A major uncertainty in numerical weather prediction (NWP) and climate prediction is the response of clouds and precipitation to changes in aerosol concentrations. Previous intercomparison work has shown that the simulation of precipitation for a given cloud drop number concentration (Nd) and the response of precipitation to changes in Nd in cloud resolving models (CRMs) and idealised kinematic frameworks is very sensitive to both the physical representation of microphysics and numerical complexity of the scheme cloud microphysics (e.g. ). The purpose of this microphysics intercomparison project is to compare detailed size resolved and bulk parametrised microphysics schemes to understand how they simulate aerosol-cloud-precipitation interactions. The first phase of this project employs the Kinematic Driver model (KiD) to compare microphysics schemes when considering the simulations of and the response of warm rain to changes in N_d. Further this intercomparison will investigate the role of in-cloud processing of aerosol

Overview of the KiD-A project

The overarching aim of KiD-A is to use the Kinematic Driver model (KiD) to compare detailed and bulk microphysics schemes in a dynamically consistent framework without dynamic feedbacks, which have complicated the interpretation of previous intercomparison projects. The main aims are:

  1. Undertake the first kinematic intercomparison of detailed microphysics schemes, i.e. size resolved bin microphysics schemes, superdroplet schemes and 2D aerosol-cloud schemes
    • This will be a sanity check to make sure that the schemes, which are consistently used to develop simpler bulk schemes produce similar results, when forced with the same dynamics.
    • Tests will be performed for a range of initial aerosol concentrations and 1D and 2D kinematic cases
    • Tests will exclude in-cloud aersol processing and focus solely on the precipitation processes, timing and amount
  2. Examine and compare in-cloud aerosol processing from both detailed microphysics and bulk microphysical representations.
    • This stage of the project will compare the detailed and bulk microphysics schemes that can include in-cloud aersol processing.
    • This will be based on only the 2D kinematic cases.
    • The results from this stage will be used to benchmark present modeling capability when considering aerosol-cloud-precipitation interactions.

Kinematic Driver Model (KiD)

The KiD model is the computational and numerical basis the KiD-A project. The source code is available from the [main KiD-A github repository](( and can be obtained by either of the following methods

It is expected that participants will add their microphysics scheme to the KiD model. This will involve writing an interface that couples the generic KiD prognostics to the microphysics variables. The documentation describes how to add a scheme to the KiD framework, while the Morrison, Thompson and TAU provide schemes are provided as examples of adding a microphysics code to the KiD model. In general, there is no need to change the microphysics code so that it will work with the KiD model. Further, diagnostic calls may be required in the microphysics code so that process rates and precipitation rates are output. Otherwise, all required diagnostics will be automatically output once a scheme is successfully coupled. The required diagnostics are discussed below.

Please do not submit results from the provided schemes (i.e., Morrison, Thompson or TAU schemes) unless you have modified the codes to enhance/change the aerosol-cloud or precipitation functionality.

KiD-A intecomparison testcases

The KiD-A intercomparison uses following testcases, which can all be found in KiD testcase directory

1D and 2D kinematic cases

Aerosol specifications for 1D and 2D case

In both cases aerosol are assumed to be soluble ammonium sulfate particles. The initial aerosol distribution is assumed to be a single mode lognormal distribution with a lognormal geometric mean diameter = 0.08 * 10-6m and a log standard deviation = 1.4. The initial aerosol number concentrations (Na) are defined in the test case descriptions below. These parameters are defined in the test case namelists. If a participants model does not include aerosol, please set the initial cloud drop number concentration (Nd) to the initial Na defined below.

1D case

2D stratocumulus (Sc 2D)

Box model tests with KiD

In order to understand potential differences between the detailed schemes, we propose running some simple box cases which will test the evolution of the drop size distrbution resulting from 1) collision-coalescence and 2) condensational growth. The aim of both of these tests is to prescribe the initial cloud drop size distribution using bulk variables of total mass, number and shape of the distribution, so that all models start from the same point. Once initialised the cases either test collision-coalescence or condensational growth, to understand how different schemes evolve given the same starting point. To implement these test each scheme will need to be able to prescribe gamma drop size distribution and discretize this onto their respective bins or into lagrangian parcels. The routine src/init_bins.F90 has been developed to do this for the 2-moment TAU scheme, and this routine has been adapted and applied to the 2d-bin scheme and a lagrangian model.

To build the box model case type

make COMPILER=gfortran CASE=ICMW_BOX all

Box - Condensational growth

Box - Collision-coalescence growth


The example diagnostic outputs (see links above) present standard diagnostic output for the KiD model. Once a microphysics scheme is coupled with the KiD framework, we believe that all cloud microphysics fields accept the process rates and precipitation rates will be automatically output. The participant will need to add a save_dg call to their scheme to store the process and precipitation rates that are output from their scheme. Details about adding diagnostics to a scheme are presented in the documentation, with examples available in the Morrison, Thompson and TAU schemes that are available with the download.

A diagnostic requirement for all schemes (bulk and detailed bin microphysics) is the provision of autoconversion, accretion and rain evaporation process rates. This requires the definition of a boundary between cloud and rain drops. For these diagnostics, we are suggesting that rates be provided for a 20 and 32 micron (radius) cut-off, where all bins below are described as cloud and all bins above are described as rain. We accept that this is artificial; however, such a cut-off permits comparison with the bulk schemes. If you are unable to provide such partitioning and process rates, or if you want to provide a best estimate partition, please do so but also provide a definition of the partition.

In addition to cloud microphysics fields, we also require the following aersol diagnostics from schemes that include aerosol and aersol processing.

Once diagnostics files are available and participants are ready to submit, participants should contact Adrian Hill ( and Zach Lebo ( ).