Análise comparativa entre medidas meteorológicas da estação convencional e automática da Fazenda Lageado no município de Botucatu - São Paulo - Brasil
Carregando...
Data
2021-07-27
Autores
Franco, José Rafael
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
O processo de substituição de uma Estação Meteorológica Convencional (EMC) por uma Estação Meteorológica Automática (EMA) envolve questões de impacto, tais como transferir a confiança entre as diferentes tecnologias e existir duas series históricas para cada elemento. O objetivo foi comparar os dados medidos pela EMC e EMA, verificando se é seguro desativar a EMC e se existe mudança significativa entre as estações. E quando possível fazer a reconstrução das séries históricas a partir da correlação nos dados convencionais. A literatura recomenda que o processo de atualização de tecnologia dos instrumentos deve ser realizado após análise comparativa das medições feitas pelas duas estações (validação). Para comparação e análise dos dados meteorológicos, por meio de regressão linear simples, foi considerado o período de 01/01/2015 a 31/12/2016, da EMC e EMA localizadas no Departamento de Engenharia Rural e Socioeconomia (22º54’S, 48º27’O e 786m) da Faculdade de Ciências Agronômicas (UNESP) de Botucatu - São Paulo - Brasil. As medidas analisadas foram: temperatura máxima (Tmax), mínima (Tmin) e média (Tmed) do ar; umidade relativa média do ar (UR); velocidade média do vento (U2) e precipitação (Prec). Para comparação foram utilizados os indicativos estatísticos d; MBE; rMBE; RMSE; rRMSE; R e R². Para reconstrução das séries históricas foi criado um modelo de regressão linear para estimar os dados da EMA com base nas medidas da EMC utilizando o método bootstrap e validação cruzada K-Fold. Os dados foram divididos aleatoriamente em 70% para criação do modelo de estimativa e 30% para validação em um laço de 1000 repetições, os mesmos indicativos da comparação foram usados para avaliar a capacidade do modelo fazer previsão. A comparação entre a EMC e EMA apresentaram os seguintes resultados, do melhor para o pior, a Prec teve um R² de 0,984, a Tmax teve um R² de 0,961, Tmed teve um R² de 0,917, a Tmin teve um R² de 0,844, a UR teve um R² de 0,811 e, por fim, a U2 teve um R² de 0,746. Apesar das diferenças no período de amostragem, e da resolução dos instrumentos, a comparação apresentou bons resultados. Os resultados apresentam que é seguro desativar a EMC, os modelos de estimativa permitiram reconstruir as séries históricas.
The process of replacing a Conventional Weather Station (EMC) by an Automatic Weather Station (EMA) involves issues of impact, such as transferring trust between different technologies and having two historical series for each element. The objective was to compare the data measured by EMC and EMA, verifying whether it is safe to disable EMC and whether there is significant change between stations. And when possible, reconstruct the historical series from the correlation with conventional data. The literature recommends that the instrument technology update process should be carried out after a comparative analysis of the measurements made by the two stations (validation). For comparison and analysis of meteorological data, through simple linear regression, the period from 01/01/2015 to 12/31/2016 was considered, by EMC and EMA located in the Department of Rural Engineering and Socioeconomics (22º54'S, 48º27'W and 786m) from the Faculty of Agronomic Sciences (UNESP) of Botucatu - São Paulo - Brazil. The measures analyzed were: maximum (Tmax), minimum (Tmin) and average (Tmed) air temperature; mean relative humidity (RH); average wind speed (U2) and precipitation (Prec). For comparison, the statistical indicators d; MBE; rMBE; RMSE; rRMSE; R and R². To reconstruct the historical series, a linear regression model was created to estimate EMA data based on EMC measurements using the bootstrap method and K-Fold cross validation. Data were randomly divided into 70% for estimation model creation and 30% for validation in a 1000-repeat loop, the same comparison indicators were used to assess the model's ability to make predictions. The comparison between EMC and EMA showed the following results, from best to worst, Prec had an R² of 0.984, Tmax had an R² of 0.961, Tmed had an R² of 0.917, Tmin had an R² of 0.844, the UR had an R² of 0.811 and, finally, U2 had an R² of 0.746. Despite the differences in the sampling period and the resolution of the instruments, the comparison showed good results. The results show that it is safe to disable EMC, the estimation models allowed reconstructing the historical series.
The process of replacing a Conventional Weather Station (EMC) by an Automatic Weather Station (EMA) involves issues of impact, such as transferring trust between different technologies and having two historical series for each element. The objective was to compare the data measured by EMC and EMA, verifying whether it is safe to disable EMC and whether there is significant change between stations. And when possible, reconstruct the historical series from the correlation with conventional data. The literature recommends that the instrument technology update process should be carried out after a comparative analysis of the measurements made by the two stations (validation). For comparison and analysis of meteorological data, through simple linear regression, the period from 01/01/2015 to 12/31/2016 was considered, by EMC and EMA located in the Department of Rural Engineering and Socioeconomics (22º54'S, 48º27'W and 786m) from the Faculty of Agronomic Sciences (UNESP) of Botucatu - São Paulo - Brazil. The measures analyzed were: maximum (Tmax), minimum (Tmin) and average (Tmed) air temperature; mean relative humidity (RH); average wind speed (U2) and precipitation (Prec). For comparison, the statistical indicators d; MBE; rMBE; RMSE; rRMSE; R and R². To reconstruct the historical series, a linear regression model was created to estimate EMA data based on EMC measurements using the bootstrap method and K-Fold cross validation. Data were randomly divided into 70% for estimation model creation and 30% for validation in a 1000-repeat loop, the same comparison indicators were used to assess the model's ability to make predictions. The comparison between EMC and EMA showed the following results, from best to worst, Prec had an R² of 0.984, Tmax had an R² of 0.961, Tmed had an R² of 0.917, Tmin had an R² of 0.844, the UR had an R² of 0.811 and, finally, U2 had an R² of 0.746. Despite the differences in the sampling period and the resolution of the instruments, the comparison showed good results. The results show that it is safe to disable EMC, the estimation models allowed reconstructing the historical series.
Descrição
Palavras-chave
Meteorologia, Climatologia, Análise de regressão, Meteorolgy, Climatology, Regression analysis