Development and implementation of an embedded automatic weather station based on IoT technology

This article describes developing and implementing an embedded automatic weather station based on the ESP8266 NodeMCU V3 microcontroller and the ThingSpeak IoT platform in Belem, PA. Temperature and humidity, wind speed and precipitation data were collected using the AM2301, SEN017 and PB10 sensors, respectively, during the period 13/05/2022-14/05/2022, taking hourly readings according to the reference station of the local National Institute of Meteorology (INMET), storing and visualizing the data using the ThingSpeak platform. To compare the data from the developed station with the reference station, statistical methods such as the Coefficient of Determination (R2), Nash -Sutcliffe Coefficient (Ef), Willmott Concordance Index (d), and error indicators Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Mean Percentage Error (PBias) were applied. The results showed that the indices were predominantly classified as 'Very Good' and 'Good' for the temperature and humidity variables, with R2 values of 0.95 and 0.89, respectively. However, the wind speed variable had a low R2 value (0.03). Due to the geographical distance of the stations and the distribution of the variables, it is impossible to infer the real accuracy of the sensors used. However, the analysis demonstrated the station's functionality, suggesting that future work should validate the station developed by extending the observation period and installing it near an automatic weather station with known instrumental error, using its data as a reference.