5.2 Sensitivity studies
5.3.3 Impact of surface albedo on the vertical weighting function
Fig. 5.12a shows a photo of Amazon surface taken from HALO during a flight. Several sur- face types are classified in a small scale such as (1) forest, (2) dry-land, (3) water body, and (4) wet-land associated with different surface albedos. It is conceivable that for measure- ments above this area, the surface albedo will change suddenly in a very small scale along the flight path. Platnick (2000) discussed that changes of the surface albedoρ will alter the wm accordingly. When the underlying surface has ρ = 0 (black surface), the transmitted
radiation will be completely absorbed by the black surface. Conversely, if it is not a black surface (ρ > 0), some portions of the transmitted radiation will be reflected back by the surface to the cloud. In this way, the reflected radiation by the surface will change thewm.
Therefore, it is important to implement a correct surface albedo in the forward simulation. Otherwise, a bias on the retrievedreff might be misinterpreted.
0 5 10 15 τ (b) 0.0 0.1 0.2 0.3 0.4 wm 6.0 6.5 7.0 7.5 8.0 z (km) 1240 nm (measured) 1640 nm (measured) 1240 nm (MODIS) 1640 nm (MODIS)
Figure 5.12: (a) A picture of Amazonian surface taken during the AC-18 flight. Four surface types are classified such as (1) forest, (2) dry-land, (3) water body, and (4) wet-land. (b) wm calculated at λ = 1240 nm (black) and 1640 nm (red). The dashed lines represent wmcalculated by assuming
the spectral surface albedo of forest measured by SMART-Albedometer while the solid lines are by using the MODIS BRDF/Albedo product.
In Sec. 5.1.1, it has been introduced that for simulating the DCC case, the MODIS BRD- F/Albedo (MCD43A3) product is applied. According to Strahler et al. (1999), both MODIS Terra and Aqua are used to generate the surface albedo in 500 meter resolution which combines registered, multi-date, multi-band, atmospherically corrected surface reflectance data from the MODIS and the multi-angle imaging spectroradiometer (MISR) instruments to fit a BRDF in seven spectral bands consisting of three visible bands centered atλ = 460
nm, 555 nm, and 645 nm, and four near-infrared bands centered atλ = 865 nm, 1240 nm, 1640 nm, and 2130 nm. Fig. 5.13 shows the spectral surface albedo derived from the MODIS BRDF/Albedo productρM,λcentered atλ = 645 nm (a), 858 nm (b), 555 nm (c), 1240 nm (d),
1640 nm (e), and 2130 nm (f). The black arrows illustrate the flight leg of HALO when measuring the DCC. From Fig. 5.13, it obvious that the DCC was situated above a hetero- geneous surface. This justifies that assuming a homogeneous surface along the whole leg is therefore inappropriate.
-67.7°E -67.6°E -67.5°E -67.4°E -67.3°E Geographic Longitude -1.3°S -1.2°S -1.1°S -1.0°S -0.9°S Geographic Latitude (a) 645 nm A B
-67.7°E -67.6°E -67.5°E -67.4°E -67.3°E Geographic Longitude -1.3°S -1.2°S -1.1°S -1.0°S -0.9°S Geographic Latitude (b) 858 nm A B
-67.7°E -67.6°E -67.5°E -67.4°E -67.3°E Geographic Longitude -1.3°S -1.2°S -1.1°S -1.0°S -0.9°S Geographic Latitude (c) 555 nm A B
-67.7°E -67.6°E -67.5°E -67.4°E -67.3°E Geographic Longitude -1.3°S -1.2°S -1.1°S -1.0°S -0.9°S Geographic Latitude (d) 1240 nm A B
-67.7°E -67.6°E -67.5°E -67.4°E -67.3°E Geographic Longitude -1.3°S -1.2°S -1.1°S -1.0°S -0.9°S Geographic Latitude (e) 1640 nm A B
-67.7°E -67.6°E -67.5°E -67.4°E -67.3°E Geographic Longitude -1.3°S -1.2°S -1.1°S -1.0°S -0.9°S Geographic Latitude (f) 2130 nm A B 0.0 0.1 0.2 0.3 Surface Albedo
Figure 5.13: Spectral surface albedo derived from the MODIS BRDF/Albedo product ρM,λcentered
at λ = 645 nm (a), 858 nm (b), 555 nm (c), 1240 nm (d), 1640 nm (e), and 2130 nm (f). The black arrows indicate the flight leg during DCC measurements from points A to B.
Changes of thewmdue to different surface albedo assumptions are investigated. For this
purpose, cloud B (see Table 5.2) is chosen to represent a DCC topped by an anvil cirrus. The wm is calculated twice by considering two approaches in determining the spectral
surface albedo ρλ. For the first approach, the spectral surface albedo of forest measured
by SMART-Albedometer ρS,λ during the ACRIDICON-CHUVA campaign which yields a
value of ρS,645 = 0.04, ρS,1240 = 0.30, and ρS,1640 = 0.13. As the second approach,ρM,λ is
employed. For the comparison with the first approach, the values of ρM,λ are averaged
along the selected flight leg which results in a value of ρM,645 = 0.04, ρM,1240 = 0.14, and
ρM,1640 = 0.08. Fig. 5.12b showswm computed atλ = 1240 nm (black) and 1640 nm (red).
The dashed lines representwmcalculated using ρS,λ while the solid lines are usingρM,λ.
When implementing a higher value of ρλ, it is found in general that the maximum at the
more weighted due to the enhanced absorption of the reflected radiation from the surface. Thus, by assuming a higherρλ, the retrievedreffis more influenced by the absorption at the
lower layers. For cloud B with decreasing particle size towards the cloud top, assuming a higher surface albedo will result in a larger retrievedreffthan by assuming a smaller surface
albedo. The opposite result is expected for clouds, where the particle size decreases toward the cloud top, such as adiabatic liquid water clouds.
By means of the wm, the differences in the weighting estimate reff∗ are calculated using
Eq. 5.6. This then allows to quantify how large the discrepancies in the retrievedreff for
implementing the two approaches of surface albedo.reff = 27.5 µm forλ = 1240 nm and reff
= 24.2 µm forλ = 1640 nm are obtained when implementing the measured surface albedo of forest (SMART-Albedometer). On the other hand, when using the surface albedo from the MODIS BRDF/Albedo product, reff = 27 µm for λ = 1240 nm and reff = 24.1 µm λ =
1640 nm are acquired. According to these findings, the two approaches in determining the surface albedo seemingly do not give a significant impact on the retrievedreff of cloud B.
The resulting differences are only 0.5 µm forλ = 1240 nm and 0.1 µm for λ = 1640 nm. Given that the cloud B is optically thick (τ = 15), the radiation is largely attenuated by the cloud itself. Consequently, only small amount of radiation are transmitted to the surface which minimizes the impact of reflected radiation from the surface. The impact is more relevant for thin clouds withτ < 5 (not shown here) but it can be minimized by using wavelengths with a higher absorption, such asλ = 1640 nm.