed best with a root mean square error of 24 W m−2. However, in cloudy conditions the model of Idso & Jackson (1969) performed best with a root mean square error of 22 W m−2, and, as it performs nearly as well as that of Garratt (1992) in clear-sky conditions, it is probably the best choice.

In strong baroclinic flows, small errors in the initial conditions may grow rapidly and lead to large forecast errors even at very short ranges. It is therefore of crucial importance that these initial conditions are determined carefully. In this paper we use two different but complementary methods to improve the operational analysis of a storm that caused loss of life and much material damage in France on 27 December 1999. The first method is based on the adjoint technique: an adjoint model-run provides the sensitivity of one aspect of the forecast (the cost function) to the initial conditions. We then apply a correction whose shape is the sensitivity so that the corrected initial state better fits the observations. The second method is based on the Quasi-Geostrophic Potential Vorticity (QGPV) inversion. A misfit between the model analysis and satellite pictures (water vapour Meteosat images, for example) leads to the subjective modification of some of the features of the potential vorticity field. The analysis is modified using the wind and temperature corrections provided by the inversion of the QGPV correction. Both methods significantly improve the forecast and may provide valuable guidance for forecasters in real time.

The clear sky UV-index is expressed as a function of two predictable quantities: the solar zenith angle and total ozone. This function is derived by fitting the measurements of total ozone and the UV-index obtained from two instruments, one in the mid-latitudes and one in the tropics. The shape of the function was chosen so that it represents the essentials of the underlying physics. This new function gives good results for all solar zenith angles between 0° and 90° and a wide range of total ozone values.

A statistical comparison of observations and forecasts using the Regional Atmospheric Modelling System (RAMS) for operations at the Savannah River Site (SRS) in South Carolina is discussed. Simulated and observed values of wind direction and wind speed, collected twice daily for a two-year period from April 1998 to March 2000, are compared in a variety of ways for five different locations in the southeastern United States. Simulated and observed turbulence intensities are also compared for a one-year period beginning in February 1999 for the SRS. Statistical measures used for comparison are the mean absolute error, mean relative error and standard deviation. Directional errors tend to increase with forecast time while speeds tend to be underestimated over the entire forecast range, especially those exceeding 8 m s−1. Seasonally, winds are predicted with greater accuracy during the winter and spring months, when speeds in the southeastern United States are higher and direction is less variable. Vertical turbulence intensity at SRS is slightly underestimated, while horizontal turbulence intensity is slightly overestimated. In general, the statistical measures show that the model produces results with reasonable skill for the wind speed, direction and turbulence.

Neural networks (NNs) have been successfully used in the environmental sciences over the last two decades. However, only a few review papers have been published, most of which cover image processing, classification, prediction and geophysical retrieval in general, while neglecting rainfall estimation issues. This paper reviews, without aiming to be exhaustive, NN approaches to satellite rainfall estimation (SRE) by providing an overview of some of the methodologies proposed. A basic introduction to NNs is provided and the advantages of using NNs in SRE are explained, illustrating how NNs can be used to complement more computational-expensive methods to generate quick and accurate results in near real time. The role of the NNs in statistical-empirical algorithms is also reviewed. The last section aims to generate some discussion through comparing the empirical and deterministic algorithmic approaches and contrasting some of the apparent drawbacks of using NNs with a statistically based view of the satellite geophysical parameter estimation.