Cloud Top Height Retrieval From MIPAS
Jane Hurley, Anu Dudhia, Graham Ewen, Don Grainger
Atmospheric, Oceanic and Planetary Physics, University of Oxford
[email protected]
ABSTRACT
MIPAS is an infrared limb-sounding Michelson interferometer onboard the ENVISAT satellite. At low tangent heights, clouds are frequently detected in the
field of view (FOV) and, when retrieving profiles of atmospheric composition, cloud-contaminated spectra are usually excluded. However, clouds themselves
are of great interest scientifically, playing an important role in the Earth’s radiation budget. The radiative effects of clouds depend upon both their micro- and
macro-physical properties, such as cloud top height, cloud depth, particle number density and effective radius. Here we present results of an investigation into
the retrieving cloud top height (Ctop) from MIPAS spectra. In this preliminary study we have used simple models of clouds in the infrared to effectively assume
thick, flat clouds to retrieve the cloud top height with a high vertical resolution of ± 0.25 km.
MODELS AND DETECTION METHODS
CASE STUDY
Colour Index (CI)
These methods for the retrieval of Ctop were applied to a set of Level 1B MIPAS
data from 1-8 August 2003. Fig. 1 shows the Ctop reported by the CI Method and
Fig. 2 compares the Ctops resulting from the three methods. The three methods
give corresponding Ctops with minimal scatter, but the PACT and RIACT
Methods give Ctops within 0.25 km of each other, implying a finer resolved Ctop.
CI work on the principle of radiance ratios between two microwindows
which respond differently to cloud. Colour index CI = LMW1/ LMW2.
•Cloud presence is determined by setting a threshold, below which it is
said to be cloudy and above which it is said to be cloud-free.
•A band: MW1 788-796 cm-1 and MW2 832-834 cm-1
threshold of 1.8. (Spang et al. 2004)
Planck Approximation of Cloud Top Height (PACT)
Assume the cloud can be modelled as a blackbody and neglect the
radiance contributed by the atmosphere.
Fig. 1: Ctops reported by CI Method.
Identify cloudy spectrum by CI Method
Fig. 2: Comparison of Ctop results.
Looking at an individual case at 17:22:03 on 4 August 2003, the CI Method flags
cloud at 13.17 km. Fig. 3 shows the PACT and RIACT Methods’ results of 12.19
km and 12.42 km respectively. Furthermore, the RIACT simulated spectrum does
a fairly good job of modelling the measured spectrum, with a mean difference
between the two of 18 nW/cm2 sr cm-1 (noise ~ 50 nW/cm2 sr cm-1).
At this tangent height, partition FOV vertically into 40 divisions
zi, at a resolution of 0.1 km
Interpolate temperature at each height partition T(zi) using
corresponding L2 temperature profile. Assume brightness
temperature TBi of a cloud at zi is T(zi)
Calculate radiance emitted by blackbody at each zi by
evaluating Planck function B at TBi in microwindow MW
(960-961 cm-1, most transparent region of A band)
Each zi is a possible Ctop. Integrate over the cloud-filled portion
of the FOV to get the total radiation emitted by the cloud:
Lmod(zi) = ∑ij=0 B(TBi , zj) wj / ∑40j=0 wj
Fig. 3: PACT Method Retrieval (left), RIACT Method Retrieval (right)
These results for Ctop compare well with EUMetSat’s SEVIRI infrared image
over the Indian Ocean taken at 18:00 on 4 August 2003, as shown in Fig. 4.
These results clearly indicate the presence of thick cloud in the area of the case
study, as highlighted by a dark circle. A comparison with EUMetSat’s SEVI
cloud top height meteorological product will be carried out.
Calculate mean radiance of measurements in chosen MW (Lmeas)
and compare this value to those modelled at each possible Ctop,.
When Lmod(zi) ≈ Lmeas the Ctop has been found as zi.
RFM Iterative Approximation of Cloud Top Height (RIACT)
Assume cloud can be modelled as a column of aerosol with volume
extinction coefficient βext = 1 km-1 starting at the Earth’s surface and
extending homogeneously upwards to Ctop.
Identify cloudy spectrum by CI Method
Fig. 4: EUMetSat SEVIRI infrared
image. Copyright @ 2005 EUMetSat.
Fig. 5: Sample EUMetSat SEVI cloud top
height map. Copyright @ 2005 EUMetSat.
CONCLUSIONS AND FURTHER WORK
At this tangent height, partition FOV vertically into 0.25 km
vertically separated levels. Each of these levels is a possible Ctop.
Use Reference Forward Model (RFM) (Dudhia 2005) to simulate
radiance emitted in FOV in the 960-961 cm-1 microwindow.
Compare RMS error for RFM runs at each possible Ctop:
the height for which RMS error is minimized is the Ctop.
REFERENCES
Dudhia, Anu, ``Reference Forward Model Software User's Manual'', www.atm.ox.ac.uk/RFM/sum
Spang, R. et al., ``Colour Indices for the Detection and Differentiation of Cloud Types in Infra-red Limb Emission Spectra'', Advances in
Space Research, 33, 2004.
This preliminary study confirms that Ctop can be successfully retrieved by
modelling clouds as having blackbody-like properties, either by estimating with
the Planck function (PACT Method) or by taking βext=1 km-1 (RIACT Method).
Both methods yield Ctops that are well in keeping with the CI Ctop, but that are
almost entirely self-consistent, indicating that the PACT and RIACT Methods
give an improved retrieval of Ctop with greatly increased vertical resolution..
The success of these simple models suggests that other parameters of interest
could be retrieved if further degrees of freedom were introduced into the simple
retrievals presented. Future work includes the retrieval of other parameters, like
cloud top temperature, and comparison with EUMetSat meteorological products.