/Application of methodologies to estimate missing data of precipitation of the weather station ISER

Abstract

In meteorology and climatology, normally not all the meteorological data are registered due to both anthropogenic and natural factors, like instruments. On the base of this problem and thanks to the precipitation data from 1973- 2013 contributed by IDEAM, the missing data values that were not recorded in the multiannual monthly precipitation series of the ISER PAMPLONA weather station (located in Pamplona, Norte de Santander - Colombia) were estimated using different methodologies. To achieve this, the software SOLAS for missing data analysis software version 4.02 was used, where the value of the missing data was estimated by eight different methodologies: Group means, Hot Deck, Predicted mean imputation, Predictive model based, propensity score method, Predictive mean matching method, Mahalanobis distance and Propensity score/predictive mean/mahalanobis distance combination method. It was determined which of these methodologies resulted to be acceptable in the moment of using them for later analysis of any kind. The graphs of annual precipitation and monthly accumulated for each methodology were obtained and compared, the graphs did not show significant differences between each other. The missing data calculated by each methodology were averaged, obtaining values of 8.89, 25.52, 25.69 and 25.80 mm corresponding to February 1984, March 1976, March 1987 and November 1995, respectively. For multiyear monthly precipitation series of the weather station PAMPLONA ISER is possible the application of any methodology mentioned, this is because the amount of missing data is less than 1%.