![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Development of fog detection algorithm using Himawari-8/AHI data at daytime |
VerfasserIn |
Ji-Hye Han, So-Hyeong Kim, Myoung-Seok Suh |
Konferenz |
EGU General Assembly 2017
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 19 (2017) |
Datensatznummer |
250142377
|
Publikation (Nr.) |
EGU/EGU2017-5992.pdf |
|
|
|
Zusammenfassung |
Fog is defined that small cloud water drops or ice particles float in the air and visibility is less
than 1 km. In general, fog affects ecological system, radiation budget and human activities
such as airplane, ship, and car. In this study, we developed a fog detection algorithm (FDA)
consisted of four threshold tests of optical and textual properties of fog using satellite and
ground observation data at daytime. For the detection of fog, we used satellite data
(Himawari-8/AHI data) and other ancillary data such as air temperature from NWP data (over
land), SST from OSTIA (over sea). And for validation, ground observed visibility data from
KMA. The optical and textual properties of fog are normalized albedo (NAlb) and
normalized local standard deviation (NLSD), respectively. In addition, differences between
air temperature (SST) and fog top temperature (FTa(S)) are applied to discriminate
the fog from low clouds. And post-processing is performed to detect the fog edge
based on spatial continuity of fog. Threshold values for each test are determined by
optimization processes based on the ROC analysis for the selected fog cases. Fog detection
is performed according to solar zenith angle (SZA) because of the difference of
available satellite data. In this study, we defined daytime when SZA is less than 85˚ .
Result of FDA is presented by probability (0 ∼ 100 %) of fog through the weighted
sum of each test result. The validation results with ground observed visibility data
showed that POD and FAR are 0.63 ∼ 0.89 and 0.29 ∼ 0.46 according to the fog
intensity and type, respectively. In general, the detection skills are better in the cases
of intense and without high clouds than localized and weak fog. We are plan to
transfer this algorithm to the National Meteorological Satellite Center of KMA for the
operational detection of fog using GK-2A/AMI data which will be launched in 2018. |
|
|
|
|
|