AKURASI PREDIKSI CURAH HUJAN HARIAN OPERASIONAL DI JABODETABEK : PERBANDINGAN DENGAN MODEL WRF

Abstract

Akurasi prakiraan curah hujan harian operasional yang dibuat oleh Badan Meteorologi Klimatologi dan Geofisika (BMKG) dikaji dengan cara diverifikasi berdasarkan kategori hujan dikotomi, lebat dan sangat lebat terhadap data dari 25 titik pengamatan di Jakarta. Prosedur yang sama juga diterapkan pada prakiraan curah hujan model Weather Research and Forecasting (WRF) dengan teknik multi-nesting yang di-downscale dari keluaran Global Forecast System (GFS). Hasilnya memperlihatkan bahwa kedua metode prediksi tersebut memiliki akurasi yang baik untuk prediksi dikotomi tetapi hampir gagal dalam memprediksi curah hujan lebat dan sangat lebat. Khususnya, kegagalan prediksi operasional dalam mendeteksi tiga kejadian hujan sangat lebat dalam periode kajian. Dalam hal ini, model WRF yang cenderung menghasilkan false alarm memperlihatkan prospek yang bagus untuk pengembangan sistem prediksi cuaca skala lokal/regional yang lebih akurat di Indonesia.

The accuracy of daily rainfall forecasts produced operationally by the Meteorological, Climatological, and Geophysical Agency (BMKG) has been assessed by verifying the prediction of dichotomous, heavy, and very heavy rain events against observed data at 25 stations in Jakarta. Similar procedure was applied to raw hindcasts performed  using the Weather Research and Forecasting (WRF) model with multi-nesting technique up to 3 km resolution downscaled from NOAA global forecast system (GFS) outputs.  The results show that both forecasts have quite favorable accuracy for dichotomous rain events but almost no meaningful score for the predictions of heavy and very heavy rain events was obtained. Particularly, none of as many as three observed very heavy rain events was predicted by the operational forecast. In this case, WRF tend produce false alarms indicating a better prospect for future development of more accurate local/regional weather forecasting system in Indonesia.