Improved precipitation forecasts in IFS Cycle 45r1

Richard Forbes


Cloud and precipitation processes are represented in the Integrated Forecasting System (IFS) with parametrizations that capture the gridscale effects of water and ice particle microphysics in the atmosphere. This includes the formation of particles, phase changes and collisions between particles as they fall. The representation of these processes in the IFS has improved significantly in recent years. In the latest upgrade, implemented in IFS Cycle 45r1 operational from 5 June 2018, changes to the parametrized processes for liquid-phase microphysics were made to address a long-standing problem of precipitation maxima just off the coastline and over lakes. This occurred in situations where the warm-rain generation process was active over a prolonged period. The results show significant improvements to the spatial pattern of rainfall and quantitative precipitation forecasts near coastlines and over lakes in these specific situations.

Warm-rain cloud and precipitation processes

Warm rain refers to the process of precipitation production through the collision and coalescence of liquid particles (cloud droplets, drizzle drops and raindrops). In the model, just less than half of the global stratiform rainfall is produced via the warm-rain process, with the rest produced via melting snow particles as they fall through the 0°C level in deeper cloud systems. The warm-rain process dominates in clouds with shallow to mid-level tops, particularly in the warmer seasons when the melting level is high.

Whether a cloud generates rain and the intensity of generated rainfall depends on many factors, including the strength of updraughts in the cloud, the size of the cloud particles, and the lifetime and depth of the cloud. To capture the wide range of observed conditions (from cloud with no rain at the surface, to drizzling stratocumulus, to widespread moderate rain, to heavily precipitating events), the model must capture the appropriate non-linearity of the warm-rain process and all factors that affect the rain generation, particle size distribution, fall of raindrops and evaporation in the sub-cloud layer.

Microphysics upgrade

The focus of the warm-rain microphysics upgrade in IFS Cycle 45r1 is on improved numerics of autoconversion (cloud droplets coalescing to form larger raindrops), accretion (raindrops colliding and coalescing with cloud droplets to form larger raindrops), fall speed and rain evaporation processes to reduce the dependence on threshold values and to make the microphysics parametrization more robust to timestep. Specific changes in 45r1 include:

  • The critical threshold for activation of the warm-rain process is removed. As the threshold was previously different over land points and open water points (sea and gridscale lakes), this change removes the discontinuity in the warm-rain process at the land/water boundary.
  • The numerical formulation of the sedimentation process (fall due to gravity) is modified to allow a change of the terminal fall speed for raindrops from a fixed 4 m/s to a realistic drop size dependence. As observed, larger drops can fall faster than smaller drops, leading to an improved representation of the rain accretion and sub-cloud evaporation processes.
  • The formulation of the warm-rain processes was made more robust to timestep, which allows the model to give similar results across the full range of timesteps used in different IFS configurations without changes in computational cost.

Impact on forecasts

The impacts of the changes in Cycle 45r1 are evident in the frequency distribution of rainfall rates and accumulations, with fewer occurrences of light rain and increased occurrence of heavier rain, generally closer to the observed distribution. However, the most noticeable change for users is in the spatial distribution of rainfall near coastlines and over lakes (resolved by the model grid) in meteorological situations dominated by the warm-rain process and persistent for several hours. The unrealistic rainfall over the water is removed or moves downwind over the land in closer agreement with observed rainfall patterns throughout the forecast range.

Ongoing research is looking at further developments to the representation of both stratiform and convective cloud and precipitation processes in the IFS for further improvements to quantitative precipitation forecasting in the future.

Example of changes to near-coastline precipitation. 24-hour accumulated precipitation (mm) on 15 May 2017 from a 00 UTC 14 May 2017 forecast in a warm-rain dominated south-westerly flow over the UK for IFS Cycle 43r3 (left), IFS Cycle 45r1 (middle) and the radar observations (right).