Abstract. Cloud liquid water path (LWP) is one of the target
atmospheric parameters retrieved remotely from ground-based and space-borne
platforms using different observation methods and processing algorithms.
Validation of LWP retrievals is a complicated task since a cloud cover is
characterised by strong temporal and spatial variability while remote
sensing methods have different temporal and spatial resolutions. An attempt
has been made to compare and analyse the collocated LWP data delivered by
two satellite instruments SEVIRI and AVHRR together with the data derived
from microwave observations by the ground-based radiometer RPG-HATPRO.
The geographical region of interest is the vicinity of St. Petersburg,
Russia, where the RPG-HATPRO radiometer is operating. The study is
focused on two problems. The first one is the so-called scale difference
problem, which originates from dissimilar spatial resolutions of
measurements. The second problem refers to the land–sea LWP gradient. The
radiometric site is located 2.5 km from the coastline where the effects of
the LWP gradient are pronounced. A good agreement of data obtained at the
microwave radiometer location by all three instruments (HATPRO, SEVIRI, and
AVHRR) during warm and cold seasons is demonstrated (the largest correlation
coefficient 0.93 was detected for HATPRO and AVHRR datasets). The analysis
showed no bias of the SEVIRI results with respect to HATPRO data and a large
positive bias (0.013–0.017 kg m−2) of the AVHRR results for both
warm and cold seasons. The analysis of LWP maps plotted on the basis of the
SEVIRI and AVHRR measurements over land and water surfaces in the vicinity
of St. Petersburg revealed the unexpectedly high LWP values delivered by
AVHRR during the cold season over the Neva River bay and over the Saimaa Lake
and the abnormal land–sea LWP gradient in these areas. For the detailed
evaluation of atmospheric state and ice cover in the considered geographical
regions during the periods of ground-based and satellite measurements,
reanalysis data were used. It is shown that the most probable reason for the
observed artefacts in the AVHRR measurements over water and ice surfaces is the
coarse resolution of the land–sea and snow–ice masks used by the AVHRR
retrieval algorithm. The influence of a cloud field inhomogeneity on the
agreement between the satellite and the ground-based data is studied. For
this purpose, the simple estimate of the LWP temporal variability is used as
a measure of the spatial inhomogeneity. It has been demonstrated that both
instruments are equally sensitive to the inhomogeneity of a cloud field
despite the fact that they have different spatial resolutions.