ARM SWS to study cloud drop size within the clear-cloud transition zone (GSFC) Yuri Knyazikhin Boston University Christine Chiu University of Reading Warren Wiscombe GSFC Thanks to Peter Pilewskie (UC) and Connor Flynn (PNNL)
Why is the transition zone? It is difficult to distinguish between cloudy and cloud-free air in remote sensing observations (e.g., Stevens & Feingold, 29). The regions around clouds are neither precisely clear nor precisely cloudy (Koren et al., 28). This problem has major climatic consequences, in particular on aerosol indirect effect studies, which demand a precise separation of clear and cloudy air (e.g., Charlson et al., 27).
ARM Shortwave Spectroradiometer The SWS measures zenith spectral radiance. Spectral Range: 35-27 nm Spectral Resolution 35-97 nm: 8nm 97-27 nm: 2nm Spectral Sampling rate: Hz Field of view:.4 48 wavelengths (adapted from Pilewskie s presentation at the last ARM STM) 2
SZA=45 o May 8, 27 4 6 g/m 2 wavelength (nm) (nm) wvl 8 2 2 6 3 2 4 6 2 8 24 3 time (sec) at 25 sec..25.5.75 zenith radiance 2758 CBH=2km
Consider the whole SWS spectrum SZA=45 o May 8, 27 wavelength (nm) (nm) wvl 4 8 2 2 6 3 2 4 6 g/m 2 6 2 8 24 3 time (sec) measured zenith radiance.3.25.2.5..5 5 cloudy 6 7 8 9 2 2 22 23 24 25 clear 4 8 2 6 2 wavelength (nm)..25.5.75 zenith radiance 2758 4
Consider the whole SWS spectrum wavelength (nm) (nm) wvl SZA=45 o May 8, 27 4 8 2 2 6 3 2 4 6 g/m 2 6 2 8 24 3 time (sec) normalized zenith radiance.8.6.4.2 5 cloudy 6 7 8 9 2 2 22 23 24 25 clear 2758 4 8 2 6 2 wavelength (nm)..25.5.75 zenith radiance 2758 Normalized to an extraterrestrial solar 5 spectrum
Spectral-invariant hypothesis Zenith radiance spectrum in the transition zone is a linear combination of cloudy and clear sky spectra with a wavelength-independent weight I transition (λ) = ai cloudy (λ) + ( a)i clear (λ), a (,), a a(λ) I transition (λ) I clear (λ) = a I cloudy(λ) I clear (λ) + ( a) (i) y(λ) = ax(λ) + b (ii) b = a 6
Checking spectral-invariance wavelength (nm) (nm) wvl 4 8 2 2 3 6 2 4 6 2 8 24 3 time (sec)..25.5.75 zenith radiance 2758 I transition (λ) I clear (λ) = a I cloudy(λ) I clear (λ) + ( a) ratios to clear ratios to clear 2 5 5 6/25 7/25 8/25 9/25 2/25 2/25 22/25 23/25 24/25 2758 35-8 nm 5 5 2 6 5 4 3 2 cloudy to clear ratio (5/25) y =.445 +.86528x R= y =.385 +.6979x R=.99998 y =.64447 +.4227x R=.99982 y =.95884 +.22x R=.99582 y =.973 +.52729x R=.99775 y =.98547 +.32384x R=.99697 y =.9857 +.25497x R=.99776 y =.9936 +.4396x R=.9932 y =.9993 +.4927x R=.96287 35-68 nm 2 3 4 5 6 cloudy to clear ratio (5/25)
Wavelength-independent function a(time) or a(space) wavelength (nm) (nm) wvl 4 8 2 2 6 3 a(t) and I(t,λ)..8.6.4 2758 a(t) [based on 35-68 nm] 44 nm 87 nm 64 nm 2 4 6 2 8 24 3 time (sec).2. 5 6 7 8 9 2 2 22 time (sec)..25.5.75 zenith radiance 2758 from Marshak et al., 29 8
Do radiative transfer calculations confirm this spectral-invariant behavior found in SWS data? Use SBDART to calculate zenith radiance at 4-22 nm wavelengths with a nm resolution Atmosphere mid-latitude summer atmosphere 3 cm integrated water vapor amount default trace gas amount Aerosol Rural aerosol type;.2 optical depth at 55 nm 8% relative humidity Cloud -4 cloud optical depth (defined at 55 nm) km altitude from Chiu et al., ACPD 2 9
Modeling: Whole spectrum Normalized zenith radiance.8.6.4.2 (a) B B2 B3 B4 B5 3. 2...5. Ratios to clear 5 4 3 2 (c) B B2 B3 τc=2. τc=. τc=.5 B4 B5 4 8 2 6 2 Wavelength (nm) 2 3 4 5 Cloudy to clear ratio SBDARD zenith radiances Ratios to clear radiance I transition (λ) I clear (λ) = a I cloudy (λ) I clear (λ) + b
Modeling: B and B5 bands B.4-.8 um B5 2.-2.2 um Ratios to clear Ratios to clear B 2 4 6 8 Cloudy to clear ratio 8 6 4 2 45 4 35 3 25 2 (d) (e) τc=2.; (.8,.9) τc=.; (.47,.52) τc=.5; (.25,.74) τc=2.; (.78,4.52) τc=.; (.43,6.33) τc=.5; (.2,5.6) 5 B5 4 42 44 46 48 5 Cloudy to clear ratio Ratios to clear 5 4 3 2 (c) B B2 B3 τc=2. τc=. τc=.5 B4 2 3 4 5 Cloudy to clear ratio B5 Ratios to clear radiance I transition (λ) I clear (λ) = a I cloudy (λ) I clear (λ) + b
Surface albedo (B) Slope function.8.6.4 (b) Black Sand Vegetated Snow Surface albedo.8.6.4 (a) Vegetated Snow Sand.2 Cloud optical depth..2 4 8 2 6 2 Wavelength ( nm) 2
