automatic weather station bmkg

SehinggaBMKG akan menyediakan data, informasi kepada para penggunanya untuk dapat dipergunakan dalam mengenali dan memahami karakteristik unsur-unsur meteorologi, klimatologi, kualitas udara, dan geofisika dengan akurasi tinggi dan tepat waktu. Automatic Weather Station didefinisikan sebagai "stasiun meteorologi di mana pengamatan 7Sensor Weather Station Sekarang Telah Memiliki Sertifikat BMKG. Senin, 21 Maret 2022. Kalibrasi adalah proses pengaturan dan pengecekan tingkat keakurasian alat ukur dengan menggunakan cara membandingkan dengan standar atau tolak ukur. Kalibrasi sangat diperlukan untuk memastikan bahwa hasil dari pengukuran yang dilakukan nantinya akurat IndonesianAgency for Meteorology Climatology and Geophysics (BMKG), Jl. Angkasa I No. 2 Kemayoran, Jakarta 10720, Indonesia kadarsah@ The performance analysis of Automatic Weather Station (AWS) on measuring meteorological parameter of a Total Solar Eclipse (TSE) of 9 March 2016 in Indonesia is conducted by comparing three LAPORANKUNJUNGAN KE BMKG PALEMBANG Termometer Tanah Gundul dan Berumput 8. Automatic Weather Station (AWS) 9. Automatic Rain Water Sampler (ARWS) 10. Oven Pan Evaporimeter 11. Termohygrograf 12. Penakar Hujan Hellman 13. Campbell Stokes 14. Massa Aerosol PM10 3.3 Cara Kerja 3.3.1 Lysimeter Cara kerja : yaitu diamati sekali saja setelah instrument(Automatic Weather Station/AWS) and air temperature measurements using manual instrument. The data that is used in this study are three-hourly data collected from February to June 2016 in 12 (twelve) synoptic stations of the Indonesian Agency for Meteorology Climatology and Geophysics (BMKG), which are Bengkulu, Dabo Singkep, Gunung 누누티비 다운로드. Aplikasi Mobile Info BMKG - Cuaca, Iklim, dan Gempabumi Indonesia Semua informasi mengenai Prakiraan Cuaca, Iklim, Kualitas Udara, dan Gempabumi yang terjadi di berbagai wilayah di Indonesia tercakup dalam satu aplikasi mobile. To improve the quality and quantity of meteorological data over Indonesia, Meteorology Climatology and Geophysics Agency of Indonesia BMKG is continuously developing automatic weather observations. BMKG has 63 units Automatic Weather Station AWS and 165 units Automatic Weather Observation System AWOS both inside and outside the BMKG Station environment. To make the control of sensor conditions easier, especially for temperature, pressure, relative humidity, and rainfall sensors, an additional system is needed to monitor and warn when problems occur with these sensors. The correlation among weather parameters data is the key to monitoring the sensor condition, these data are going to be trained and tested with the Artificial neural network ANN method. Then, the sensor condition normal or error indicated can be well detected based on AWS’s data. The quality improvement of automatic weather station data is expected to increase the utilization of the may be subject to copyright. Discover the world's research25+ million members160+ million publication billion citationsJoin for free Journal of Physics Conference SeriesPAPER • OPEN ACCESSTemperature, pressure, relative humidity and rainfall sensors early errordetection system for automatic weather station AWS with artificialneural network ANN backpropagationTo cite this article P Wellyantama and S Soekirno 2021 J. Phys. Conf. Ser. 1816 012056View the article online for updates and content was downloaded from IP address on 09/03/2021 at 0630 Content from this work may be used under the terms of the Creative Commons Attribution licence. Any further distributionof this work must maintain attribution to the authors and the title of the work, journal citation and under licence by IOP Publishing LtdThe 10th International Conference on Theoretical and Applied Physics ICTAP2020Journal of Physics Conference Series 1816 2021 012056IOP Publishingdoi pressure, relative humidity and rainfall sensors early error detection system for automatic weather station AWS with artificial neural network ANN backpropagation P Wellyantama1 and S Soekirno1 1Physics Department, University of Indonesia, Depok, West Java, Indonesia E-mail pradawellyantama Abstract. To improve the quality and quantity of meteorological data over Indonesia, Meteorology Climatology and Geophysics Agency of Indonesia BMKG is continuously developing automatic weather observations. BMKG has 63 units Automatic Weather Station AWS and 165 units Automatic Weather Observation System AWOS both inside and outside the BMKG Station environment. To make the control of sensor conditions easier, especially for temperature, pressure, relative humidity, and rainfall sensors, an additional system is needed to monitor and warn when problems occur with these sensors. The correlation among weather parameters data is the key to monitoring the sensor condition, these data are going to be trained and