New measurements

CPR: EarthCARE's Doppler cloud radar

In addition to measuring radar reflectivity factor, EarthCARE's 94 GHz Cloud Profilling Radar (CPR) also infers the vertical velocity of particles in the atmosphere from the Doppler shift in the frequency of the returned signal. This velocity consists of the sum of the terminal fall speed of the particles and the vertical motion of the air. Most of the time the vertical air motion averaged over the 750-m radar footprint is small so the measurement can be treated as the mean terminal fall speed of the particles but weighted by their backscattering cross-section at 94 GHz. Since radar scattering is dominated by the larger particles, the radar tends to see the larger precipitation particles even when cloud is present in the same volume.

In the examples below there is a sharp increase in fall speed at the melting layer where snow melts to rain. In rain, EarthCARE's radar-only C-CLD and synergistic ACM-CAP  retrieval algorithms essentially use the Doppler velocity to infer the mean raindrop size (since larger drops fall faster) from which they can make a more accurate estimate of rain rate than possible from radar reflectivity alone. In snow, the fallspeed is indicative of the density of the particles, which can in principle be used to improve snow rate retrievals as well as estimating how much riming has occurred. Comparison of Doppler velocity with simulated values by numerical models provides a valuable constraint on the representation of precipitation processes in models. 

The cases above also show regions of convection where the air motion exceeds the terminal fall speed of the particles. There is great interest in using measurements of the strength and width of updrafts to evauate simulations by cloud resolving models and improve our understanding and forecasting of these extreme weather events. The CPR employs a variable pulse repetition frequency through its orbit, leading to a "folding velocity" that varies between ±5 and ±6 m/s. In the weak updraft in the Mediterranean convective cloud at a latitude of 43ºN, the Doppler velocity appears to fold at most once, implying an updraft velocity of no more than 10 m/s. In the much more intense Benin storm at 10.75ºN, the Doppler velocity folds many times. The EarthCARE team is working an unfolding algorithm but this is a challenging task given the (expected) presence of some random noise in the raw measurements. 

Users should also be aware that 94 GHz radar pulses can not only be strongly attenuated in convective clouds, but also be scattered multiple times before returning to the receiver. This is the case at a latitude of 10.75oN in the Benin storm above: the longer path length of radiation subject to multiple scattering at the top of the cloud leads to signals appearing to originate from lower in the cloud. In such situations, inferring updraft velocities from the measurements will only be possible near cloud top. Nonetheless, there is clearly some tantalizing new information present in these measurements and it is an intriguing research challenge to fully exploit it.

In addition to its Doppler capability, the CPR has a few additional advantages over the CloudSat radar (2006-2023). Its lower orbit and larger antenna makes it more sensitive and its smaller footprint reduces the magnitude of multiple scattering. Its oversampling of the signal at a resolution of 100 m rather than 250 nm improves the profiling capability even though the pulse width is the same at 500 m. Finally, the shape of the pulse drops off very sharply reducing the intensity of the blind zone near the surface, meaning that the radar is already fully sensitive 500 m above the surface.

ATLID: EarthCARE's high spectral resolution lidar

The most important property of a cloud or aerosol layer that determines how much it interacts with sunlight is its extinction coefficient. Estimating the profile of extinction coefficient from a standard backscatter lidar, such as CALIPSO, is ambiguous because we don't know exactly the ratio between extinction and backscatter, and we must account for attenuation of the lidar signal as it penetrates through the cloud or aerosol layer. A further ambiguity arises because the measured lidar backscatter is the sum of the contribution from air molecules (the Rayleigh channel) and particles (commonly referred to as the Mie channel, although Mie scattering is strictly limited to scattering by spheres).

A high spectral resolution lidar (HSRL) can infer the particle extinction coefficient unambiguously by separating out the particle and air contributions to the backscatter. It does this by exploiting the fact that cloud and aerosol particles move at only a few m/s while air molecules move at hundreds of m/s. This means that air molecules change the frequency of the reflected laser light much more than particles, enabling the molecular signal to be isolated via an appropriate spectral filter.

The first HSRL in space was the wind lidar Aeolus (2018-2023), although with its large range gates, long averaging time and slanted path it was not ideally suited to observing clouds and aerosols. The second HSRL space is ACDL (launched in 2022), although its data are not freely available. There fore, EarthCARE's Atmospheric Lidar (ATLID) is the first spaceborne HSRL targeting clouds and aerosols to release its data publicly. Each of these lidars operate in the ultraviolet at 355 nm, where Rayleigh scattering is strong.

The figure below illustrates the separate backscatter signals from particles and from air molecules (the "Mie channel" and "Rayleigh channel", respectively) for a cirrus cloud and a layer of smoke over a stratocumulus cloud. The location of these layers is mapped precisely by the particle backscatter, but this is not enough to quantify the extinction coefficient unambiguously. In clear skies (e.g. at 63.5oN) the air backscatter increases towards the surface due to the increase in atmospheric density, but when the pulse passes through a cloud or aerosol layer we observe partial or total attenuation of the signal. Since we know the density of the atmosphere very well, we know what the air backscatter would have been without the presence of clouds or aerosols. The difference between that and the measured value enables the profile of extinction coefficient to be retrieved unambiguously, as explained in the description of EarthCARE's A-EBD algorithm.

The ability of ATLID to measure global profiles of aerosol extinction coefficient is a game changer for evaluating air quality forecasts. Indeed, current operational methods for evaluating such aerosol forecasts with MODIS or ground-based AERONET sites work only in clear skies and only provide the column optical depth. The smoke in the figure above is would be impossible to measure with MODIS or AERONET because it overlies an optically thick liquid cloud, but this poses no problems for ATLID. ATLID can also infer the type of aerosols particle using the combination of the backscatter-to-extinction ratio derived from the HSRL and the depolarization, as shown by Figure 8 of Illingworth et al. (2015).

For clouds of optical depth greater than around 3, the air backscatter is totally extinguished, in which case the HSRL is best used in combination with the radar by a synergistic retrieval algorithm such as ACM-CAP, which makes use of the HSRL capability of the lidar.