Quantifying precipitation intermittency for Bergen, Norway, from measurements and models across a wide range of time scales

Intermittency of rainfall is an important property, for example in the context of urban flooding. There is currently a lack of information about the ability of numerical weather prediction models to represent precipitation intermittency for different weather situations, in particular at high resolution in space and time. Here we present a new way to quantify rainfall intermittency based on a near-continuous, high-resolution precipitation dataset from Bergen, Norway, one of the rainiest cities in Europe. 

We quantify precipitation intermittency from a precipitation dataset acquired at the Geophysical Institute, Bergen, spanning the period 2019-2022 at a 1 min time resolution. Precipitation rates were obtained from a Total Precipitation Sensor TPS-3100 (Yankee Environmental Systems Inc., USA) and a Parsivel2 disdrometer (OTT Hydromet GmbH, Germany). In addition, we use precipitation output at 1 min resolution from the regional high-resolution weather forecasts model HARMONIE-AROME for selected events. Precipitation intermittency is then identified for a range of minimum inter-event times (MIT) from 1 min to 24 h, and precipitation event durations from 1 min to 33 days. Next, the precipitation events for different intermittencies are related to average meteorological characteristics during the events with respect to air temperature, pressure, wind speed, rain rate and amount, and corresponding weather regimes.  

We compile the intermittency information into a 2-dimensional heat map that can be considered as a characteristic fingerprint for precipitation in Bergen. Particular frequency maxima and minima appear to be related to different precipitation processes and weather regimes. A scale gap between 30 min and 2 h event duration for MIT larger than 12 h indicates that separate factors control precipitation processes at these time scales. Weather regimes show a clear influence on the precipitation characteristics, with a markedly higher probability for long-duration rain events in the zonal flow regime for longer event durations at high MITs compared to the Scandinavian trough regime. A comparison between precipitation intermittency simulated by HARMONIE-AROME shows reasonable agreement with observed event characteristics for events lasting more than 1h, while events with durations of 30 min and less are poorly represented.