Homogeneity of monthly mean air temperature of the United Republic of Tanzania with HOMER

Abstract

The long-term climate datasets are widely used in a variety of climate analyses. These datasets, however, have been adversely impacted by inhomogeneities caused by, for example relocations of meteorological station, change of land use cover surrounding the weather stations, substitution of meteorological station, changes of shelters, changes of instrumentation due to its failure or damage, and change of observation hours. If these in homogeneities are not detected and adjusted properly, the results of climate analyses using these data can be erroneous. In this paper for the first time, monthly mean air temperatures of the United Republic of Tanzania are homogenized by using HOMER software package. This software is one of the most recent homogenization software and exhibited the best results in the comparative analysis performed within the COST Action ES0601 (HOME). Monthly mean minimum (TN) and maximum (TX) air temperatures from 1974 to 2012 were used in the analysis. These datasets were obtained from Tanzania Meteorological Agency (TMA). The analysis reveals a larger number of artificial break points in TX (12 breaks) than TN (5 breaks) time series. The homogenization process was assessed by comparing results obtained with Correlation analysis and Principal Component analysis (PCA) of homogenized and non-homogenized datasets. Mann-Kendal non-parametric test was used to estimate the existence, magnitude and statistical significance of potential trends in the homogenized and non-homogenized time series. Correlation analysis reveals stronger correlation in homogenized TX than TN in relation to non-homogenized time series. Results from PCA suggest that the explained variances of the principal components are higher in homogenized TX than TN in relation to non-homogenized time series. Mann-Kendal non-parametric test reveals that the number of statistical significant trend increases higher with homogenized TX (96%) than TN (67%) in relation to non-homogenized datasets.