Slope function.8.6.4 (d) Aerosol effect (B) Rural Urban Oceanic Aerosol optical depth Single scattering albedo.3.2..8.6 (a) (b) Rural Urban Oceanic Rural Urban Oceanic.2 Cloud optical depth. Assymetry factor.4.9.8.7 (c) Rural Urban Oceanic.6 4 8 2 6 2 Wavelength (nm)
Different drop size in B and B5 6 4 2 (.2,5.6) B5 Ratios to clear 8 3 2 (.25,.74) (.2,.82) (.2,2.28) B 4µm;B 4µm;B5 6µm;B 6µm;B5 5 4 5 6 Cloudy-to-clear ratio τ c =3. cloudy 6μm τ c =. clear 5 4μm The high sensitivity of the intercept to cloud drop size can be used to understand cloud growth/evaporation processes in the transition zone 4
Application to aerosol effect on cloud morphology LWP (g/m2 on log-scale) clean polluted distance (m) distance (m) Koren et al. (GRL, 29): The polluted field dries out more rapidly with distance from cloud, and in the cleaner cloud field there are more numerous and larger weak-cloud elements. Jiang et al. (JGR, 29) these responses are a result of stronger evaporation at cloud edges in the case of polluted clouds. 5
(Potential) observational evidence: 2NFOV Case Case 2 Case 3 2NFOV FOV =.2 wavelengths: 673 nm (RED) 87 nm (NIR) 6
Summary - The slope and intercept of the spectral-invariant relationship is mostly sensitive to cloud properties and NOT sensitive to surface type and aerosol properties - At visible wavelengths, both the slope and intercept primarily depend on cloud optical depth; at waterabsorbing wavelengths, the intercept depends on cloud absorption properties - These results suggest a new cloud retrieval method for the transition zone even with insufficient knowledge about spectral surface albedo and aerosol properties 7
Cloud optical depth 4 3 2 (a) 3D effect Slope function.8.6.4 (a) SBDART MC,sun in the east MC,sun in the west B Ratios to clear -.5.5.5 2 2.5 3 Distance (km) 8 (b) SBDART 7 (.22,.8) MC,sun in the east 6 (.7,.85) 5 4 (.42,.62) 3 MC,sun in the west Slope function.2.8.6.4 Cloud optical depth SBDART MC,sun in the east MC,sun in the west B5. 2 2 4 6 8 2 Cloudy to clear ratio.2 Cloud optical depth. 8
Discussion There is a wavelength-independent function a(time) or a(space) a(t) that characterizes the transition zone (TZ) between cloudy and clear areas. The TZ spectrum is fully determined by this function and zenith radiances spectra of clear and cloudy sky areas. t Model simulations support this statement. However, we do not have yet a clear theoretical understanding of the observed (and simulated) phenomenon. High temporal resolution meas. in the TZ can be well approx. by the lower temporal resolution plus a linear interpolation. Missing (or saturated) spectral data in the TZ can be well retrieved using meas. of the whole spectra of clear and cloudy sky areas. 9
Discussion 2 (applications) If spectral-invariance of the TZ is established, function a(t) can be studied using only two wavelengths, e.g. high temporal resolution 2NFOV (with Red and Near IR channels). The TZ between ice clouds and cloud-free area is longer and smoother than between water and cloud-free area. Koren et al. (29) and Jiang et al. (29) found much sharper cloud edges in polluted environments compared to their cleaner counterparts. Thus a(t) can serve as a characteristic of pollution in a field with small Cu clouds. 2
Discussion 3 (remote sensing) -The slope and intercept of the spectral-invariant relationship is mostly sensitive to cloud properties and not sensitive to surface type and aerosol properties -At visible wavelengths, both the slope and intercept primarily depend on cloud optical depth; at water-absorbing wavelengths, the intercept depends on cloud absorption properties - These results suggest a new cloud retrieval method for the transition zone even with insufficient knowledge about spectral surface albedo and aerosol properties 2
Possible application to cloud thermodynamic phase: 2NFOV water ice TSI images every 3 sec Ice vs water clouds.8 4, 6 km, ice 5, km, water 2NFOV function a.6.4 Function a(dist) FOV =.2 wavelengths: 673 nm (RED) 87 nm (NIR).2 4 8 2 6 distance (m) 22
Spectral-invariance y(λ) = ax(λ) + b 23