ABSTRACT
Zolnier and Campbell-Norman models are compared in this study for hourly estimation of air temperature of two Automatic Weather Stations (Ecologia Agrícola and São Cristóvão), both localized in Rio de Janeiro Metropolitan Region. Hourly temperature observations over 2008-2010 period were evaluated for estimation on seasonal and annual time scales. Air temperature observed and estimated values comparison for both models were based on simple linear regression analysis (determination coefficient - R2) and Standard Error Estimation (SEE). Independent of time scale available (seasonal or annual), Zolnier model highlighted on estimations of air temperature daily variation, with fewer dispersion estimations (R2 = 0.87-0.89) and fewer SEE (1.44-1.84 ºC) in relation to Campbell-Norman model (R2 = 0.71-0.63 and SEE 2.23-2.75 ºC). The coefficient of determination (R2) indicates better precision estimation for models for Ecologia Agrícola, with upper values for spring season (0.91), summer (0.87) and autumn (0.87), and value equal (0.85) for R2 in São Cristóvão for winter. Zolnier model could be adopted for estimation of hourly air temperature in Rio de Janeiro Metropolitan region.
Key words:
empirical methods; simple linear regression; time series
