| Title | Cloud detection and thin cloud calibration in noaa avhrr images with fuzzy logic |
| |
| Author | Yong, D; Gower, J F R |
| Source | Canadian Journal of Remote Sensing vol. 26, no. 1, 2000 p. 54-63, https://doi.org/10.1080/07038992.2000.10874754 |
| Year | 2000 |
| Alt Series | Natural Resources Canada, Contribution Series 20181759 |
| Publisher | Informa UK Limited |
| Document | serial |
| Lang. | English |
| Media | paper; on-line; digital |
| File format | pdf |
| Subjects | geophysics; remote sensing |
| Program | Canada Centre for Remote Sensing Divsion |
| Released | 2014 07 31 |
| Abstract | A large data set ofsea surface temperatures derived from NOAA - AVHRR satellite images, and measured by surface meteorological buoys are compared for the time period from April to August 1997 in the
B.C. coastal region. The data are screened for thick cloud using thresholds of visible light reflectance and apparent surface temperature derived from bands 1 and 5 of the AVHRR. Agreement between the two sets of water temperature is about 1.25°e
RMS. Improved agreement is achieved when the standard global retrieval formula for satellite SST is modified in accordance with the conditions of this region. Residual errors are assumed due to presence of undetected thin cloud. Following the
principle offuzzy mathematics, thin cloud and clear sky are defined in terms of the difference of SST between satellite image and buoy data. The fuzzy implication relation R is determined between thin cloud / clear sky and the data from different
bands of the AVHRR (ch2, ch2/ch 1 and T3 - T4). The cloud distribution is classified using R, and new retrieval formulae for SST are derived under thin cloud and clear sky separately. It is shown that the classifiedformulae perform better than the
standard global NOAA model in this area. The retrieval accuracy ofSST is improved significantly (from 1.1re to 0.69°C) and the satellite image is used more effectively, since a larger thin cloud area is recovered. |
| GEOSCAN ID | 312114 |
|
|