A method for simulating realistic DSD fields with given areal rain-rate or reflectivity—application to rainfall disaggregation

Rainfall disaggregation is an important issue in atmospheric, hydrological and climate research. Its primary objective is to downscale (say at 0.5 x 0.5 km²) a given rainfall field at relatively coarse resolution (e.g., 5 x 5 km²) by preserving its spatial structure and some given space-integral value like the areal rain rate or reflectivity for each considered pixel. Not surprisingly, most disaggregation techniques are stochastic, because this allows to generate multiple scenarios for a single input field. The distinctive feature of this approach is that each generated field is different at the fine scale but indistinguishable from the original at the coarser scale.

So far, most of the disaggregation techniques have been focusing on rain rate of reflectivity measurements. Few attempts have been made to include the drop size distribution (DSD) into the disaggregation process. In this presentation, a stochastic DSD simulator is used to generate realistic DSD fields at high spatial resolution (0.5 x 0.5 km²) with realistic spatial structures which all preserve a given reflectivity value at coarser scale. The conditioning values and resolution are set from the output of numerical weather prediction models, ground based weather radar or satellite measurements. As an illustration, examples of simulated DSD fields for different rainfall regimes and spatial structures are given for a GPM footprint of 5 x 5 km².