Remotesensing Measurements and Limitations Visible and Infrared Imagery

Imagers are the "workhorse" sensors for spaceborne cloud and aerosol retrievals, as they provide high global coverage of observations at high spatial resolution. The two different kinds of algorithms are based on (a) reflection of visible light (daytime only) and (b) emission of thermal infrared radiation from cloud top (day and night). Operational retrievals are based on lookup tables, created from radiative transfer models for a given geometry and a set of parameters such as optical thickness, effective radius, cloud-top pressure, and thermodynamic phase. The measured reflected (emitted) radiance is compared to lookup table values, and the corresponding cloud parameters are assigned based upon the modeled values, which approximate most closely the measurements. Emission-based retrievals rely on the temperature contrast between the surface and the (colder) clouds. Therefore, they are capable of detecting high-altitude thin (even subvisible) cirrus for which reflectance-based retrievals fail. Systematic errors in these retrievals arise from vertical and horizontal spatial cloud variability. One example is the albedo bias: because of the nonlinear relationship between optical thickness and cloud albedo, a heterogeneous cloud with the same mean optical thickness as a homogeneous cloud has always a lower albedo than its homogeneous counterpart, which leads to an underestimation of its optical thickness. The effective radius retrievals also become unreliable, especially for broken clouds. The cloud cover derived from imagery depends largely on the instrument spatial resolution (ranging from a few meters to 20 km) and sensitivity of the retrieval to various types of clouds, as well as on cloud optical thickness. Cloud vertical structure is not resolved by nadir-viewing imagers; multi-layer clouds represent a problem especially if the uppermost layer is partly transparent. Ice clouds are particularly difficult for imagery, not only because they can be extremely thin and inhomogeneous but also because of the ice crystal shape; currently, techniques for utilizing multiangle (McFarlane et al. 2005) or polarized (Sun et al. 2006) observations are being explored.

The "classical" AVHRR was used for cloud and aerosol retrievals on a variety of platforms and is still flown on geostationary platforms. Since its first deployment in 1978, both the spatial and spectral resolution of imagers increased: MODIS, flown on NASA's polar orbiting sun-synchronous satellites Terra and Aqua, provides cloud retrievals at 1 km spatial resolution. Aqua is part of the so-called "A-Train," a satellite constellation of various platforms in short sequence with a daily afternoon overpass. MISR (Kahn et al. 2005) was specifically designed to improve aerosol and cloud retrievals by a combination of cameras with different viewing angles onboard Terra. POLDER (Riedi et al. 2001) utilizes the information from polarized refl ectance for cloud phase discrimination, crystal shape detection, and aerosol retrievals. Over the last decade, significant progress has been made in the aerosol retrieval from these three platforms, especially over bright land surfaces. However, they are currently only possible under completely cloud-free conditions.

Whereas standard imagers rely on information from a limited number of wavelength bands, spectral imagers allow for novel retrieval techniques. In the thermal window, AIRS provides vertical profi les of water vapor, carbon dioxide, and temperature. In the solar spectral range, the capabilities of spectral imaging for combined aerosol-cloud retrievals or the attribution of climate change to various forcing agents are being explored.

Radar

Radar (radio detection and ranging) relies on scattering of microwave radiation by cloud drops and precipitation. The intensity of the back-scattered signal depends on the distance between the radar and the drops, the drop size distribution, and the wavelength of the emitted pulses. In the Rayleigh limit of Mie scattering theory, it is proportional to 1-4 and (D6). Therefore, sensitivity to small cloud drops can only be achieved by using shorter wavelengths. However, at short wavelengths, the signal becomes quickly attenuated as a result of increased scattering and absorption near the edge of the radio window (1 = 1 cm). Hence, the choice of the wavelength depends on the targets: weather radars (5-10 cm wavelength) penetrate non-precipitating clouds whereas cloud radars (1 cm) are sensitive to the smaller-size cloud drops but have shorter range. Most weather radars are scanning and polarized systems, operating in pulse mode. The distance to the cloud is determined from the elapsed time for the pulse's round trip, where the maximum range is determined by the pulse separation. In addition, Doppler radars determine the speed and direction of hydrometeors relative to the radar system. The range in detectable speeds is inversely proportional to the pulse separation time. Signal polarization is used to detect the vertical and horizontal dimension and thus the shape of the scattering object. Radars are used for water as well as for ice clouds. They excel in detecting cloud structure while retrievals of ice and liquid water content profiles are better constrained when combined with data from other instruments.

Radar systems are most commonly deployed at the ground. The first spaceborne radar cloud system was the precipitation radar (f = 13.8 GHz, 1 ~ 2 cm) onboard the TRMM satellite. More recently (2006), the cloud profiling radar (CPR; f = 94 GHz, 1 ~ 3 mm) was launched on CloudSat (Stephens et al. 2002). The frequency, peak power, and dynamic range were chosen to reconcile sufficient sensitivity to small cloud drops and the ability to profile moderately thick clouds. Figure 4.8 shows a CloudSat image from November 9, 2006, as the Canadian Convair 580 was flying underneath during the Canadian CloudSat/CALIPSO Validation Project. A comparison between the satellite radar (W-band) and the aircraft radar (Ka-band) shows some significant differences attributable to wavelength and resolution. However, overall, the CloudSat radar did an excellent job of characterizing the cloud and detecting light precipitation, and it provides coverage over much larger areas than other methods. Figure 4.9 shows the PMS 2-DC ice crystal images from when the aircraft and satellite were observing the same cloud. These diagrams illustrate the general point that in-situ aircraft with remote and in-situ sensors can be very useful for both testing and validating satellite instrumentation.

