REMOTE SENSING OF CLOUD ALBEDO FROM BACKSCATTERED
SUNLIGHT IN CLOUDY ATMOSPHERE
A. Hünerbein, R. Preusker and J. Fischer
Freie Universität Berlin, Institut für Weltraumwissenschaften Carl-Heinrich-Becker-Weg 6-10, 12165 Berlin, Germany
ABSTRACT
Clouds play an important role in modifying the earth's radiation budget. One part to understand the cloud effects are at which amount clouds reflect incoming sunlight back to space. An algorithm is developed for estimating the cloud albedo which is applicable to MODerate resolution Imaging Spectroradiometer (MODIS) data. The method is based on narrow-to-broadband conversion and radiative transfer model simulations. The model results provide a basis for the derivation of broadband albedo from top of the atmosphere MODIS measurements by use of an artificial neural network. This product was compared with Clouds and Earth's Radiant Energy System (CERES) which is also onboard the Terra satellite. The results of the comparison show an excellent agreement in terms of spatial means, being lower than 13 W/m² for the European region. The algorithm provides a higher spatial resolution than CERES products. This makes it possible to study the cloud albedo for smaller grid scales in detail.Furthermore, the cloud albedo from MODIS has been used to validate a climate model.
1. INTRODUCTION
To obtain reflected shortwave fluxes from the top of atmosphere (TOA), the measured radiances from a narrow-field-of-view scanning radiometer in a particular sun-earth-observer viewing geometry need a conversion to estimate the TOA shortwave flux. Previous Earth Radiation Budget experiments (e.g., ERBE, CERES) have been using Angular Distribution Models (ADMs) to take into account the angular dependence of the radiation field (Suttles et al. 1988, Loeb et al. 2003). Monitoring the change of radiative energy budget at TOA with a higher spatial resolution was one of the motivations to choose narrowband observations from MODIS.
On board Terra MODIS (King et al. 1992) provides global coverage of the earth every one to two days since 2000. MODIS has 19 spectral bands in the solar spectrum with a spatial resolution of one kilometre at nadir. These measurements have been used to derive the upward shortwave radiation above clouds. An algorithm has been developed on the basis of the conversion of MODIS narrowband radiances to broadband flux to derive the shortwave upward flux above clouds.
To solve the inversion problem of the narrowband reflectance from MODIS data, we used a radiative transfer model.
2. DATA
The MODIS sensor operates on two NASA platforms (National Aeronautics and Space Administration), Terra and Aqua. Terra’s orbit around the earth is timed in such a way that it crosses the equator from north to south in the morning. Level-1b geo-referenced and calibrated data are being provided by the NASA Distribution Active Archive Centre (DAAC) and have been used for this study.
The used radiances from the bands centred at 0.46, 0.55, 0.64, 0.83, 0.91, 0.93, 0.94, 1.24, 1.63 and 2.12 µm have been extracted and converted to visible and near-infrared reflectance. The spatial resolution is approximately 1km at nadir.
The information from the bands at 0.91, 0.93 and 0.94 µm is being used to derive the integrated water vapour content above clouds, which is based on an absorption technique (e.g., Albert et al. 2001; Goa and Kaufman, 1998).
3. METHODS
The algorithm for the retrieval of the solar upward flux is based on a narrowband to broadband conversion. The algorithm follows two steps:
1) a narrowband-to-broadband conversion of a series of cases with radiative transfer calculation and simulation of MODIS observation of these cases
2) an inversion to retrieve solar upward flux from actually observed radiance measurements The steps will be explained in detail in the following:
1)
Radiative transfer simulations are applied to simulate the complete radiation field in a plan-parallel atmosphere using the matrix operator method (Plass et al., 1973, Fell and Fischer, 2001). Gaseous
absorption within the solar spectrum is treated using the k-distribution technique for transmission (Goody and Young, 1989; Bennartz and Fischer, 2000) with the absorption line parameters of the main atmospheric absorbers from the HITRAN dataset (Rothman et al., 2003).
The atmosphere is modelled by homogeneous layers characterized by the single scattering parameters of each scattering component in the atmosphere (molecules, aerosols and cloud particles). The adding-and-doubling model MOMO (Fell and Fischer, 2001) is used to calculate the reflectance in 200 individual narrowbands to cover the entire solar spectral domain including the used 10 MODIS bands.
The radiances are calculated for a discrete set of solar zenith angles, viewing zenith angles and azimuth differences. To represent the natural variability of the atmosphere, arbitrary chosen cases are being
simulated. The cases depend on the physical properties of the clouds like stratification, particle radius, cloud height and optical thickness as well as ground reflection, aerosols and water vapour. The radiation flux was calculated for each case.
2)
The aim of the inversion is to generate an operation that relates the MODIS measurements at all considered bands [I1,..,I10] and the observed geometrical parameters (solar zenith angle [ ], viewing zenith angle [ ], and
the azimuth difference [φ]) as well as surface albedo [s], to upward solar flux [F].
