3D cloud geometry

Dataset description

The cloud reconstruction is performed by using the images of the two polarization resolving cameras of the specMACS instrument. The algorithm is based on identifying points on the clouds surface and to re-find them in following images. As a first result, point clouds are obtained. Furthermore, a Poisson surface reconstruction is used to derive the cloud surfaces.

For each flight segment three NETCDF4 files are provided. Their ideas are briefly described in the following:

• "points_segement-id.nc": This file contains all the points that are located by the algorithm. The most important variables are:
  • "loc": location of the sample points in a three dimensional cartesian coordinate system with its origin in the center of the circles flown by HALO
  • "height": height of the sample points above WGS84 in meters
  • "lat": latitude of the sample points in degree north (WGS84)
  • "lon": longitude of the sample points in degree east (WGS84)
  • "viewpoint": average position of the cameras from where a located point was observed (reference point)

• "centroids_segment-id.nc": This is a reduced data file from the file above. It contains so-called centroids. The sample points from the points.nc file were therefore sorted in 100x100x100m^3 cubes and their locations averaged. Next to the above mentioned variables in the points.nc file contains a "local_sp" variable which gives the location of the centroid in a so-called local reference frame. The origin of this local reference frame is centered in the middle of the considered flight segment. Furthermore, the file contains a variable "count_boxes" which gives the number of original points per box and "count_2d" which is the the sum over all counts in a vertical column.

• "centroid_mesh_segment-id.nc": This file provides the data of the cloud surfaces obtained from the centroid-file. The Poisson surface reconstruction gives a mesh of triangles which give the actual cloud surface. The file contains the following variables:
  • "vertices": the location of the vertices of the triangles constructed from the Poisson surface reconstruction
  • "triangles": this variable contains the numbers of the three vertices of the considered triangle
  • "point_distance": minimal distance of a triangle to a point of the point clouds

For some flight segments, an extra "mesh_flight-id.nc" file is added. This provides the results of the Poisson surface reconstruction with the original points as an input. If there is no such file, the data had to be reduced for the actual surface reconstruction. In general, the obtained surfaces should be very similar and should not lead to large differences in further computations.

Important notes and known problems:

• The data provided here are only a few examples of the proceeded data from the EUREC4A field campaign situated in Barbados in early 2020. They are chosen with regard to the considered examples during the EUREC4A Hackathon from July 28 to July 30.
• As the aircraft was flying above the clouds, the provided cloud heights are in general the cloud top heights. Nevertheless, due to three dimensional effects some lower points on the sides of the clouds might be tracked, too.
• The estimations of the height might be influenced by the movement of the clouds, leading to an overestimation, when the aircraft flew in the opposite direction as the clouds and an underestimation, when the directions are parallel. For more information and possible correction of this error do not hesitate to contact us by for example writing an email to Lea Volkmer

Changes

The algorithm developed to reconstruct the clouds in three dimensional space from the specMACS data is described in the following paper: Kölling, T., Zinner, T., and Mayer, B.: Aircraft-based stereographic reconstruction of 3-D cloud geometry, Atmos. Meas. Tech., 12, 1155–1166. It can be found under this DOI. Next to the actual point clouds, the centroids were added to reduce the data to work with.

Download

HALO-0205

• points_HALO-0205_sl1 direct file download: points_HALO-0205_sl1.nc (file size ca. 800 MB) or via OPeNDAP
• centroids_HALO-0205_sl1 direct file download: centroids_HALO-0205_sl1.nc (file size ca. 800 MB) or via OPeNDAP
  
• centroid_mesh_HALO-0205_sl1 direct file download: centroid_mesh_HALO-0205_sl1.nc (file size ca. 200 MB) or via OPeNDAP
  
• points_HALO-0205_sl2 direct file download: points_HALO-0205_sl2.nc (file size ca. 800 MB) or via OPeNDAP
• centroids_HALO-0205_sl2 direct file download: centroids_HALO-0205_sl2.nc (file size ca. 800 MB) or via OPeNDAP
  
