GAP FILLING PROCEDURES OF CLIMATOLOGICAL SERIES IN THE STATE OF PERNAMBUCO

Abstract

This study aimed to compare the applicability of three methods of filling gaps in rainfall and temperature data from thirteen automatic weather stations (AWS) in the state of Pernambuco, from January to December 2019. The methods used were arithmetic mean, regional weighting, and simple linear regression. The data estimated by filling techniques have been subjected to comparison using R² and descriptive statistical analysis. The estimated data of air temperature presented R² equal or very close to 1 for the three methods. On the other hand, the estimated data of rainfall showed values similar or closer to the real data only to regional weighting (R² = 1) and linear regression (R² = 0.99) methods. The smallest values ​​of standard deviation (1.70) for temperature were obtained with linear regression. The regional weighting method and unfilled data showed greater uniformity for precipitation. The analyzed methods to estimate the climatic variables, air temperature, and precipitation, on a monthly scale, were efficient to fill in missing data in the evaluated AWS. The simple linear regression method is more efficient and adequate, followed by regional weighting, to fill in missing data in climate databases.