tested with the Artificial neural network ANN method. Then, the sensor condition normal or error indicated can be well detected based on AWS’s data. The quality improvement of automatic weather station data is expected to increase the utilization of the data. 1. Introduction Indonesia is a very large archipelago country with an area of about km2, Indonesia has 17,508 islands and a long coastline of about 81,000 km [1]. In Indonesia, weather information has an important role both, to plan and to operate daily life in various sectors. From the construction development, economy, social, transportation, tourism, health, etc. In the construction development sector for buildings, airports and ports require information about wind direction, wind speed, and tides, in the economic sector, the analysis of inflation in a region requires information on wave height, the tourism sector requires weather forecast data, temperature, humidity, wave height, and the land, sea, and air transportation sector requires weather information data, air pressure, wave height, and significant weather maps. The Meteorology Climatology and Geophysics Agency BMKG has 183 Meteorological Stations that observe and provide weather information spread across Indonesia. Weather observations are carried out manually or by using human power to observe weather parameters using conventional weather instruments and there are also automatic observations using digital weather instruments. Of the 183 meteorological stations, 62 use fully automatic observation, and the rest use a conventional instrument. BMKG has 63 units meteorological AWS automatic weather station and 165 units AWOS automatic weather observation system spread throughout Indonesia, both inside and outside of the Meteorological Station zone. Some digital instruments usually unnoticed if there is a problem with the values generated by the sensor, if they are not compared to other instruments or if there is no event that validates the value. This makes the The 10th International Conference on Theoretical and Applied Physics ICTAP2020Journal of Physics Conference Series 1816 2021 012056IOP Publishingdoi process plays an important role to maintain data quality. BMKG always calibrates the equipment every 6 months, but between the 6 months it does not rule out the possibility of potential problems in measuring values, especially for electronic or digital equipment. The eligibility conditions for meteorological instruments adhere to the regulations of the World Meteorological Organization WMO CIMO of 2014, where the measurement tolerances are 1 temperature maximum of 2 humidity maximum 3%, 3 air pressure maximum of hPa, 3 maximum wind speed of m/s, 4 wind direction maximum 5o, 5 rainfall maximum 5%, 6 sun radiation maximum 5%. To make the control of sensor conditions easier, especially temperature, pressure, humidity, and rainfall sensors, we need a system that can monitor and detect when problems occur with these sensors. The correlation among weather parameters is the key to controlling the sensor conditions to be trained and tested using the ANN backpropagation method. This ANN system design works by learning the correlation and pattern of each sensor data during the training phase. In the testing phase, the condition of the test data will be predicted. If any sensor outputs a value that is unusual or different from the pattern studied by ANN, the system will give a warning indicating sensor failure. With better quality weather observation data, it will improve the quality of providing weather information, so that the use of weather information becomes more accurate and useful. In a study [2] entitled Machine Learning-Based Calibration of Low-Cost Air Temperature Sensors Using Environmental Data, and research [3] entitled Temperature error correction based on BP neural network in meteorological wireless sensor network, they tried to approach a calibration using software and models, but only limited to the temperature sensor. In this study, we try to do the same approach, but for more sensors. The next approach to sensor error detection is studied based on the correlation pattern among sensors, this was done in a study [4] entitled Soft Sensors for Instrument Fault Accommodation in Semiactive Motorcycle Suspension Systems. The detection of a condition in classification has been carried out in a research conducted by [5] entitled Intelligent Multi-Sensor Control Device for Recognition of Gas-Air Mixture Samples with the Use of Artificial Neural Networks, which classifies and detection odors with electronic noses using ANN. From the researches above, the ANN model has good results, so this paper will try to apply the ANN-BP method for an early detection approach for error indication of more than one sensor on AWS in a result that is classified as error or normal. 