Freshwater from precipitation is an indispensable resource for life

and a driver for many biological and geological processes. In this

dissertation we present four papers and an investigation in progress

using conventional and multiscale global climate models with the

long term goal of understanding the physical mechanisms responsible

for the observed precipitation trends and distributions. We use the

multiscale model because, in contrast to conventional models, it has

an explicit treatment of cloud and precipitation processes. We test

the hypothesis that the multiscale approach should, in principle,

generate more realistic estimates of tracer concentration and

convective precipitation fields which are known to be sensitive to

cloud-scale physical processes.

Atmospheric convection associated or not with precipitation produces

vertical transport of surface-emitted compounds which can then affect

convection itself, and hence precipitation, by altering the atmospheric

radiative budget or through cloud microphysical processes. In two

papers we evaluate the multiscale method on simulating the global

transport of passive tracers and dust contrasting model outputs and

in situ observations. We force our climate simulations with

observationally derived meteorology. Therefore we can attribute the

differences between the multiscale and the conventional methods to the

differences in vertical transport. The multiscale model simulates

realistic atmospheric concentrations of passive tracers and is, to a

small degree, better than the conventional model in the lower and

upper troposphere. Comparing global maps of passive tracers and dust

from simulations, we find that the multiscale model simulates higher

concentrations in the upper troposphere. This characteristic should

emerge from the overall more efficient vertical transport from

synthetic thermals in the explicit cloud treatment.

We evaluate the multiscale model and a group of the latest versions of

standard climate models on the intensity distribution of

precipitation. In two papers we focus our analysis on sub-daily

precipitation for a highly convective region and contrast model output

with observationally derived precipitation rates. The multiscale model

greatly improves the simulation of convective precipitation

distribution with respect to its conventional counterpart. A comparison

of the precipitation efficiency between the two models suggests that

the improvement should be due to the higher rate of production of cloud

water from vertical advection from the multiscale model as opposed to

the rate of conversion of cloud water to precipitation from cloud

microphysical processes. Extending the analysis of the precipitation

intensity distribution to other state of the art climate models, we

find that the models which implement mechanisms for inhibiting

convective precipitation at low level of instability are more

realistic.

We present work in progress where the multiscale model is the

tool of choice for understanding the physical processes responsible

for the observed precipitation trends in the region of the Indian

monsoon and for attributing these trends to anthropogenic aerosol

emissions. In addition to the realistic representation of convection and

transport, the multiscale model can simulate better than other models

the intraseasonal precipitation variability thought to be important

for the dynamics of the monsoon.