The hyperspectral and polarization resolving imager specMACS during EUREC4A

The hyperspectral and polarization resolving imager specMACS has been operated during the EUREC4A field campaign on board the HALO research aircraft in January and February 2020 in the trade wind region east of Barbados. The field campaign aims to Elucidate the Couplings Between Clouds, Convection and Circulation (Bony et al. 2017).

Instrument Setup

The specMACS instrument was set up for EUREC4A as a combination of two hyperspectral line imagers and two polarization resolving 2D cameras. The assembled system is shown in Figure 1. The instrument was mounted on board the HALO research aircraft in downward looking perspective for cloud remote sensing applications.

Hyperspectral Sensors

The hyperspectral line imagers of the specMACS system (Figure 2) are line imagers and operate by projecting a slit onto a diffraction grating G which provides spectral separation of the incoming light to the image sensors (Figure 3). The sensors are sensitive in the VNIR (400nm - 1000nm) and the SWIR (1000nm - 2500nm) wavelength range respectively. The sensors have been operated according to the push broom principle with 30 Hz sampling rate or about 7 meter traveling distance of the aircraft. See Ewald et al. 2016 for further details on the specMACS hyperspectral sensors.

specMACS assembled
Figure 1: Assembled specMACS system.
specMACS VNIR and SWIR cameras
Figure 2: VNIR and SWIR sensors removed.
specMACS ray path
Figure 3: Optical path of the specMACS hyperspectral sensors (from Ewald et al. 2016).

Polarization Resolving Sensors

The new addition for the EUREC4A field campaign are two polarization resolving 2D cameras. The two Phoenix 5.0 MP Polarized cameras include four directional pixelated wire grid polarizers in front of the image sensors. Using measurements from these sensors, one can reconstruct the first three components (I, Q, U) of the incoming Stokes vectors across the captured image. Figure 4 shows the intensity I on the left and the degree of linear polarization, derived from the Stokes vector on the right.

The cameras are mounted with partially overlapping field of view, pointing at approximately 20° off-nadir to both sides in the across track direction. This way, the field of view observed by the instrument is relatively large and covers 82° along track and 110° across track. During the field campaign, the polarization resolving imagers have been operated at 8 Hz sampling rate or about 25 meter traveling distance of the aircraft.

polarization resolving image
Figure 4: Example image of a polarization resolving camera. The left panel shows an ordinary RGB image, while the right image shows the uncorrected RGB degree of linear polarization of the same scene. The difference between these two images shows that the polarization resolving channels enhance the signal from the cloudbow strongly.

Cloud Surface Geometry

Using the RGB images from the polarization resolving imagers, it is possible to retrieve 3D cloud geometry information. As indicated in Figure 5, two images are taken in consecutive time steps t1 and t2. If the same part of the cloud surface is identified in the two images, the location on the cloud surface in 3D space can be obtained from the triangle given by the travelled distance d and the two viewing directions v1 and v2. Furthermore, a small possible misspointing m must be taken into account. Details on the method can be found in Kölling et al. 2019.

stereo geometry
Figure 5: Stereo geometry (from Kölling et al. 2019)

Figure 6 shows an example of a towering cumulus cloud which developed on 09 Feb 2020 at 10:06 UTC during the EUREC4A field campaign. Figure 7 shows the point cloud derived from this cloud, with color indicating height above WGS84 in meters, indicating that the tower rose more than 4 km above the shallow cumulus layer at 2 km. While robust results can be obtained from the lower layers, the side of the tower facing the camera shows only little contrast and finding image correspondences becomes challenging, leading to less cloud location points.

specMACS RGB image from polarized camera
Figure 6: RGB image
specMACS 3D reconstruction
Figure 7: Cloud height, camera perspective

When changing the view to a perspective from higher above (Figure 8), the 3D cloud structure becomes more apparent. In this case, the tops of the tower can not be retrieved as the tower grew out of the cameras field of view. It can also be seen that some points from behind the tower can be found as the reconstruction method also uses images from before and after the image shown in Figure 6.

specMACS 3D reconstruction
Figure 8: Cloud height, view from above

Currently the data from the EUREC4A field campaign is being processed, cloud surface location data will be made available for all observed clouds during the campaign.

Note that a prior version of this method, based on a different camera model and data from the NARVAL-II field campaign, has been presented at the EGU GA 2017. While the previous setup provided valuable input to the retrieval development, the current setup used for EUREC4A provides an increased field of view and due to the improved camera control system a greater level of detail.

Campaign Quicklooks and other Products

Most of the data acquired during the EUREC4A field campaign has already been processed into quicklooks, which can be found on the macsServer. As processing is not yet finished, further products will be added in due course.

Further examples and products are presented by Pörtge et al. during this years EGU GA.

References