|Title||Rainfall variability in the Netherlands from radars, rain gauges, and disdrometers|
|Author(s)||Beek, R. van de|
|Source||University. Promotor(en): Remko Uijlenhoet, co-promotor(en): Hidde Leijnse. - S.l. : s.n. - ISBN 9789461736437 - 128|
Hydrology and Quantitative Water Management
|Publication type||Dissertation, internally prepared|
|Keyword(s)||regen - neerslag - radar - regenmeters - hydrologie - meteorologie - schattingen - schatting - meettechnieken - meetsystemen - nederland - rain - precipitation - rain gauges - hydrology - meteorology - estimates - estimation - measurement techniques - measurement systems - netherlands|
Chapter 1. This thesis presents studies on the variability of precipitation in the Netherlands from datasets collected by radars, rain gauges and disdrometers. Accurate rainfall estimates are highly relevant in hydrology, meteorology and climatology as precipitation has a large impact on society. Precipitation has been studied extensively in the past, although it is impossible to describe all processes and behavior involved. This thesis attempts to add to the knowledge on precipitation. In the first chapter a short overview of rainfall variability at different scales is presented together with the most common instruments for measuring precipitation.
Chapter 2. The spatial variability of daily rainfall accumulations is studied. Ninety-day averaged semi-variograms are created based on a 30-year data set gathered by automatic stations operated by the Royal Netherlands Meteorological Institute (KNMI). This is complemented by a one-year dataset of 10 gauges within a 5 km radius around CESAR (Cabauw Experimental Site for Atmospheric Research) in the center of the Netherlands. It is shown that it is possible to derive an average semi-variogram that describes the climatology of daily precipitation for each day of the year.
Chapter 3. The study of chapter 2 is extended by investigating accumulation intervals shorter than daily scales. These are at 12, 8, 6, 4, 3, 2 and 1-hour accumulation intervals. It is shown that at shorter temporal scales the behavior of semi-variograms of precipitation still shows a clear seasonal trend. At hourly and two-hourly accumulation intervals the signal of the range becomes fairly constant during the summer due to the limited accumulation period, the frequent occurrence of convective precipitation, and measurement errors. This illustrates the lower limit of using cosine functions to describe variogram parameters. By fitting a power-law function through the different cosine parameters it is possible to describe the semi-variance of precipitation at scales between 1 and 24 hours using a limited set of equations.
Chapter 4. Different sources of error affecting rainfall estimates by weather radar are identified. By focussing on precipitation near a C-band radar some of these sources are reduced, which allows to focus on a limited set of error sources. These are radar calibration, ground clutter, wet radome attenuation and variations in rain drop size distribution. An event that caused high precipitation amounts in a band along the center of the Netherlands and more than 50~mm near the radar between the 25th and the 27th of August 2010 is studied. Without any correction and by applying a standard Marshall-Palmer Z-R relation the radar is found to underestimate by approximately 50% with respect to the rain gauge measurements. Using the sun for calibration a correction of 1 dB is applied. Clutter is corrected by subtracting a clear sky clutter map as this proves to provide better results than the standard doppler filter. Wet-radome attenuation is corrected by looking at the amount of attenuation at a known strong clutter pixel near the radar. Disdrometer data near the radar are used to derive accurate Z-R relations specific to the precipitation of the event. These corrections combined provide very promising results with a slight overestimation of the quantitative precipitation estimates (QPE) from the radar by 5 to 8%.
Chapter 5. An extensive dataset of 195 precipitation events measured by an X-band radar (SOLIDAR) is used to study precipitation at a high spatial resolution of 120 m and a high temporal resolution of 16 s. This study shows the benefit of using such high resolution X-band radars over flat terrain. The errors in the radar measurements are first assessed and corrected as well as possible by considering different techniques. These errors are calibration, ground clutter and attenuation. Finally, five strongly different precipitation events are studied in detail to illustrate the strengths and weaknesses of the X-band weather radar.
Chapter 6. The variability and possible measurement methods of precipitation have been studied. It was shown that precipitation spatial and temporal variability has a clear statistical signal by analyzing variograms for different accumulation intervals. Weather radars were also shown to be able to give good estimates of precipitation at ground level as well as detailed information on the spatial variability. Some recommendations are given to perform follow up studies. For chapters 2 & 3 it is recommended to use a larger and more detailed dataset, which also incorporates Belgian and German data. This would allow the study of anisotropy in the semi-variograms as well as extending the analysis to accumulation times shorter than 1 hour and longer than 24 hours. For chapter 4 it is recommended to study pixels located further away from the radar. While other error sources would become more pronounced it would be possible to study the applicability of the proposed corrections at longer ranges. Studying the wet-radome attenuation with several strong clutter pixels near the radar would allow the study of wind-effects on wet-radome attenuation, possibly allowing corrections using measurements of (Doppler) wind-speed and direction. Finally, in chapter 5 it is recommended to study the successor of SOLIDAR, IDRA, which is currently operational at CESAR. This radar is a polarimetric radar, allowing a more detailed study of precipitation together with the data from other instruments at this location and the C-band radar of KNMI, which is located close to this location at approximately 23 km.