Autoria: LUIZ, A. J. B.; SILVA, F. A. M. da

Resumo: Abstract: The use of simulation models is increasingly frequent in research. The greater capacity, both in memory and processing, as well as the great availability, accessibility and amount of data collected and stored, induces and sometimes makes the use of this tool mandatory. On the other hand, when calibrated, the simulation models allow the realization, in seconds and at almost zero cost, of countless estimates of future behavior, in the most different scenarios, which, otherwise, would require a lot of time and resources to be obtained. Crop growth models and climate models are among the most widespread and used in agricultural research. Some metrics have been frequently used by users of these models to assess their efficiency. However, the best techniques are not always employed. In this work, the simulation results of air and soil temperature data were analyzed by two different models: ETA-HADGEM and STICS, respectively. The metrics usually indicated: EF (modelling efficiency) and RMSE (root mean square error), were compared with results of regression analysis and sign test for bias. The introduction of bias correction was also applied and interpreted. It is concluded that the metrics, if used in isolation and without in-depth analysis of the characteristics of the simulated variables, may result in a misinterpretation of the efficiency of the models.

Ano de publicação: 2024

Tipo de publicação: Artigo de periódico

Unidade: Embrapa Meio Ambiente

Palavras-chave: Climate models, Climatologia, ETA-HADGEM, Modelo de Simulação, Regression analysis, Regression model, STICS, Sign test, Simulation models, Temperatura do Solo, bias