Supplement Finding Climate Characteristics Associated with Primary Modes of Global Climate Variability (original) (raw)

Climate classification revisited: from Köppen to Trewartha

Climate Research, 2014

The analysis of climate patterns can be performed separately for each climatic variable or the data can be aggregated, for example, by using a climate classification. These classifications usually correspond to vegetation distribution, in the sense that each climate type is dominated by one vegetation zone or eco-region. Thus, climatic classifications also represent a convenient tool for the validation of climate models and for the analysis of simulated future climate changes. Basic concepts are presented by applying climate classification to the global Climate Research Unit (CRU) TS 3.1 global dataset. We focus on definitions of climate types according to the Köppen-Trewartha climate classification (KTC) with special attention given to the distinction between wet and dry climates. The distribution of KTC types is compared with the original Köppen classification (KCC) for the period 1961−1990. In addition, we provide an analysis of the time development of the distribution of KTC types throughout the 20th century. There are observable changes identified in some subtypes, especially semi-arid, savanna and tundra.

Class Thresholds Pre-Definition by Clustering Techniques for Applications of ELECTRE TRI Method

Energies

The sorting problem in the Multi-criteria Decision Analysis (MCDA) has been used to address issues whose solutions involve the allocation of alternatives in classes. Traditional multi-criteria methods are commonly used for this task, such as ELECTRE TRI, AHP-Sort, UTADIS, PROMETHEE, GAYA, etc. While using these approaches to perform the sorting procedure, the decision-makers define profiles (thresholds) for classes to compare the alternatives within these profiles. However, most such applications are based on subjective tasks, i.e., decision-makers’ expertise, which sometimes might be imprecise. To fill that gap, in this paper, a comparative analysis using the multi-criteria method ELECTRE TRI and clustering algorithms is performed to obtain an auxiliary procedure to define initial thresholds for the ELECTRE TRI method. In this proposed methodology, K-Means, K-Medoids, Fuzzy C-Means algorithms, and Bio-Inspired metaheuristics such as PSO, Differential Evolution, and Genetic algorith...

The Use of Decision Threshold Adjustment in Classification

Standard classification algorithms are generally designed to maximize the number of correct predictions (concordance). The criterion of maximizing the concordance may not be appropriate in certain applications. In practice, some applications may emphasize high sensitivity (e.g., clinical diagnostic tests) and others may emphasize high specificity (e.g., epidemiology screening studies). This paper considers effects of the decision threshold on sensitivity, specificity, and concordance for four classification methods: logistic regression, classification tree, Fisher's linear discriminant analysis, and a weighted k-nearest neighbor. We investigated the use of decision threshold adjustment to improve performance of either sensitivity or specificity of a classifier under specific conditions. We conducted a Monte Carlo simulation showing that as the decision threshold increases, the sensitivity decreases and the specificity increases; but, the concordance values in an interval around the maximum concordance are similar. For specified sensitivity and specificity levels, an optimal decision threshold might be determined in an interval around the maximum concordance that meets the specified requirement.

Climate patterns of political division units obtained using automatic classification trees

