Download Folie 1 - ACTRiS-2

Uncertainties and recent improvements of liquid water path
observations by ground-based microwave radiometers with
some examples from long-term Cloudnet station records
Pospichal, B. 1, U. Löhnert1, N. Küchler1, and J. Bühl2
1Institute
of Geophysics and Meteorology, University of Cologne, 2Leibniz Institute for Tropospheric Research (TROPOS), Leipzig
1. Importance of cloud liquid water path observations
4. Conditions for (sub-)adiabatic clouds
One major parameter when describing cloud properties is the amount of
water which is contained in a cloud, the so-called cloud water path which
can be separated into liquid (LWP) and ice water path (IWP). Groundbased microwave radiometers (MWR) have been widely used for
measuring LWP, since these instruments can directly provide this quantity
from the observed emission caused by cloud droplets. In the frequency
range used (below 100 GHz), emission from ice particles can be neglected.
However, in the presence of drizzle or rain, the measurements get
unreliable due to the additional presence of scattering on the larger
particles.
For ideal clouds, the liquid water content (LWC) can be described by the
adiabatic liquid water content which corresponds to the amount of
condensation by cooling a saturated atmospheric parcel. In many studies,
clouds are assumed to follow the adiabatic LWC or have a constant subadiabatic LWC. We can show from Cloudnet observations that real clouds
are mostly subadiabatic with a strongly varying sub-adiabatic factor fad (Fig.
3). The degree of adiabaticity for all single-layer liquid water clouds varies
between the stations and between the seasons which might be caused by
different climatic regimes. On the other hand, differences between the
years are an indicator for retrieval or calibration uncertainties. (Fig. 3c)
Within Cloudnet, MWR are a crucial component of the measurement setup
to categorize and observe cloud properties. Therefore, we provide an error
characterization and suggest possible improvements for measuring LWP.
a
b
c
2. Liquid water clouds in Cloudnet
The Cloudnet program provides continuous cloud classification at stations
where a combination of ground-based remote sensing observations
(millimeter cloud radar, lidar ceilometer and MWR) are available. The LWP
observations from MWR are there a crucial part for analyzing liquid clouds.
Below, an example for a Cloudnet target classification is shown (Fig. 1)
where a persistent liquid cloud can be observed between 06 and 14 UTC.
In this study, we focus on pure liquid clouds, i.e. observations when only
cloud droplets (light blue in Fig.1) were classified throughout the whole
atmospheric column. These clouds are determined by their boundaries,
detected by lidar (cloud base) and radar (cloud top) as well as their LWP.
The adiabatic LWP between cloud base and cloud top can be calculated
and then compared with the observed LWP. It can be shown that the
degree of adiabaticity can vary strongly.
Fig. 1: Cloudnet target classification for LACROS site in Krauthausen during HOPE campaign around Jülich.
Fig. 3: Statistics of Adiabaticity (sub-adiabatic conditions) for different stations (a), for different months at Leipzig (b) and
different years at Leipzig (c).
One indicator for sub-adiabatic conditions is the
cloud depth. Thicker clouds tend to be less
adiabatic because of more entrainment processes.
Fig. 4 (right): Observed (scaled) LWP as function of the adiabatic LWP between
the cloud boundaries detected by ceilometer and radar for all single layer clouds
at Leipzig. Adiabatic conditions are shown by black line.
5. Reduction of Uncertainties for MWR observations
LWP is usually derived by statistical methods from observed brightness
temperatures. Most MWR use the water vapour absorption band at 22.235
GHz and a window region around 30 GHz to retrieve LWP with a statistical
accuracy of about 25 g m-2 (Fig. 5a). Other errors include systematic
biases due to calibration uncertainties as well as instrument failures. In any
case, a detailed error characterization as well as careful calibration
monitoring is crucial for accurate LWP observations.
Fig. 5: Left: Theoretical LWP retrieval
performance using K-Band channels
only. Right: adding of 89 GHz
observations to LWP retrieval
3. Cloud statistics
Cloud frequency depends on the location (see Tab 1 for the stations and
time periods used for the comparisons). The distribution of three main
variables is shown in Fig. 2. Cloud thickness is generally quite uniformly
distributed with slightly thicker clouds at maritime sites. Cloud base height,
however, tends to be rather variable. Liquid water path shows different
distributions, especially for Potenza and Jülich which have higher average
LWP than the other stations.
The newly developed W-Band radar (Fig. 6a) includes a 89 GHz passive
channel which can increase the information content to substantially
improve the LWP observations (Fig. 5b), in addition to the usual MWR
observations (Fig. 6b) at JOYCE in Jülich. We are furthermore improving
the automatic quality control of the MWR data and will serve as a
calibration center for MWR.
.
http://joyce.cloud
Fig. 2: Probability distributions of cloud thickness, cloud base height and LWP for the stations and time periods
mentioned in Tab. 1
The differences can be partly
explained by climatic conditions.
Another
important
source
for
systematic biases are observation
and calibration errors which can lead
to substantial differences in derived
cloud properties. The MWR errors
will be discussed here.
Station
Time period
Days
Cloudy 30-second
periods (per day)
Chilbolton
2003-2014
2344
531801 (227)
Mace Head
2009-2013
908
214432 (236)
Lindenberg
2004-2015
3459
636444 (184)
Leipzig
2011-2016
938
183865 (196)
Potenza
2009-2014
910
143037 (157)
Jülich
2011-2016
1746
407689 (233)
Tab. 1 Stations and time periods used for comparison plots
ACTRIS General meeting, Granada, 1- 3 February 2017
Fig. 6: Left: New W-Band radar with 89 GHz channel at JOYCE, right: HATPRO MWR.
 Cloudnet products depend on accurate observations
 Comparison of stations and periods (sub-adiabatic
conditions)
 LWP from Microwave radiometers one uncertainty source
 LWP can be improved by adding more frequencies