Classification by multiple regression - a new approach towards the classification of extremes
Journal Title: Meteorology Hydrology and Water Management Research and Operational Applications - Year 2016, Vol 4, Issue 1
Abstract
There are numerous algorithmic classification methods that address the connections between different scales of the atmosphere, such as EOFs, clustering, neural nets. However, their relative strength encompasses the mean conditions. Extremes are poorly covered. A novel approach towards the identification of linkages between large-scale atmospheric fields and local extremes of meteorological parameters is presented. The principle is that objectively selected fields can be used to circumscribe a local meteorological parameter by way of regression. For each day, the regression coefficients form a pattern, source of a classification based on similarity. Several classes are generated which contain extreme atmospheric conditions and from which local meteorological parameters are computed, yielding an indirect way of determining local extremes from large-scale information. Not only can local meteorological parameters be subjected to this classification procedure. It can be extended to extreme indicators, e.g., threshold exceedances, yielding the relevant atmospheric fields to describe those indicators and grouping days with “favourable atmospheric conditions”. This approach can be extended by investigating networks of stations from a region and describing, e.g., the probability for threshold exceedances at a given percentage of the network. Not only can the method be used as a tool to objectively identify days in the current climate with extreme conditions. The method can be applied to climate projections, using the previously found parameter-specific combinations of atmospheric fields. From those fields, which constitute the modeled future climate, local time series can be generated which are then analyzed with respect to the frequency and magnitude of future extremes. The method has sensitivities (i) due to the degree to which there are connections between large-scale fields and local meteorological parameters (measured, e.g., by the correlation) and (ii) due to the varying quality of the different fields (geopotential, temperature, humidity et cetera) projected by the climate model.
Authors and Affiliations
Arne Spekat, Wolfgang Enke, Frank Kreienkamp
Keywords
Related Articles
- EP ID EP572905
- DOI 10.26491/mhwm/62366
- Views 90
- Downloads 0
Floods and droughts as a result of deformability of the geological environment
The water content of the rivers is formed by atmospheric precipitation and underground waters. Influence of underground waters on water content of the rivers cannot be measured. It is shown that the volume of underground...
Models, manifestation and attribution of climate change
Results from a series of five surveys among five groups of international climate scientists about their evaluation of elements of climate models and of climate change are presented. The first survey was done in 1996, the...
The importance of hourly nutrient concentration variability in terms of assessment of the surface water state in Słupia pilot river
Water eutrophication still causes many environmental and social problems, and so better quantification of the amount of nutrient loads discharged into surface waters is required. The basis for eutrophication analysis is...
Developing criteria for a stakeholder-centred evaluation of climate services: the case of extreme event attribution for storm surges at the German Baltic Sea
Science-based knowledge about climate-related hazards is an inevitable part of the knowledge basis needed for many stakeholders’ decision-making. Despite continuous advances in climate science, much of this knowledge is...
The choice of optimal lag for Kriging interpolation of NWP model forecast
In this paper a Kriging method is reviewed and a way of its application in numerical weather prediction is proposed. The basic principles of the Kriging are shown; the main advantage is its accuracy, but at the same time...