RESUMEN Este artículo propone una metodología para descubrir patrones en datos climatológicos, particularmente temperaturas y precipitación, observados en unidades políticas subnacionales, usando un algoritmo de clasificación automática (un árbol de decisión producido por el algoritmo C4.5). Por lo tanto, los patrones representan árboles de clasificación, en el supuesto de que: 1) cada unidad de división política contiene al menos una estación climatológica y 2) los periodos de registro de las estaciones son relativamente similares en duración y en sus años iniciales y finales. Se produce una serie de modelos de clasificación mediante © 2016 Universidad Nacional Autónoma de México, Centro de Ciencias de la Atmósfera. This is an open access article under the CC BY-NC-ND License (http://creativecommons.org/licenses/by-nc-nd/4.0/). 360 S. R. Coria et al. el uso de diferentes subconjuntos de un conjunto de datos experimentales. Este conjunto de datos contiene información de 3606 estaciones climatológicas en México cuyos periodos de registro tienen diversas dura-ciones, años iniciales y finales. La variable objetivo (dependiente) en todos estos modelos es el nombre de la unidad política (es decir, el estado). Los predictores son 36 características mensuales por cada estación climatológica: 12 corresponden a una temperatura mínima, 12 a una temperatura máxima y 12 a la precipi-tación acumulada. También se usó la altitud como predictor adicional a los 36 mencionados, pero sólo para cuantificar su contribución adicional al modelado. Los resultados muestran que los árboles de clasificación son modelos eficaces para describir y representar los patrones no triviales que caracterizan a las unidades de división política, con base en sus temperaturas y precipitación mensual. Uno de los hallazgos destacables es que la precipitación acumulada de mayo es la característica con el mayor poder discriminatorio en esta tarea de caracterización, lo cual es consistente con el trasfondo teórico de la climatología mexicana. Además, los árboles de clasificación ofrecen alta expresividad a personas poco familiarizadas con aprendizaje automático. ABSTRACT This article proposes a methodology to discover patterns in observed climatologic data, particularly temperatures and rainfall, in subnational political division units using an automatic classification algorithm (a decision tree produced by the C4.5 algorithm). Thus, the patterns represent classification trees, assuming that: (1) every political division unit contains at least one climatological station, and (2) the recording periods of the stations are relatively similar in duration and in their initial and ending years. A series of classification models are produced by using different subsets from an experimental dataset. This dataset contains information from 3606 climatological stations in Mexico with recording periods whose durations, initial and ending years are diverse. The target (dependent) variable in all these models is the name of the political unit (i.e., the state). The predictors are 36 monthly features per each climatological station: 12 features corresponding to a minimum temperature, 12 to a maximum temperature, and 12 to cumulative rainfall. The altitude feature is also used as one of the predictors, in addition to the other 36; however, it is used only to quantify its additional contribution to the modelling. The results show that classification trees are effective models for describing and representing non-trivial patterns to characterize the political division units based on their monthly temperatures and rainfalls. One of the remarkable findings is that the cumulative rainfall of May is the feature with highest discrimination capability to the characterization task, which is consistent with the theoretical background on Mexican climatology. In addition, classification trees offer higher expressivity to non-experts in machine learning.

Temperature/precipitation method for meteorological classification

This work deals with combinations of monthly precipitation air temperature extremity in two dry areas of the Czech Republic (Břeclav, Kladno district) for the period 1961–2010. Precipitation extremity is based on comparison of monthly value with 2nd, 10th, 25th, 75th, 90th and 98th percentile (1961–1991 reference period). Evaluation of extremity of air temperature is based on comparison with standard deviation (1961–1991). Combination of air temperature and precipitation ab/sub/normality was expressed as four quadrant diagram. There are more normal months in Břeclav then in Kladno (196 vs. 184). Amount of months with predisposition of drought occurrence is higher in Kladno (45 vs. 43), but severity is opposite (105 vs. 108). Quadrant of abnormal precipitation and subnormal temperature contains more months with higher severity in Břeclav then in Kladno (27/58 vs. 21/49).

Two-dimensional, threshold-based cloud type classification using MTSAT data

Remote Sensing Letters, 2012

A new two-dimensional threshold diagram (2D-THR) has been developed based on maximum likelihood cloud classification results, which can readily be applied for Multi-functional Transport Satellite (MTSAT) split window datasets. Because 2D-THR was trained using northern summer 2010 data for Japan and its surrounding area, it is typically suitable only for summer. Comparison of snapshot cloud type distributions showed that 2D-THR images and the corresponding night-time microphysical colour composite images as well as 2D-THR images and Japan Meteorological Agency (JMA) cloud type images are in good agreement. A time series inter-comparison of the hourly 2D-THR cloud classification results with the JMA cloud type classification data product was performed by calculating spatial correlation of cloud percentage for 1 • × 1 • grid cells. For cumulonimbus, high-level, middle-level and low-level clouds over tropical and subtropical areas in the northwestern Pacific Ocean region, the spatial correlation between 2D-THR and JMA is moderate. Thus, 2D-THR cloud classification algorithm can be applied in both regions.

The Köppen climate classification as a diagnostic tool for general circulation models

Climate Research, 1993

The Koppen climate classification was applied to the output of atmospheric general circulation models and coupled atmosphere-ocean circulation models. The classification was used to validate model control runs of the present climate and to analyse greenhouse gas warming simulations The most prominent results of the global warming con~putations were a retreat of regions of permafrost and the increase of areas with tropical rainy climates and dry climates.