New methods and software for cloud detection and classification at high and midlatitudes using Advanced Very High Resolution Radiometer (AVHRR) data are developed for use in a wide range of meteorological, climatological, land surface, and oceanic applications within the Satellite Application Facilities (SAFs) of the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT), including the SAF for Nowcasting and Very Short Range Forecasting Applications (NWCSAF) project. The cloud mask employs smoothly varying (dynamic) thresholds that separate fully cloudy or cloud-contaminated fields of view from cloud-free conditions. Thresholds are adapted to the actual state of the atmosphere and surface and the sun–satellite viewing geometry using cloud-free radiative transfer model simulations. Both the cloud masking and the cloud-type classification are done using sequences of grouped threshold tests that employ both spectral and textural features. The cloud-type classification divides the cloudy pixels into 10 different categories: 5 opaque cloud types, 4 semitransparent clouds, and 1 subpixel cloud category. The threshold method is fuzzy in the sense that the distances in feature space to the thresholds are stored and are used to determine whether to stop or to continue testing. They are also used as a quality indicator of the final output. The atmospheric state should preferably be taken from a short-range NWP model, but the algorithms can also run with climatological fields as input.

Algorithms for cloud detection (cloud mask) and classification (cloud type) at high and midlatitudes using data from the Advanced Very High Resolution Radiometer (AVHRR) on board the current NOAA satellites and future polar Meteorological and Operational Weather Satellites (METOP) of the European Organisation for the Exploitation of Meteorological Satellites have been extensively validated over northern Europe and the adjacent seas. The algorithms have been described in detail in Part I and are based on a multispectral grouped threshold approach, making use of cloud-free radiative transfer model simulations. The thresholds applied in the algorithms have been validated and tuned using a database interactively built up over more than 1 yr of data from NOAA-12, -14, and -15 by experienced nephanalysts. The database contains almost 4000 rectangular (in the image data)-sized targets (typically with sides around 10 pixels), with satellite data collocated in time and space with atmospheric data from a short-range NWP forecast model, land cover characterization, elevation data, and a label identifying the given cloud or surface type as interpreted by the nephanalyst. For independent and objective validation, a large dataset of nearly 3 yr of collocated surface synoptic observation (Synop) reports, AVHRR data, and NWP model output over northern and central Europe have been collected. Furthermore, weather radar data were used to check the consistency of the cloud type. The cloud mask performs best over daytime sea and worst at twilight and night over land. As compared with Synop, the cloud cover is overestimated during night (except for completely overcast situations) and is underestimated at twilight. The algorithms have been compared with the more empirically based Swedish Meteorological and Hydrological Institute (SMHI) Cloud Analysis Model Using Digital AVHRR Data (SCANDIA), operationally run at SMHI since 1989, and results show that performance has improved significantly.

Atmospheric interaction distorts the surface signal received by a space-borne instrument. Images derived from visible channels appear often too bright and with reduced contrast. This hampers the use of RGB imagery otherwise useful in ocean color applications and in forecasting or operational disaster monitoring, for example forest fires. In order to correct for the dominant source of atmospheric noise, a simple, fast and flexible algorithm has been developed. The algorithm is implemented in Python and freely available in PySpectral which is part of the PyTroll family of open source packages, allowing easy access to powerful real-time image-processing tools. Pre-calculated look-up tables of top of atmosphere reflectance are derived by off-line calculations with RTM DISORT as part of the LibRadtran package. The approach is independent of platform and sensor bands, and allows it to be applied to any band in the visible spectral range. Due to the use of standard atmospheric profiles and standard aerosol loads, it is possible just to reduce the background disturbance. Thus signals from excess aerosols become more discernible. Examples of uncorrected and corrected satellite images demonstrate that this flexible real-time algorithm is a useful tool for atmospheric correction.