(
I
1,
..,
I
10,
θ
,
φ
,
s
)
→
(
F
)
The inverse model consists of a multi-layer perceptron neuronal network. The simulated dataset is used for neuronal network training. An independently created test dataset is simulated to calculate the accuracy of the neural network. The resulting root-mean-square error of 7%, the bias of 4% and the correlation of 0.98 demonstrates that the neuronal network was correctly trained.
In the following an example illustrates application of the algorithm to a MODIS dataset acquired on 13 September 2001, 10:05 UTC (Figure 1). The shortwave flux at TOA is compared with a RGB-colour composite.
FIGURE 1. RGB-colour composite (left) and the shortwave flux (SWF) at the top of the atmosphere above
clouds from MODIS (right) for September the 13th 2001 at 10:05 UTC
A strong depression was located over Western Europe and another one was located over the Baltic Sea. A belt of clouds (Cu, Sc) covers North Europe (fig.1, left). Strong convection takes place in the warm sector of the depression. This effect can be seen in figure 1 (right), where we find the highest radiation flux.
The developed algorithm has been validated against the measurements from CERES (Clouds and Earth‘s Radiant Energy System) (Wielicki et al. 1996). The CERES instrument was designed to measure the Earth’s radiation budget and is also on board the Terra platform. For our study we use the ERBE-like instantaneous TOA Estimates (ES-8). This is 24-hour collection of scanner fluxes from a single satellite. Fluxes at top of the atmosphere are produced from unfiltered radiances by the ERBE inversion algorithm and ERBE angular distribution models (ADM’s) to correct the anisotropies over an observed scene. The ES-8 also includes the ERBE scene identification.
Both instruments are on the same satellite. That has the advantage that the time shifts between the
measurements is negligible. We have chosen 12 overpasses over Europe for the intercomparison. The first step of our preparation to compare both datasets was to transfer the datasets onto a common spatial resolution. Therefore we used the CERES pointspreat weighting function (Smith, 1994) to smooth the MODIS shortwave flux for each pixel. The geographical information from CERES are used to select the nearest smoothed MODIS pixel for the intercomparison. Additional we used the MODIS MOD35 cloud mask and processed the data like the shortwave flux. The criterions to select a pixel for intercomparison were:
1) ERBE scene identification had to be mostly cloudy and 2) The smoothed cloud mask fraction (MOD35) had to be 1.
In order to evaluate the differences between the CERES and MODIS shortwave flux we have calculated the following statistical characteristics: a bias [12.3 W/m²], a root mean square error [44.2 W/m²] and a
correlation of 0.93. In Figure 2 the 2d histogram of the corresponding scatter plot is shown, which reveals that an excellent agreement is reached. With a bias of 12.3 W/m² the MODIS shortwave flux slightly underestimates the CERES shortwave flux. Further investigations are planned with the so-called CERES SSF fluxes.
FIGURE 2: scatter plot from the comparison of CERES and MODIS shortwave radiation flux [W/m²] at TOA
above clouds
5. EXAMPLE OF AN APPLICATION
One objective of satellite experiments is to validate climate model outputs with the measurements. The developed algorithm is used for an intercomparison with a regional climate model (BALTIMOS). The intent was to evaluate the representation of the shortwave radiation in a cloudy atmosphere, simulated by the BALTIMOS model system. The comparison focussed on the Baltic Sea catchment area. This comparison is being carried out within the context of the BALTIMOS project. The purpose here is to reveal the differences in modelled and observed upward solar radiative flux at the top of the atmosphere. A detailed description of the comparison study is provided in Hünerbein (2005).
First a case study will be described and than the results of the annual cycle will be presented.
Figure 3 illustrates a frontal system over Europe for May the 2nd 2002 at 10:00 UTC from the observation and the simulation. An occluded front had been passing Europe in a bow from north Scandinavia to the Alps. The low-pressure centre at the surface was located over the north-east Atlantic Ocean. A less significant front has lain east of the considered region. Both the observation and the simulation show the occluded front which covers main regions of north Europe. The climate simulation catches the frontal system in general, but the distribution and the position were different to the observation. The easterly front was not simulated at all. Furthermore, it has to be kept in mind that the comparison was undertaken with simulations in a climate mode. Therefore only mean values are comparable.
The modelled and the observed annual cycle are shown in Fig. 4. In Fig. 4 (left) all cloudy cases and in Fig.4 (right) only thick clouds are chosen for the comparison. Information over the cloud types was not available for this study. Therefore we defined a threshold to separate thin clouds, which are associated with cirrus, from all clouds. Dealing with cirrus clouds is always problematic for regional models. The albedo of cirrus is estimated up to 0.4 [Jacobson, 1999, von Storch, 1999]. We used this threshold as an additional parameter for the comparison. The curves of the monthly mean reflect the solar insolation. During the months May, June, July and August the model overestimates the shortwave upward flux above clouds (Fig.4, left). Where as in Fig. 4 (right) the regional model and the MODIS shortwave flux are in good agreement with each other.
