A Simulation Study Exploring the Effects of Sensor Spatial Resolution on Estimates of Cloud Cover from Satellites

Abstract

The effect of sensor spatial resolution on estimating the amount of clouds covering the earth were investigated by simulating various cloud distributions and sizes, and measuring the known cloud amount with resolution of different sizes cloud-no cloud threshold technique often applied in automatic data processing. Cloud amount statistics have been tabulated for a three-orders-of-magnitude range in the ratio (R) of areal cloud size to areal resolution size for seven cloud amounts between 6 and 90%. Three different cloud patterns were used. These were 1) a regularly spaced pattern of homogeneous dots arranged in rows and columns (to simulate cloud streets), 2) a randomly spaced pattern of the same dots (to simulate randomly oriented cumulus clouds), and 3) a heterogeneous cloud size distribution irregularly spaced (to simulate a view of different cloud types and sizes). Two cloud amount estimation techniques were tested. Cloud amounts of 100% (method 1) and 50% (method 2) were assigned to partially filled resolution elements. Using criteria applicable to some studies carried out in the past, it is shown that cloud amount estimations can be in error by as much as 86% and 38%, respectively, for the two methods. Nomograms have been developed which substantially improve the estimate of the true cloud cover for R < 100 provided that R can be determined. Good agreement was found when a check was performed on whether or not the simulated cloud patterns were representative of real cloud patterns. The check was made by testing the cloud cover estimate nomograms constructed with the simulated data against similar curves prepared from cloud fields extracted from high-resolution photographs obtained from an Apollo flight.