Aerosol Optical Depth (AOD) Retrieval from Noaa AVHRR in an Alpine Environment: Validation Using AERONET Data

• Main Authors: A. Hauser, D. Oesch, S. Wunderle, Hallerstrasse Bern Switzerland
• Other Authors: The Pennsylvania State University CiteSeerX Archives
• Format: Text
• Language: English
• Subjects: Aerosol Robotic Network
• Online Access: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.2.5833; http://www.eumetsat.de/en/area2/proceedings/eump39/docs/5_12_hauser.pdf

The aim of this study is the retrieval of aerosol optical depth from the National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) sensor over land. The region of interest covers central Europe ranging from 50N to 40.5N and from 0E to 17E including the European Alps. On the temporal scale, we limit the data set to afternoon NOAA-16 passes of the entire year 2002. In this region, there are sixteen stations from the Aerosol Robotic Network (AERONET) at which we can compare the ground-based versus the space borne measurements. The most crucial parameter in the retrieval procedure is the estimate of a correct surface reflectance since inaccuracies of 0.01 can result in AOD variations of 0.1. Surface reflectance has been estimated by extracting the minimum reflectance within 10 intervals of the satellite zenith angle within two-month intervals. This method eliminates the varying reflectance with varying satellite zenith angle but the extracted surface reflectance still contains an aerosol signal. The aerosol optical depth retrievals from NOAA AVHRR are analysed in detail for the AERONET site Ispra (Italy). The temporal characteristics of the aerosol optical depth are well preserved, independent of the viewing geometry. However, there is an offset between the ground-based measurements and the satellite observations. The reason for this offset has to be further investigated. Improvements of the cloud masking are expected to decrease the amount of scatter. We can conclude that aerosol retrieval over land using AVHRR is a challenging task but it is possible to extract some valuable results.