IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 45, NO. 7, JULY 2007 2287
Information Content of Millimeter-Wave
Observations for Hydrometeor Properties
in Mid-Latitudes
Mario Mech, Susanne Crewell, Member, IEEE, Ingo Meirold-Mautner,
Catherine Prigent, and Jean-Pierre Chaboureau
Abstract—For future remote sensing applications the potential
of the millimeter wavelength range for precipitation observations
from geostationary orbits is investigated. Therefore, a database
consisting of hydrometeor profiles from various mid-latitude pre-
cipitation cases over Europe and corresponding simulated bright-
ness temperatures at 18 microwave frequencies was built using the
cloud resolving model Méso-NH and the radiative transfer model
MicroWave MODel. The information content of the database was
investigated by applying simple statistical methods, as well as
developing first-order retrieval approaches. The results show that,
particularly for snow and graupel, the total column content can
be retrieved accurately with relative errors smaller than 25% in
dominantly stratiform precipitation cases over land and ocean
surfaces. The performance for rain-water path is similar to the one
for graupel and snow in light precipitation cases. For the cases with
higher precipitation amounts, the relative errors for rain-water
path are larger particularly over land. The same behavior can be
seen in the surface rain rate retrieval with the difference that the
relative errors are doubled in comparison to the rain-water path.
Algorithms with reduced number of frequencies show that window
channels at higher frequencies are important for the surface rain
rate retrieval because these are sensitive to the scattering in the
ice phase related to the rain below. For the frozen hydrometeor
retrieval, good results can be achieved by retrieval algorithms
based only on frequencies at 150 GHz and above which are suit-
able for geostationary applications due to their reduced demands
concerning the antenna size.
Index Terms—Hydrometeor, millimeter wave radiometry, re-
mote sensing, retrieval, satellite applications.
I. I NTRODUCTION
P
RECIPITATION results from a complex chain of
processes implicating a high variability in space and time.
Therefore, the observation of precipitation with a sufficient
temporal and spatial resolution for many applications, such as
numerical weather forecast models or hydrological purposes, is
still a challenging task.
Manuscript received June 6, 2006; revised April 4, 2007. This work was
supported by EUMETSAT under Contract EUM/CO/04/1311/KJG (“Simu-
lation study of precipitating clouds from geostationary orbits with passive
microwaves”).
M. Mech and S. Crewell are with the Institute for Geophysics and
Meteorology, University of Cologne, 50923 Cologne, Germany (e-mail:
mech@meteo.uni-koeln.de).
I. Meirold-Mautner and C. Prigent are with the LERMA, l’Observatoire de
Paris, 75014 Paris, France.
J.-P. Chaboureau is with the Laboratoire d’Aérologie, Université Paul
Sabatier and CNRS, 31400 Toulouse, France.
Digital Object Identifier 10.1109/TGRS.2007.898261
Traditional precipitation observation systems such as rain
gauges measure precipitation at single points, mostly as accu-
mulated daily sums with an unsatisfactory coverage particularly
over ocean. In well-developed countries ground-based radar
networks provide precipitation observations with a temporal
resolution of about 5 min over large areas with high spatial
resolution. Relating the measured reflectivities of the radar
sampling volume to the surface rain rate can lead to substantial
errors of the order of a factor of 2 [1], particularly at distances
larger than 100 km. Due to the different error factors involved
in the measurement process, a homogeneous quality radar
precipitation composite will not be available in the near future,
even for Europe.
With satellite-based observations in the lower microwave
regions promising results were accomplished over ocean from
polar orbiters (see [2] and [3]). Their algorithms for retrieving
precipitation rates use frequencies below 40 GHz and are based
on the brightness temperature increase through emission by the
precipitation layer over the radiatively cold ocean. An algo-
rithm applicable over ocean surfaces, including the emission
signal and the information contained in the signal influenced
by scattering by frozen hydrometeors above the precipitating
layer, was developed by Bauer and Schlüssel [4]. By using
window channels, the surface rain rate and also the hydrometeor
profiles were retrieved (see [5] or [6]). Over land, the emission-
based approach is not useful, since land surface emissivity is
too large and heterogeneous within the radiometer footprint.
Therefore, to reduce the sensitivity to surface emissivity, higher
frequencies are applied. In the millimeter wavelength region,
information can be gained by examining the signal originating
in the precipitating layer and modulated by scattering by large
frozen hydrometeors above. Since the formation of precipita-
tion in the mid-latitudes mostly follows the cold rain process
in the solid form as snow and graupel which melts to rain at
lower altitudes, there is a relation between the microwave signal
originating in the upper atmospheric ice layer and the surface
rain rate. This indirect approach was implemented for the
85.5-GHz window channel in various algorithms (see [7]–[12]).
The potential of millimeter (30–300 GHz) and
submillimeter-wave (300+ GHz) channels for passive space-
borne remote sensing of the troposphere and lower stratosphere
has been numerically investigated by Gasiewski [13]. He found
that high-frequency window channels (340 or 410 GHz) show
a potential for observing clouds with very low liquid water
contents and that these high frequencies can map thick clouds
0196-2892/$25.00 © 2007 IEEE