2. Method Figure1. Schematic of the AWS sensor condition early detection system. The 10th International Conference on Theoretical and Applied Physics ICTAP2020Journal of Physics Conference Series 1816 2021 012056IOP Publishingdoi design of the early detection system begins with the design of the ANN backpropagation model, the model is built with pattern recognition in training data on observations of weather parameters, temperature, humidity, pressure, and rain at Tanjung Priok Maritime Station for 4 years, from 2017 until 2020. The data training is carried out using Rstudio software. The composition of data for training is 80% data. The training is carried out so that the network can recognize the patterns generated from the input and output pairs. The data input consists of weather parameters, temperature, humidity, pressure, and rain, and the output is a label of the sensor's condition, normal or an error indication. After the model produces the best accuracy in training and testing data, then the ANN model is used to estimate and to detect the condition of the AWS sensors, especially pressure, temperature, humidity, and rain sensors. The details of the research steps are Preprocessing data Before the data was processed using ANN, the data were compiled and conditioned, with a composition of ± 50% actual data and ± 50% in the form of synthetic data. The synthetic data mean the actual data that has been added and subtracted in value according to WMO CIMO regulation 2014 to obtain data in the form of damaged sensor label values. Figure 2. AWS Tanjung Priok. ANN Design ANN design is done by determining the amount of input data used in training, the number of hidden layers used and the number of outputs desired. The data used as input are temperature, humidity, pressure, and rain observation data at the Tanjung Priok Maritime Meteorological Station from 2017 to 2020, with details of the network architecture as follows Figure 3. ANN architecture of temperature and humidity. The 10th International Conference on Theoretical and Applied Physics ICTAP2020Journal of Physics Conference Series 1816 2021 012056IOP Publishingdoi 4. ANN architecture Air pressure. Figure 5. ANN rainfall architecture. Figure 6. Research algorithm. The 10th International Conference on Theoretical and Applied Physics ICTAP2020Journal of Physics Conference Series 1816 2021 012056IOP Publishingdoi Pattern Recognition training. In the training process, the maritime meteorological station’s conventional weather observation data for 4 years are arranged into 2 output conditions, namely 1 Output conditions "sensor in normal conditions", where all input values are the original values of weather observations for the past 4 years. 2 The output condition is "problematic sensor", where all input values are added and also subtracted from the value that exceeds the tolerance limits of the CIMO World Meteorological Organization WMO No. 8 of 2014, where the measurement tolerance is as follows a temperature maximum b Humidity maximum 3%, c Air pressure maximum of hPa , d Rainfall maximum 5%. The input and output data during the training are in the form of 1 Input temperature, humidity, and pressure data, the output temperature sensor label indication is damaged or normal, 2 Input temperature, humidity, and pressure data, the output humidity sensor label indicates damaged or normal . 3 Temperature, humidity, and pressure data input, the output pressure indication is damaged or normal. 4 Temperature, humidity, and rain data input, output rain label indication of damage or normal in all rain categories except 1-3mm rain which has additional pressure data input. Testing and estimation Data testing is carried out aimed to determine whether the network can recognize patterns of training data from the input data provided. If the resulting error value has reached the target, the resulting output can be used as estimation data. The model validation value is obtained from the accuracy coefficient with the following value interpretation Table 1. The relation between accuracy coefficient and interpretation [6] - 20 % - % - % - % - 100 % Very low Low Moderate High Very high The estimation is done after the pattern recognition process is carried out by the network when the training is complete and the model has been tested with good accuracy values. Input data consist of AWS Tanjung Priok’s temperature, humidity, pressure, and rain data and the output is a classification of sensor conditions a Normal, or b The temperature sensor is indicated as damaged, or c The humidity sensor is indicated as damaged, or d The pressure sensor is indicated as damaged, or e The rain sensor is indicated to be damaged. 