Toward a standard procedure for validation of satellite-derived cloud liquid water path: a study with SEVIRI data

Differences between satellite-derived and ground-based values of cloud liquid water path ([LWP.sub.sat] and [LWP.sub.gr], respectively) in validation studies are partly associated with the validation itself, in particular with scale differences and parallax. This paper aims at establishing standards for validation procedures to minimize these contributions to the differences. To investigate this, LWP values were collected as computed from ground-based microwave radiometer (MWR) summer measurements made at two Cloudnet sites and from the spaceborne Spinning Enhanced Visible and Infrared Imager (SEVIRI) instrument. The large number of all-sky sample pairs (~2500 after selection) formed an essential condition for the present study. The best validation method was determined by optimum statistical agreement between [LWP.sub.sat] and [LWP.sub.gr]. The method consists of (i) computation of [LWP.sub.sat] by averaging LWP over the pixels surrounding the ground station by means of a Gaussian weight function with a length scale defining the validation area, (ii) computation of [LWP.sub.gr] by averaging the MWR measurements with a Gaussian weight function, by using a time scale that is considerably longer than the time in which the clouds move across the validation area (by a factor of 3-15), and (iii) correcting for parallax. The authors argue that the best length scale for averaging the satellite data is equal to the image resolution. The improvement resulting from the parallax correction was significant at the 99.5% level, but its effect was not significant for a subset of the data for relatively homogeneous cloud fields. Also, there was no significant improvement when, instead of taking a constant, the time scale for averaging the ground data was adjusted to the instantaneous wind field.