Prakiraan Cuaca Jangka Pendek Terkalibrasi Menggunakan Metode Geostatistical Output Pertubation

Cuaca merupakan suatu kondisi udara di suatu tempat pada waktu yang relatif singkat yang meliputi kelembaban, kondisi suhu, serta tekanan udara sebagai komponen utama. Hal tersebut menyebabkan cuaca dapat berubah setiap saat. Seiring bekembangnya zaman kebutuhan informasi cuaca saat ini dapat dirasakan semakin penting dibutuhkan. Adapun upaya yang dilakukan oleh lembaga BMKG dalam meningkatkan hasil prakiraan cuaca yaitu mengembangkan Numerical Weather Prediction (NWP). Geostatistical Output Pertubation (GOP) merupakan suatu metode prakiraan cuaca yang diperoleh hanya dari satu luaran deterministik saja, seperti halnya NWP. Akan tetapi, GOP dapat memodifikasi suatu luaran tersebut, sehingga mampu mengkoreksi bias berdasarkan hubungan spasial yang diidentifikasi dari error model. GOP memiliki parameter spasial yang berfungsi untuk memodifikasi hasil prakiraan cuaca sehingga mampu memanfaatkan informasi spasial. Temperatur udara menjadi fokus penelitian, karena memiliki hubungan yang cukup erat dengan elemen cuaca lainnya. Analisis dilakukan dengan mengkalibrasi prakiraan suhu udara di tujuh stasiun meteorologi wilayah Surabaya dan sekitarnya. Tahap awal yang dilakukan yaitu melakukan processing data, kemudian data training digunakan untuk mendapatkan penaksir koefisien regresi selanjutnya menghitung semivarogram empiris dan mengestimasi parameter spasial. Untuk periode training selama 30 hari, prakiraan temperatur udara GOP di 7 stasiun meteorologi menghasilkan RMSE yang sebelumnya bernilai 4,35 menjadi 2,03 pada temperatur maksimum. Dapat disimpulkan bahwa metode GOP mampu mengkoreksi bias dari NWP. =============================================================================================================================== Weather is an air condition in a place for a relatively short time which includes humidity, temperature conditions, and air pressure as the main components. This causes the weather to change at any time. Along with the development of the times, the need for weather information at this time can be felt to be increasingly important. The efforts made by the BMKG agency in improving weather forecasting results are developing a Numerical Weather Prediction (NWP). Geostatistical Output Pertubation (GOP) is a weather forecasting method that is obtained from only one deterministic output, just like the NWP. However, GOP can modify an output, so that it is able to generate a large ensemble based on the spatial relationships identified from the error model. The GOP method can be classified as a multivariate modeling because this method considers a spatial effect simultaneously, even though it only uses one predictor at the modeling stage. Even though in reality the GOP method is applied in the case of NWP deterministic model calibration. GOP has spatial parameters that function to modify weather forecast results so that they are able to utilize spatial information. Air temperature is the focus of research, because it has a fairly close relationship with other weather elements. The analysis was carried out by calibrating air temperature forecasts at seven meteorological stations in Surabaya and its surroundings. The initial stage is to perform data processing, then the training data is used to obtain an estimator of the regression coefficient then calculates the empirical semivarogram and estimates the spatial parameters. For a training period of 30 days, the GOP air temperature forecast at 7 meteorological stations resulted in an RMSE that was previously 4.35 to 2.03 at the maximum temperature. It can be concluded that the GOP method is able to correct the bias of the NWP