Climatology of extreme rainfall from rain gauges and weather radar

Extreme rainfall events can have a large impact on society and can lead to loss of life and property. Therefore, a reliable climatology of extreme rainfall is of importance, for instance, for the design of hydraulic structures. Such a climatology can be obtained by abstracting maxima from long rainfall records. Subsequently, a probability distribution is fitted to the selected maxima, so that rainfall depths can be estimated for a chosen return period, which can be longer than the rainfall record. In this thesis, the Generalized Extreme Value (GEV) distribution is used to model annual rainfall maxima.

Using a new methodology, rain gauge data from 12 stations in the Netherlands are employed to derive rainfall depth-duration-frequency (DDF) curves, which describe rainfall depth as a function of duration for given return periods. Often, uncertainties are not incorporated in the design of hydraulic structures, which can lead to a risk of under design. Therefore, uncertainties in the DDF curves are estimated as well.

Weather radars are widely used in real-time quantitative precipitation estimation over large areas with high temporal and spatial resolutions not achieved by conventional rain gauge networks. A 10-year radar-based climatology of rainfall depths for durations of 15 min to 24 h is derived for the Netherlands. Since radar data can be vulnerable to a number of errors, they are adjusted using rain gauges. Verification shows that the radar data set has a high quality.

In general, only few digitized time series from rain gauges are available for subdaily durations. This hampers the study of regional variability in extreme rainfall and the estimation of extreme areal rainfall, which can be overcome by using weather radar. The climatological radar rainfall data set is utilized to obtain annual rainfall maxima for durations of 15 min to 24 h and the size of a radar pixel. GEV distributions are fitted to these annual maxima. For most durations, significant regional differences in extreme rainfall in the Netherlands are found. Subsequently, rainfall DDF curves are constructed. The radar-based extreme rainfall statistics are in good agreement with those obtained from rain gauges, although an underestimation is found for short durations. The uncertainties in radar-based DDF curves are small for short durations and become rather large for long durations.

Employing the climatological radar data set, annual maxima are obtained for area sizes ranging from a radar pixel to approximately 1700 km2. A single equation is derived from which rainfall depths for a chosen return period and area size can be calculated for different durations: the areal DDF curve. Extreme areal rainfall statistics based on rain gauge data agree well with those derived from radar data.

The main result of this thesis is that, after adjustment with rain gauges, weather radar data can be used to derive a climatology of extreme (areal) rainfall including the uncertainties and can be used to study regional differences in extreme rainfall.