Daniel Watters *1,2, Alessandro Battaglia 1,2,3 & Richard P. Allan 4
1 Earth Observation Science Group, Department of Physics and Astronomy, University of Leicester, Leicester LE1 7RH, UK; dcw17@leicester.ac.uk
2 National Centre for Earth Observation, University of Leicester, Leicester LE1 7RH, UK
3 DIATI, Politecnico di Torino, Torino, Italy
4 Department of Meteorology and National Centre for Earth Observation, University of Reading, Reading RG6 6BB, UK
The NASA Precipitation Measurement Mission’s 20-year observational record (June 2000 – May 2020) is used to evaluate the diurnal cycle of precipitation from state-of-the-art Coupled Model Intercomparison Project (CMIP6) models and ECMWF Re-Analysis 5 (ERA5, which combine model simulations with observations via data assimilation). NASA’s global-gridded product (IMERG – Integrated Multi-satellitE Retrievals for Global Precipitation Measurement), which is generated by combining spaceborne radar, microwave radiometer, infrared sensor and ground-based gauge measurements, provides high spatial and temporal resolution (0.1°, half-hourly) estimates that are ideal for assessing model performance. IMERG estimates are coarsened to the respective spatial and hourly resolution of the CMIP6 and ERA5 products, and their respective diurnal parameters (mean, amplitude and time of maximum) are compared across multiple decades of June-July-August (IMERG: 2000-2019; CMIP6: 1979-2008; ERA5: 2000-2019) in the 60°N-S range. The diurnal amplitude and time of maximum are relatively consistent whether the multi-decade or the coincident 2000-2008 period is used. Key findings include that CMIP6 and ERA5 lag behind the 16-24 local solar time of maximum precipitation observed over land regions, where diurnal variability is stronger compared to over ocean. CMIP6 and ERA5 have weaker diurnal amplitudes over ocean compared to IMERG observations, with areas of underestimate and overestimate over land. In particular, different CMIP6 models are found to compare differently to IMERG observations, with the NCAR model having a smaller lag in the time of maximum over land compared to the CNRM model. This evaluation can aid in identifying and addressing systematic errors in models to improve their realism.
Fig 1. Local solar time (LST) of maximum precipitation for June-July-August from a) IMERG observations (2000-2019, 0.25° x 0.25°) and b) CMIP6’s CNRM-CM6-1-AMIP simulation (1979-2008, 1.40°x1.41°). Shaded regions are those that are arid (daily mean precipitation less than 0.275 mm) or have weak diurnal amplitudes (amplitude less than 30% of the mean).