In the past two decades, track forecast errors for TCs have decreased, but errors in intensity and precipitation forecasts have stayed constant. Through their impacts on the distributions of latent heating that in turn can affect storm dynamics, microphysical processes can impact the intensity of TCs and hence must be well represented in mesoscale models simulating TCs. Cloud parameterization schemes in mesoscale models make assumptions about the shape of assumed exponential and gamma size distributions for different ice species, such as graupel, snow and cloud ice. These assumptions impact the rates of processes such as riming, aggregation, sublimation, melting, evaporation and collection of hydrometeor species by another, thus affecting the spatial distributions of latent heating generated, which in turn can affect the simulated intensity.
In-situ observations of cloud particles were obtained during the NASA African Monsoon Multidisciplinary Analyses (NAMMA). Ten second averages of size distribution functions were calculated from the cloud and precipitation imaging probes. Observed size distributions of ice particles (125 μm < D < 3 mm) were fit to gamma functions using a non-linear fitting routine. These fits to observed size distributions were used to determine the dependence of the slope, shape, and y-intercept of gamma distributions that describe the hydrometeor size distributions on temperature, total water content, vertical velocity and stage of TC evolution. As NAMMA sampled a tropical storm, a tropical depression, 2 tropical waves that developed into hurricanes and 3 that did not, the observations provide a unique data set to determine whether constant relationships can be used to describe hydrometeor distributions in terms of model prognostic variables, or whether dependences on additional environmental parameters need to be considered.