This dataset contains a collocation of three different A-Train cloud phase products on a 2x30 grid (latxlon): The GCM-Oriented Cloud Calipso Product (GOCCP v3.0), the liDAR-raDAR (DARDAR-MASK v2.0) product, and the Polarization & Anisotropy of Reflectances for Atmospheric Sciences coupled with Observations from a Lidar (PARASOL) + Aqua/MODIS combination (PM-L2) product.

To better understand the differences in cloud glaciation around the globe, liquid and ice cloud tops around the globe were counted, as observed from four different space-borne instruments (lidar, radar, radiometer and polarimeter) during four years. The top pixels of each product were selected and binned into 3 K temperature intervals. With this data, an algorithm was applied to determine the most probable thermodynamic phase at cloud top based on the different cloud-phase products on a 2x30 grid (latxlon). The "GDP2" ensemble was produced using the mode between the three different binary classifications (ice/liquid), provided that all three products are available at each timestep. The "GDP3" ensemble is a subselection of the GDP2 ensemble, containing only timesteps where all three products agree. The "GP"(GOCCP and PM-L2), "DP"(DARDAR and PM-L2) and "GD"(GOCCP and DARDAR) ensembles contain only timesteps where two products agree.