Costa, Ana Cristina Marinho daRibeiro, Sara Cristina da SilvaMateu Mahiques, JorgeCaceres, Julia Desiree Velastegui2018-04-102018-04-102017-03-01http://hdl.handle.net/10362/34211Dissertation submitted in partial fulfilment of the requirements for the degree of Master of Science in Geospatial TechnologiesHistorical climate records are relevant since they provide significant information in different environmental studies. However, few climate records are free of non-natural irregularities that imply a problem in the quality of climate data. Considering this problem, many methods have been developed to homogenize climate data. The present project focused on the evaluation of the GSIMCLI (geostatistical simulation for the homogenization of climate data) homogenization method, which is based on a geostatistical stochastic approach, the direct sequential simulation (DSS). The GSIMCLI method was used to homogenize simulated monthly temperature data. This project also included the study of different alternatives for the modelling of the empirical variogram that is part of the parameters in the DSS algorithm. The efficiency of the method was assessed through the calculation of performance metrics, in order to be compared with other homogenization procedures. The literature review on variography was a relevant contribution for the challenging task of modelling small networks’ data. Nonetheless, the results provide evidence that the artificial data of the benchmark data set used lacks a spatial autocorrelation structure. Hence, it was not surprising that GSIMCLI underperformed other homogenization methods. Annual temperature data sets achieved better homogenization results than monthly data sets.engDirect Sequential SimulationGeostatistical SimulationHomogenizationGSIMCLISimulated temperature datamaster thesis201895994