The Diurnal Cycle of Precipitation and Its Relation with Surface Variables over North America in ISD Station Data and in a Convection Permitting Climate Model.

Convective precipitation is more frequent in the afternoon when the surface temperature is at its highest and favors atmospheric instability. The resulting diurnal cycle of precipitation is therefore an interesting indicator of the representation of deep convection, its intensity and the frequency with which it is triggered. The timing of the convective diurnal cycle is generally better represented by high resolution models using an explicit representation of convection. However, its amplitude is generally exaggerated due to the frequency overestimation of the convection onset and its intensity.

The objective of this study is to investigate how well measured near-surface variables, such as 2 m temperature and humidity, affect the representation of convective precipitation and its diurnal cycle.

The analysis compares hourly station observations of 2 m temperature, humidity and precipitation from the Integrated Surface Database (781 stations used over North America) with 2.5 km simulations of the of the Canadian Regional Climate Model version 6 (CRCM6/GEM5). The analysis is performed for the summer months (June, July, and August) between 2015 and 2020.

First, a clustering method is used to separate stations in groups with similar precipitation diurnal cycle characteristics (amplitude, timing and intermittency). Each group shows quite similar precipitation forcing and for groups with a significant diurnal cycle, the underlying mechanisms such as late afternoon convection or sea breeze can be identified.

Second, northeastern North America stations (250 stations) are compared to simulation data to investigate the dependency of precipitation to 2 m temperature and humidity using an intensity-frequency decomposition method. The model tends to overestimate the precipitation frequency and underestimate its intensity for low-to-moderate values of temperatures and relative humidity (persistent drizzle). However, the overestimation of late afternoon convective precipitation can be explained by an overestimation of the m temperature in the afternoon and a too frequent triggering of convection at equivalent temperature and humidity conditions.

The clustering method trained on observation data is also used on simulation data to provide insights in the underlying mechanisms that are missing or too active in the model. For the studied region, the afternoon convection appears to be strong for a region that is too wide in the model compared with the station data