• centroid_mesh_HALO-0205_sl2 direct file download: centroid_mesh_HALO-0205_sl2.nc (file size ca. 200 MB) or via OPeNDAP
  
• points_HALO-0205_sl3 direct file download: points_HALO-0205_sl3.nc (file size ca. 800 MB) or via OPeNDAP
• centroids_HALO-0205_sl3 direct file download: centroids_HALO-0205_sl3.nc (file size ca. 800 MB) or via OPeNDAP
  
• centroid_mesh_HALO-0205_sl3 direct file download: centroid_mesh_HALO-0205_sl3.nc (file size ca. 200 MB) or via OPeNDAP
• points_HALO-0205_sl4 direct file download: points_HALO-0205_sl4.nc (file size ca. 800 MB) or via OPeNDAP
• centroids_HALO-0205_sl4 direct file download: centroids_HALO-0205_sl4.nc (file size ca. 800 MB) or via OPeNDAP
• centroid_mesh_HALO-0205_sl4 direct file download: centroid_mesh_HALO-0205_sl4.nc (file size ca. 200 MB) or via OPeNDAP

HALO-0130

• points_HALO-0130_sl1 direct file download: points_HALO-0130_sl1.nc (file size ca. 800 MB) or via OPeNDAP
• centroids_HALO-0130_sl1 direct file download: centroids_HALO-0130_sl1.nc (file size ca. 800 MB) or via OPeNDAP
  
• centroid_mesh_HALO-0130_sl1 direct file download: centroid_mesh_HALO-0130_sl1.nc (file size ca. 200 MB) or via OPeNDAP
  
• mesh_HALO-0130_sl1 direct file download: mesh_HALO-0130_sl1.nc (file size ca. 200 MB) or via OPeNDAP
  
• points_HALO-0130_sl2 direct file download: points_HALO-0130_sl2.nc (file size ca. 800 MB) or via OPeNDAP
• centroids_HALO-0130_sl2 direct file download: centroids_HALO-0130_sl2.nc (file size ca. 800 MB) or via OPeNDAP
  
• centroid_mesh_HALO-0130_sl2 direct file download: centroid_mesh_HALO-0130_sl2.nc (file size ca. 200 MB) or via OPeNDAP
  
• mesh_HALO-0130_sl2 direct file download: mesh_HALO-0130_sl2.nc (file size ca. 200 MB) or via OPeNDAP
  
• points_HALO-0130_sl3 direct file download: points_HALO-0130_sl3.nc (file size ca. 800 MB) or via OPeNDAP
• centroids_HALO-0130_sl3 direct file download: centroids_HALO-0130_sl3.nc (file size ca. 800 MB) or via OPeNDAP
  
• centroid_mesh_HALO-0130_sl3 direct file download: centroid_mesh_HALO-0130_sl3.nc (file size ca. 200 MB) or via OPeNDAP
• mesh_HALO-0130_sl3 direct file download: mesh_HALO-0130_sl3.nc (file size ca. 200 MB) or via OPeNDAP
  
• points_HALO-0130_sl4 direct file download: points_HALO-0130_sl4.nc (file size ca. 800 MB) or via OPeNDAP
• centroids_HALO-0130_sl4 direct file download: centroids_HALO-0130_sl4.nc (file size ca. 800 MB) or via OPeNDAP
• centroid_mesh_HALO-0130_sl4 direct file download: centroid_mesh_HALO-0130_sl4.nc (file size ca. 200 MB) or via OPeNDAP
• mesh_HALO-0130_sl4 direct file download: mesh_HALO-0130_sl4.nc (file size ca. 200 MB) or via OPeNDAP

Help needed?

If you have any questions concerning the provided data please feel free to contact us! This could be, e.g., by sending an e-mail to Lea Volkmer.

Literature

• The algorithm developed to reconstruct the clouds is described by Kölling, T., Zinner, T., and Mayer, B.: Aircraft-based stereographic reconstruction of 3-D cloud geometry, Atmos. Meas. Tech., 12, 1155–1166, 2019. It can be found under this DOI.
• Further descriptions can be found in the PhD thesis of Tobias Kölling. It is avalable under this link.