Lidar and Ceilometer

Lidar (light detection and ranging) is similar to radar but works with pulses of visible and near-infrared light. Since the wavelength is much shorter than microwave radiation, the back-scattered signal is sensitive to aerosol particles (typical sizes on the order of 0.1-1 pm, compared to precipitation size particles on the order of 0.1-1 mm), but the optical penetration depth is comparatively low (r ~ 3...4). The lidar return is determined by the aerosol and molecular extinction profi le in a nontrivial manner. The high spectral

CloudSat W-band radar

CloudSat W-band radar

18:04:25Z 18:04:25Z 18:04:25Z 18:04:25Z 18:04:25Z 18:04:25Z

Time (UTC)

18:04:25Z 18:04:25Z 18:04:25Z 18:04:25Z 18:04:25Z 18:04:25Z

Time (UTC)

Canadian Convair Ka-band radar

Canadian Convair Ka-band radar

Latidude (deg)

17:58:39Z 18:01:19Z 18:03:59Z 18:06:39Z 18:09:19Z 18:11:58Z

Time (UTC)

Latidude (deg)

17:58:39Z 18:01:19Z 18:03:59Z 18:06:39Z 18:09:19Z 18:11:58Z

Time (UTC)

Figure 4.8 Top: CloudSat radar image (W-Band) from November 9, 2006, while passing over southern Ontario at the latitudes and time indicated. The horizontal line across the image shows the level the instrumented Canadian Convair 580 was flying at the time; the vertical line shows the location where it is estimated the two platforms coincide. The bright pixel near the intersection is likely a reflection from the aircraft itself. Bottom: Ka-band radar data from the aircraft with the images matched to correspond with each other. The Convair was flying along the same path as the satellite, but obviously at a slower speed. The Convair radar data is of higher resolution but it matches the properties of the CloudSat image in the main features. (Image selected and produced by Dave Hudak and Peter Rodriguez of Environment Canada.)

resolution lidar (HSRL) separates the molecular and aerosol contribution by detecting the Doppler broadening of the return signal caused by thermal motion of molecules (Shipley et al. 1983). Raman lidars (Ansmann et al. 1990) utilize inelastic molecular scattering of light for determining the concentration of atmospheric gases and aerosol extinction profiles independently. The

Cloud Properties from In-situ and Remote-sensing Measurements 95 OAP-2DC 18:04:19.20

^i'lh i utikllrrr l, r i . <**» i I**, i * tit i. i i u. «

i

.ri rl as l* ; im

l u (- l w l r l r i-1 r li iW i fc 5 i ■ i w i i i i ■

OAP-2DC 18:04:28.43

1 U 1

jum i*r

ni'LLfi'i.Lr^rf

wr~

' l rr Pii l

OAP-2DC 18:04:37.20

ULPrir

MfcTiri»LLrr>r r

OAP-2DC 18:04:41.61

ki

rrk*MTtu

r

ri*

ui *uru.

Figure 4.9 PMS 2-DC imagery taken at the time the CloudSat radar was imaging the Convair 580. Vertical bars represent 800 ^m. The ice crystals observed were somewhat irregular in shape. (Image produced by Alexei Korolev of Environment Canada.)

Figure 4.9 PMS 2-DC imagery taken at the time the CloudSat radar was imaging the Convair 580. Vertical bars represent 800 ^m. The ice crystals observed were somewhat irregular in shape. (Image produced by Alexei Korolev of Environment Canada.)

depolarization of the lidar return gives information about the shape of the scat-terers. A laser ceilometer is a simple type of a lidar where the cloud base is defined at the altitude where the back-scattered signal increases beyond a certain threshold. This threshold may give erroneous cloud base readings for high aerosol optical thickness below the cloud. Optical ceilometers determine cloud base geometrically by measuring the angle of reflectance with a scanning light source and a horizontally separated photocell.

In 2006, CALIOP (Winker et al. 2007) was added to NASA's A-Train onboard CALIPSO, in close coordination with CloudSat and Aqua. It provides vertical profiles of aerosols and thin clouds.

Microwave Radiometers

Passive microwave radiometers detect naturally emitted radiation from clouds and precipitation as well as from the surface, which can be related to LWP and precipitation rate. LWC profiles can be derived in conjunction with radars. The humidity and temperature profile of the atmosphere as well as surface properties must be taken into account for the retrievals. The errors associated with LWP can be as high as 25 g m-2 (typical values for thin water clouds range from 10-100 g m-2). The uncertainty attributable to the atmospheric profile can be substantially decreased by the use of dual-frequency radiometers (e.g., Liljegren et al. 2001). These operate at frequencies from 10-30 GHz where the signal is largely determined by precipitation and atmosphere and clouds are virtually transparent. At higher frequencies, the signal is increasingly attenuated, mainly by clouds and water vapor. The addition of high-frequency channels enables the retrieval of the cloud LWP in presence of precipitation, increases the sensitivity to low LWP clouds, and improves the retrievals with respect to the atmospheric profile (Di Michele and Bauer 2006). In particular, the water vapor and temperature profile within clouds can be retrieved.

0 0

Post a comment