A PEER-TO-PEER COMPUTING SYSTEM FOR GAP FILLING IN CLIMATE RECORDS

Resumen

This work presents a computing model for filling gaps in climate records obtained from automatic weather stations. A peer-to-peer distributed computing system is used for sharing climate records among the stations and a geographic criterion is adopted for selecting climate records from nearby stations to fill gaps in records. The model was evaluated using real climate records and a simulated data set, with gaps in the series, as inputs. The results obtained were compared to real and to simulation-generated data for estimating the quality of the data generated by the proposed model. The computing model achieved accurate results for the months considered in the case study and it was able to generate climate data that can statistically be considered as belonging to the same population of the real climate data studied.