3. Result and Discussion Test result Temperature Sensor. After the data training was carried out, then testing was carried out with the remaining 20% of the data, with the target data being the previously known sensor conditions. In the testing temperature sensor conditions, obtained a very high accuracy value is 99%, false negative prediction is “normal”, which it should “error indication” value is and false positive prediction is “error indication”, which it should “normal” value is with the graph of the independent variable contribution as follows The 10th International Conference on Theoretical and Applied Physics ICTAP2020Journal of Physics Conference Series 1816 2021 012056IOP Publishingdoi 7. Contribution of the independent variable, temperature sensor label output. The figure above shows the intensity of the contribution of the independent variable in training and testing for the output temperature sensor condition label, where the highest contribution is the value of the temperature sensor itself. Humidity Sensors. After the data training was carried out, testing was carried out with the remaining 20% of the data, with the target data being in the form of previously known conditions. In testing the temperature sensor conditions, a very high accuracy value was obtained false negative prediction is “normal”, which it should “error indication” value is and false positive prediction is “error indication”, which it should “normal” of with a graph of the independent variable contribution as follows Figure 8. Contribution of the independent variable, humidity sensor label output. The figure above shows the intensity of the contribution of the independent variable in training and testing for the output humidity sensor condition label, where the highest contribution is the value of the humidity sensor itself. Pressure Sensor. After the data training was carried out, testing was carried out with the remaining 20% of the data, with the target data being in the form of previously known conditions. In testing the temperature sensor conditions, obtained a very high accuracy value of 100%, false negative prediction is “normal”, which it should “error indication” value is 0% and false positive prediction is “error The 10th International Conference on Theoretical and Applied Physics ICTAP2020Journal of Physics Conference Series 1816 2021 012056IOP Publishingdoi which it should “normal” value is 0%, with a contribution graph independent variable as follows Figure 9. Contribution of the independent variable, pressure sensor label output. The figure above shows the intensity of the contribution of the independent variable in training and testing for the output Pressure sensor condition label, where the highest contribution is the value of the Pressure sensor itself. Rain Sensor. After the data training was carried out, testing was carried out with the remaining 20% of the data, with the target data being in the form of previously known conditions. In testing the temperature sensor conditions, obtained a very high accuracy value on average of 82%, an average false negative prediction is “normal”, which it should “error indication” value is and false positive prediction is “error indication”, which it should “normal” value is with details a Rainfall 1-3 mm, the test accuracy is 77%, false-negative and false-positive b Rainfall 3-20 mm testing accuracy is 82%, false-negative 0%, and false-positive c Rainfall 20-50 mm has 82% accuracy testing, 0% false-negative and false-positive. d Rainfall above 50 mm has 91% accuracy testing, false-negative 0%, and false-positive With the graph of the independent variable contribution as follows Figure 10. Contribution of the independent variable, 1-3mm, and 3-20mm rain sensor label output. The 10th International Conference on Theoretical and Applied Physics ICTAP2020Journal of Physics Conference Series 1816 2021 012056IOP Publishingdoi 11. Contribution of the independent variable, rain sensor label output 20-50mm and> 50mm. The Figure above shows the intensity of the contribution of the independent variable in training and testing for the output rain sensor condition label, where the highest contribution is the value of the rain sensor itself. Estimation Results After the training and data testing process, based on the high accuracy results above, the sensor condition estimation process is carried out. The data to be estimated is the latest AWS Tanjung Priok data on October 16 - 18, 2020 with the following results Table 2. The estimation results of the AWS Tanjung Priok sensor condition label. Estimated of sensor condition labels Error Indication for Pressure sensor Error Indication for Pressure sensor The 10th International Conference on Theoretical and Applied Physics ICTAP2020Journal of Physics Conference Series 1816 2021 012056IOP Publishingdoi on the model obtained from training and tested with previous data, and used to estimate the AWS Tanjung Priok sensor data for 16-18 October 2020, it was found that almost all were