The Impact of Time Difference between Satellite Overpass and Ground Observation on Cloud Cover Performance Statistics

Remote Sensing

Jᆝrzej S. Bojanowski¹  Reto Stཬkli²  Anke Tetzlaff²  Heike Kunz²  Alexander Kokhanovsky² 

[1] Federal Office of Meteorology and Climatology MeteoSwiss, Climate Services, Operation Center 1, P.O. Box 257, CH-8058 Zürich-Flughafen, Switzerland;

关键词: validation;    collocation;    SYNOP;    AVHRR;    sampling error;    cloud retrieval;    ESA-Cloud-CCI;    APCADA;    skill score;    Hanssen-Kuiper’s discriminant;   

DOI  :  10.3390/rs61212866

来源: mdpi

【 摘 要 】

Cloud property data sets derived from passive sensors onboard the polar orbiting satellites (such as the NOAA’s Advanced Very High Resolution Radiometer) have global coverage and now span a climatological time period. Synoptic surface observations (SYNOP) are often used to characterize the accuracy of satellite-based cloud cover. Infrequent overpasses of polar orbiting satellites combined with the 3- or 6-h SYNOP frequency lead to collocation time differences of up to 3 h. The associated collocation error degrades the cloud cover performance statistics such as the Hanssen-Kuiper’s discriminant (HK) by up to 45%. Limiting the time difference to 10 min, on the other hand, introduces a sampling error due to a lower number of corresponding satellite and SYNOP observations. This error depends on both the length of the validated time series and the SYNOP frequency. The trade-off between collocation and sampling error call for an optimum collocation time difference. It however depends on cloud cover characteristics and SYNOP frequency, and cannot be generalized. Instead, a method is presented to reconstruct the unbiased (true) HK from HK affected by the collocation differences, which significantly (t-test p < 0.01) improves the validation results.

【 授权许可】

CC BY   
© 2014 by the authors; licensee MDPI, Basel, Switzerland.