A Review of Self-Organizing Map Applications in Meteorology and Oceanography

A Review of Self-Organizing Map Applications in Meteorology and Oceanography 255 example, observed and modeled sea level pressure, geopotential height at different pressure levels, air temperature, humidity, precipitation, evaporation, snow, sea ice, etc. Geographically, the SOM meteorological applications are found around the world: the Americas, Africa, Asia, Europe, Arctic and Antarctic (Table 1).The rest of this section is roughly organized by meteorological data type in SOM applications. Sea level pressure and geopotential height dataThe SOM is popular in synoptic climatology, especially in analyzing sea level pressure and geopotential height (Table 1).It is often used to summarize and describe the synoptic patterns of atmospheric circulation as indicated by sea level pressure and geopotential height at different levels, and to relate the characteristic circulation patterns with other meteorological variables.For example, Hewitson & Crane (2002) used SOM to describe synoptic atmospheric circulation changes with time as seen from sea level pressure and to relate the sea level pressure patterns with the precipitation time series.Cassano et al. ( 2006) used the SOM to produce a 55 yr synoptic climatology of daily sea level pressure patterns for the western Arctic, and to study circulation patterns associated with air temperature and high wind extremes.Schuenemann et al. (2009) applied the SOM to the 40-yr European Centre for Medium-Range Weather Forecasts Re-Analysis daily sea level pressure data to objectively identify synoptic sea level pressure patterns over the North Atlantic region.Schuenemann & Cassano (2010a, b) examined the changes of synoptic weather (sea level pressure) patterns from the 15 climate models, and related the SOM extracted circulation patterns with Greenland precipitation in the 20th and 21st centuries.Johnson & Feldstein (2010) presented an SOM analysis that illustrated coupled variability between the North Pacific sea level pressure field and outgoing longwave radiation in the tropical Indo-Pacific region so as to shed light on the relationship between the North Pacific continuum and tropical convection.Reusch et al. (2007) used the SOM to analyze the monthly mean sea level pressure for North Atlantic climate variability.A review of SOM classifications of atmospheric circulation patterns within synoptic climatology is provided in Huth et al. (2008), and an overview in remote sensing applications is seen in Filippi et al. (2010). Air temperature, humidity, and wind dataMultiple variables can be simultaneously handled in the SOM algorithm.Thus, the SOM is often used to examine the patterns of co-variability among several meteorological variables.Cavazos (2000) used the SOM to explore the daily atmospheric variables (circulation and humidity) for climate anomalies of extreme precipitation events over the Balkan region.The SOM was used to discover meaningful intraseasonal evolution of North American monsoon from multiple daily atmospheric variables (850 hPa meridional winds, 700 hPa specific humidity, 500 hPa geopotential heights, and 850-500 hPa thickness), and to reveal interaction of the atmospheric variables during the monsoon evolution (Cavazos et al., 2002).The SOM was also used to classify the midtroposphere variables (700 hPa air temperature, geopotential height and specific humidity) for generalized atmospheric patterns, and to reconstruct the ice-core-based synoptic patterns of climate in Antarctic region (Reusch et al., 2005).SOM classification of the meteorological station data is seen in Raju & Kumar (2007), in which multiple variables (temperature, humidity, wind, sunshine hours and solar radiation, etc) are analyzed.Khedairia & Khadir (2008) also performed a classification analysis of meteorological data of Annaba region (North-East of Algeria) from www.intechopen.comSelf Organizing Maps -Applications and Novel Algorithm Design 256 1995 to 1999 using the SOM and k-means clustering methods.Tambouratzis & Tambouratzis (2008) analyzed long-term (43 years) meteorological data from 128 weather stations in Greece.