FIGURE 3. Shortwave flux (SWF) at TOA (W/m²). Observed (MODIS, left) and modelled (BALTIMOS, right)
SWF for May the 2nd 2002 at 10:00 UTC
FIGURE 4. The annual cycle for the observed and modelled shortwave flux. The bars are the standard
deviation. MODIS values (cross), coupled BALTIMOS (triangle), uncoupled BALTIMOS (square); left) for all clouds; right) without thin clouds.
6. CONCLUSIONS
In this paper, we have introduced an algorithm to derive the shortwave flux at TOA above clouds which is applicable for MODIS. We have validated the derived shortwave flux by comparison with those of the CERES instrument. The overall agreement between MODIS and CERES shortwave flux at TOA above clouds with a bias of 12.3 W/m² is excellent. Furthermore we have shown parts of a validation study with a regional climate model.
ACKNOWLEDGEMENTS
The MODIS data were received from the Dundee Satellite Receiving Station and the DLR-DFD in Oberpfaffenhofen. All CERES products used for this study were obtained from the NASA Langley Distribution Active Archive Center.
This research was funded by the German BMBF under contract number 01 LD 0027 in the framework of DEKLIM/BALTIMOS. Partial support of this research was also given by grants from the “Berlin Programm zur Förderung der Chancengleicheit für Frauen in Forschung und Lehre”.
REFERENCES
ALBERT, P., BENNARTZ, R. and FISCHER, J. (2001). Remote sensing of atmospheric water vapour from backscattered sunlight in cloudy atmosphere. J. Atmos. Oceanic Technol., 18, pp. 865-874.
BENNARTZ, R. and FISCHER, J. (2000). A modified k-distribution approach applied to narrow band water vapour and oxygen absorption estimates in the near infrared. J. Quant. Spectrosc. Radiat. Transfer, 66, pp. 539-553.
FELL, F., and FISCHER, J. (2001). Numerical simulation if the light field in the atmosphere-ocean system using the matrix-operator method. J. Quant. Spectroscop. Radiat. Transfer, 69, pp. 351-388.
GAO, B.-C. and KAUFMAN, J. (1998). The MODIS near-IR water vapour algorithm. Algorithm Technical Background Doc. ATBD-MOD-03, 25 pp.
GOODY, R. M. and YOUNG, Y.L. (1989). Atmospheric Radiation (Theoretical Basis). Oxford University Press, 519pp.
HÜNERBEIN, A., FISCHER, J., LORENZ, P. and JACOB, D. (2005) Comparison of solar radiative flux above clouds from MODIS with regional climate model simulations. In special issue:BALTIMOS – a fully coupled modelling system for the Baltic Sea and its drainage basin. submitted to Theoretical and Applied Climatology JACOBSON, M. Z. (1999). Fundamentals of atmospheric modelling. Cambridge: Cambridge University Press, 656 pp.
KING, M. D., KAUFMAN, Y. J., MENZEL, W.P. and TANRÉ, D.(1992). Remote sensing of cloud, aerosol, and water vapour properties from the Moderate Resolution Imaging Spectrometer (MODIS). IEEE Trans. Geosci. Remote. Sensing, 30, pp.1-27.
LOEB, N. G., MANALO-SMITH, N., KATO, S., MILLER, W.F., GUPTA, S. K., MINNIS, P., and WIELICKI, B. A. (2003). Angular Distribution Models for Top-of-Atmosphere Radiative Flux Estimation from the Clouds and the Earth’s Radiant Energy System Instrument on the Tropical Rainfall Measuring Mission Satellite. Part I: Methodology. Journal of Applied Meteorology, 42, pp. 240-265.
PLASS, G. N., KATTAWAR, G. W. and CATCHINGS, F.E. (1973). Matrix operator theory of radiative transfer. 1: Rayleigh scattering. Appl. Opt., 12, pp.14-329.
ROTHMAN, L. S. and CO-AUTHORS (2003) The HITRAN molecular spectroscopic database: edition of the 2000 including updates through 2001. Journal of Quantitative Spectroscopy & Radiative Transfer, 82, pp. 5-44.
SMITH, G. L. (1994) Effects of time response on the point spread function of a scanning radiometer, Appl. Opt., 33, pp.7031-7037.
SUTTLES, J. T., and CO-AUTHORS (1988) Angular radiation models for earth-atmosphere systems. Vol. I. Shortwave radiation. NASA Rep. RP-1184, NASA, Washington, DC, 144 pp.
WIELICKI, B.A. and CO-AUTHORS (1995) Clouds and Radiant Energy System (CERES): An Earth observing System Experiment, Bull. Amer. Meteor. Soc., 77, pp. 853-868.
VON STORCH, H., S. GÜSS, and M. HEIMANN (1999) Das Klimasystem und seine Modellierung. Berlin: Springer Verlag, 255 pp.
