Quality enhancement of quantitative precipitation estimates

Accurate quantitative precipitation estimates are important for a wide range of applications. Therefore KNMI is working on enhancing the quality of its radar- and rain gauge-based precipitation products.

It is well-known that radar quantitative precipitation estimates suffer from several error sources. The aim of this project is not only to find suitable correction methods for most of these, but also to design methods for estimating of the uncertainty of the corrected precipitation estimates. An important aspect of the validation of these methods is that we will not only focus on average performance, but also on the performance with respect to extreme precipitation. This is very important because of the large potential impact such extreme precipitation can have on society.

Part of this project will focus radar-only precipitation estimates. The aim of this part is to remove as many errors from the radar data as possible, using knowledge of the physics of both radar measurement and the precipitation itself. The specific aspects that will be studied are:

  1. Correction for signal attenuation and the uncertainty therein;
  2. Correction for the effects of vertical variation of precipitation and estimation of uncertainties;
  3. Use of uncertainty information from literature and this project in merging data from several radars;
  4. Correction for the effect of fast-moving rainstorms.

Correction of clutter in radar data is part of another project (https://www.knmi.nl/research/observations-data-technology/projects/optimizing-the-use-of-dual-polarization-weather-radar-data). Results of that project are used here to enhance quality, and to estimate uncertainties from the amount of clutter that was removed.

The second part of this project will focus on merging radar precipitation estimates with rain gauge data. KNMI has two rain gauge networks, one with 32 automatic rain gauges that report in real-time, and one with 320 manual rain gauges operated by volunteers that report daily precipitation amounts every day at 8:00 UTC. The challenge is to optimally use both of these networks, with their different densities and temporal resolutions, in precipitation estimates. In addition to the KNMI rain gauge networks, networks operated by others are potentially promising new sources of precipitation measurements, provided that we can determine their quality. The following specific aspects are therefore studied in this project:

  1. Automated real-time quality control methods for rain gauge data, providing both flagging and quality information (see also https://www.knmi.nl/research/observations-data-technology/projects/quantifying-quality-of-crowd-sourced-weather-data);
  2. Optimal methods for merging radar and rain gauge data, depending on the temporal resolution and the density of the network;
  3. Use of quality information in both radar and rain gauge precipitation estimated to optimally merge these data sources.
Figure 1: Sources of error affecting radar measurement of precipitation (courtesy of Markus Peura of FMI).