Towards improved ceilometer-based observations of sky cover

With the shift to automated meteorological observations, ceilometers have been used to determine sky cover amount from the surface. This areal measure is obtained from a point observation by measuring the fraction of time that skies directly overhead are cloudy and assuming it represents the fraction of coverage. Operationally, the hourly sky cover observation is obtained by averaging the previous 30 minutes of ceilometer observations and double-weighting the most recent 10 minutes in order to account for rapidly changing conditions.

A decade-long analysis containing more than 29,000 daytime cloud observations was undertaken using data from the Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP) climate research facility. Observations from a ceilometer were compared to a collocated Total Sky Imager (TSI), a visible wavelength camera that obtains sky cover through digital image analysis. Observations were sorted into the standard bins of clear, few, scattered, broken, and overcast based on the percentage of cover.

Overall, only 58% of the observations were placed into the same category by both instruments. The ceilometer recorded close to 70% more clear sky observations than the TSI does, likely due to limited ability of the ceilometer to detect high and thin clouds. Digital image processing techniques and statistical analysis will be used on cases with large discrepancies to determine the cloud types and meteorological conditions that are prone to causing observation errors, and a new model for calculating sky cover based on a regression of ceilometer and surface meteorology observations will be presented.