DAILY PLUVIOMETRIC DATA HOMOGENIZATION:A METHODOLOGICAL CONTRIBUTION

Abstract

Several non-climatic interferences can produce noise in a series of climatic data, especially in the daily unit, with relevance to Climatology. In face of this, the present paper aimed to evaluate the efficiency of the RHtest, technique of homogenization of daily rainfall data, for the Brazilian tropical context. At the international level, the demand for data quality control is increasing, however, given the possibility of altering the observed datas, many doubts arise regarding homogenization; for this reason, the data series (1967-2015) of three rain gauges in the state of Rio de Janeiro, in different topographic contexts, were submitted to the test. From the comparison of the observed and homogenized data series, using Spearman correlation, frequency distribution in rainfall classes and climatic extremes indexes (RClimDex), the potential application of the technique to the Rio de Janeiro was confirmed. The results showed that there is an expressive ordinal correlation between the observed and homogenized data (greater than 99%) and with a 95% significance. In addition, the technique, removing the noises from the series, favored that other climatic indexes presented a tendency with statistical significance. On the other hand, the frequency distribution showed a possible interference of the relief in the results in relation to the Petrópolis rain gauge (leeward), in which the concentration of rain reflects in variations of the threshold; however, the observed differences between the observed and homogenized datas, in terms of extremes, were not greater than 21 mm to Petrópolis rain gauge.