Histograms of phase of first harmonic of the diurnal cycle (time given in LT) for grid cells in the tropical band (20S-20N) for TRMM (a & d), EC-Earth-LR (b & e) and EC-Earth-HR (c & f). Top row: land and bottom row for ocean grid cells. Only grid cells with amplitudes larger than 0.2mm are take into account.
Histograms of phase of first harmonic of the diurnal cycle (time given in LT) for grid cells in the tropical band (20S-20N) for TRMM (a & d), ECMWF w/o deep convection (b & e) and ECMWF w/ deep convection (c & f). Top row: land and bottom row for ocean grid cells. Only grid cells with amplitudes larger than 0.2mm are take into account. Results using HadGEM-GC3.1 Observations datasets used are: CMORPH (1998-2018), available up to 60°N at 30min frequency, and has been regridded to the N1280 model grid; TRMM (1999-2010), available up to 50°N at 3 hr frequency; and GPM (2015-2019), available up to 60°N at 30min frequency. It should be noted that using explicit convection at the 10km scale is rather unrealistic – this requires the model to uplift a whole gridbox column of air when convection takes place, not at all like the much smaller scales on which real convection happens. However, as a sensitivity study it still has value to suggest where convection schemes have problems, and what the consequences of these problems may be for the climate simulation. More detail of the explicit convection setup can be found in Field et al. (2018). Similar simulations, with an older model configuration, and analysis can be found in Xxxxx et al. (2015). Mean properties of precipitation