in normal condition, 2 conditions indicated that the pressure sensor had an error, on October 17 at and UTC, which can be seen at those 2 times the pressure value suddenly decreased significantly, but other weather parameters were still in conditions not much different from the previous time. 4. Conclusion The sensor condition, especially temperature, humidity, pressure, and rain on AWS Tanjung Priok can be estimated using the ANN backpropagation method, where the accuracy results between the model output and the target during training and testing show very high values. Based on this model, the estimation results of the AWS Tanjung Priok sensor conditions on 16-18 October 2020 are almost all in normal conditions, 2 conditions indicated that the pressure sensor had an error, on October 17 at and UTC, this can be seen at the 2 times the pressure value decreased significantly, but other weather parameters are still not much different from the previous time. Based on the results of this estimation, it is hoped that it can serve as a warning to the nearest Maritime Meteorological Station so that checks can be carried out as soon as possible and if damage occurs, replacement or repair of sensor hardware can be carried out so that the quality of AWS data can always be maintained. Acknowledgment This research was supported by the grant of PITTA Publikasi Internasional Terindeks Untuk Tugas Akhir Mahasiswa of Universitas Indonesia under the contract number NKB-1005/ We would like to acknowledge the Indonesian Agency for Meteorology Climatology and Geophysics for supporting data and facilities. References [1] Dahuri R 2004 Pengelolaan Sumber Daya Wilayah Pesisir dan Lautan Secara Terpadu, Edisi Revisi Jakarta Pradnya Paramita [2] Yamamoto K, Togami T, Yamaguchi N, Ninomiya S 2017 Machine Learning-Based Calibration of Low-Cost Air Temperature Sensors Using Environmental Data. Sensors 176 1290 [3] Wang B 2017 Temperature error correction based on BP neural network in meteorological wireless sensor network. Int. J. Sensor Networks 234 [4] Capriglione D, et al 2020 Soft Sensors for Instrument Fault Accommodation in Semiactive Motorcycle Suspension Systems IEEE transactions on instrumentation and measurement 69 5 [5] Kulagin V P, et al 2017 Intelligent Multi-Sensor Control Device for Recognition of Gas-Air Mixture Samples with the Use of Artificial Neural Networks IEEE [6] Sugiyono 2008 Metode Penelitian Kunatitatif Kualitatif dan R&D Bandung Alfabeta ... Artificial Neural Networks ANNs are frequently used in meteorology science CIE and cloud classification [40,41], solar irradiance and wind speed forecasting [42][43][44][45][46][47], atmospheric pollution distribution [48,49], and rainfall [50,51]. ANN classification models serve to classify input information into certain categories or targets. ...Digital sky images are studied for the definition of sky conditions in accordance with the CIE Standard General Sky Guide. Likewise, adequate image-processing methods are analyzed that highlight key image information, prior to the application of Artificial Neural Network classification algorithms. Twenty-two image-processing methods are reviewed and applied to a broad and unbiased dataset of 1500 sky images recorded in Burgos, Spain, over an extensive experimental campaign. The dataset comprises one hundred images of each CIE standard sky type, previously classified from simultaneous sky scanner data. Color spaces, spectral features, and texture filters image-processing methods are applied. While the use of the traditional RGB color space for image-processing yielded good results ANN accuracy equal to other color spaces, such as Hue Saturation Value HSV, which may be more appropriate, increased the accuracy of their global classifications. The use of either the green or the blue monochromatic channels improved sky classification, both for the fifteen CIE standard sky types and for simpler classification into clear, partial, and overcast conditions. The main conclusion was that specific image-processing methods could improve ANN-algorithm accuracy, depending on the image information required for the classification WangZhi DengKe XuTao LiuIn recent years, meteorological environment has become a topic of concern to people. Various meteorological disasters threaten human life and production. Accurate and timely acquisition of meteorological data has become a prerequisite for dealing with various aspects of production and life, and also laid a foundation for weather prediction. For a long time, meteorological data acquisition system combined with modern information technology has gradually become a hot spot in the field of meteorological monitoring and computer research. The continuous development of NB-IoT technology has brought new elements to the research of meteorological monitoring system. This paper designs a weather station system based on NB-IoT, including data acquisition module, main controller module, NB-IoT wireless communication module, energy capture module, low power consumption scheme, measurement of air temperature is strongly influenced by environmental factors such as solar radiation, humidity, wind speed and rainfall. This is problematic in low-cost air temperature sensors, which lack a radiation shield or a forced aspiration system, exposing them to direct sunlight and condensation. In this study, we developed a machine learning-based calibration method for air temperature measurement by a low-cost sensor. An artificial neural network ANN was used to balance the effect of multiple environmental factors on the measurements. Data were collected over 305 days, at three different locations in Japan, and used to evaluate the performance of the approach. Data collected at the same location and at different locations were used for training and testing, and the former was also used for k-fold cross-validation, demonstrating an average improvement in mean absolute error MAE from to by applying our method. Some calibration failures were noted, due to abrupt changes in environmental conditions such as solar radiation or rainfall. The MAE was shown to decrease even when the data collected in different nearby locations were used for training and testing. However, the results also showed that negative effects arose when data obtained from widely-separated locations were used, because of the significant environmental differences between paper describes the development and experimental verification of an Instrument Fault Accommodation IFA scheme for front and rear suspension stroke sensors in motorcycles equipped with electronic controlled semi-active suspension systems. In particular, the IFA scheme is based on the use of Nonlinear Auto-Regressive with eXogenous inputs NARX Neural Networks NN employed as soft sensors for feeding the suspension control strategy back with measurement even in presence of faults occurred on the sensors. Different NN architectures have been trained and tuned by considering real data acquired during several measurement campaigns. The performance has been compared with that of the well-known Half-Car Model HCM. Very satisfying results allow the Soft sensor to be really integrated into fault-tolerant control systems. In experimental road tests an implementation of the proposed IFA scheme on a low-cost microcontroller for automotive applications, showed to be in real-time. In the paper these experimental results are shown to prove the good performance of the IFA scheme in different motorcycle operating conditions. Baowei WangXiaodu GuLi MaShuangshuang YanUsing meteorological wireless sensor network WSN to monitor the air temperature AT can greatly reduce the costs of monitoring. And it has the characteristics of easy deployment and high mobility. But low cost sensor is easily affected by external environment, often leading to inaccurate measurements. Previous research has shown that there is a close relationship between AT and solar radiation SR. Therefore, We designed a back propagation BP neural network model using SR as the input parameter to establish the relationship between SR and AT error ATE with all the data in May. Then we used the trained BP model to correct the errors in other months. We evaluated the performance on the datasets in previous research and then compared the maximum absolute error, mean absolute error and standard deviation respectively. The experimental results show that our method achieves competitive performance. It proves that BP neural network is very suitable for solving this problem due to its powerful functions of non-linear fitting. NASA/ADS Abstract To improve the quality and quantity of meteorological data over Indonesia, Meteorology Climatology and Geophysics Agency of Indonesia BMKG is continuously developing automatic weather observations. BMKG has 63 units Automatic Weather Station AWS and 165 units Automatic Weather Observation System AWOS both inside and outside the BMKG Station environment. To make the control of sensor conditions easier, especially for temperature, pressure, relative humidity, and rainfall sensors, an additional system is needed to monitor and warn when problems occur with these sensors. The correlation among weather parameters data is the key to monitoring the sensor condition, these data are going to be trained and tested with the Artificial neural network ANN method. Then, the sensor condition normal or error indicated can be well detected based on AWS’s data. The quality improvement of automatic weather station data is expected to increase the utilization of the data. Publication Journal of Physics Conference Series Pub Date February 2021 DOI Bibcode 2021JPhCS1816a2056W

automatic weather station bmkg