This study revealed the behavior of heavy rainfall before landslide event based on the Weather Research Forecasting (WRF) model. Simulations were carried out to capture the heavy rainfall patterns on 27 November 2018 in Kulonprogo, Yogyakarta. The modeling was performed with three different planetary boundary layer schemes, namely: Yonsei University (YSU), Sin-Hong (SH) and Bougeault and Lacarrere (BL). Our results indicated that the variation of rainfall distribution were small among schemes. The finding revealed that the model was able to capture the radar’s rainfall pattern. Based on statistical metric, WRF-YSU scheme was the best outperforming to predict a temporal pattern. Further, the study showed a pattern of rainfall development coming from the southern coastal of Java before 13:00 LT (Local Time=WIB=UTC+7) and continued to inland after 13:00 LT. During these periods, the new clouds were developed. Based on our analysis, the cloud formation that generated rainfall started at 10:00 LT, and hit a peak at 13:00 LT. A starting time of cloud generating rainfall may be an early indicator of landslide.
Rainfall intensity thresholds only do not take advantage of the awareness of the slope's hydrological processes, so they appear to produce large false and missed alert rates, decreasing the credibility of early warning systems for landslides. This study analyzes this dilemma by modeling the behavior of slopes to precipitation, including the potential effect of soil moisture uncertainty given by numerical modeling. For the simulation of soil moisture during the study period and event rainfall thresholds of an extreme event used to describe the intensity of a rainfall event, the Weather Research and Forecasting (WRF) model is used. The three days simulation conducted during a landslide event in Samigaluh, Kulon Progo on 28 November 2018. The four Planetary Boundary Layer (PBL) parameters in the WRF model are compared to understand each character, i.e., Yonsei University (YSU), Mellor-Yamada-Janjic (MYJ), Shin-Hong (SH), and Bougeault-Lacarrère (BL). To evaluate the precipitation as simulated by WRF, we use observation data from rain gauge and the Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS). In general, all parameters have an underestimation of precipitation. Each PBL parameter's response to rainfall is different. Both MYJ and SH schemes are closer to observation than others for day 1 and day 2 of simulation, daily precipitation. For all PBL schemes, increased soil moisture is seen, suggesting that the soil is wetter and more vulnerable to landslide events. As an early warning predictor of landslides in terms of rainfall parameters, the SH method is very useful in this analysis. For early warning of landslides, a short period (<6 hours) of precipitation with a high accumulation of precipitation would be very beneficial.
ABSTRAKMetode prediksi berbasis statistik prakiraan iklim khususnya curah hujan saat ini telah banyak dikembangkan, salah satunya adalah HyBMG. HyBMG merupakan model prediksi iklim berbasis statistik yang dikembangkan oleh Pusat Penelitian dan Pengembangan (Puslitbang) Badan Meteorologi Klimatologi dan Geofisika (BMKG). Ada 3 metode prediksi univariat yang diujikan dalam aplikasi HyBMG yaitu Adaptive Neuro-Fuzzy Inference System (ANFIS), Autoregressive Integrated Moving Average (ARIMA), dan Transformasi Wavelet. Namun demikian masih ada kendala dalam model ini yaitu running model masih dilakukan satu per satu untuk tiap lokasi dan metode, sehingga pada saat akan melakukan running untuk beberapa titik (lokasi) observasi membutuhkan waktu yang cukup lama. Oleh karena itu, untuk mengatasi hal tersebut dan menghasilkan informasi melalui prediksi iklim yang berkualitas diperlukan model prediksi iklim yang memiliki performa tinggi. Pengembangan HyBMG 2.07 dilakukan agar dapat dijalankan dengan menggunakan data spasial, yang dalam hal ini adalah data curah hujan bulanan seluruh wilayah Indonesia dari hasil penginderaan jauh satelit Tropical Rainfall Measuring Mission (TRMM). Waktu yang diperlukan untuk menjalankan model menjadi jauh lebih cepat. Hasil validasi menunjukkan bahwa prediksi curah hujan bulanan dari ketiga metode yang digunakan masih underestimate bila dibandingkan dengan data observasinya. Namun, metode-metode yang digunakan mampu memprediksi lebih baik untuk wilayah dengan hujan tipe monsun (Jawa, Bali, dan Nusa Tenggara) daripada untuk wilayah dengan tipe hujan ekuatorial dan lokal. Kata Kunci: prediksi iklim, HyBMG, ANFIS, ARIMA, Wavelet ABSTRACTThere are numerous statistics methods to forecast rainfall and tools developed based on these methods. The Center for Research and Development of Agency for Meteorology, Climatology and Geophysics (BMKG) has developed a statistics-based model for rainfall prediction, namely HyBMG. There are 3 univariate prediction methods provided in the HyBMG application, namely Adaptive Neuro-Fuzzy Inference System (ANFIS), Autoregressive Integrated Moving Average (ARIMA), and Wavelet Transformation. However, the application does not have capability to simultaneously predict rainfall over all stations using all methods and hence timeconsuming. Therefore, to produce reliable and timely climate information and predictions, a high-performance climate prediction model is needed. We improve the model by devising it to run spatial data, i.e. measured monthly rainfall data obtained from the Tropical Rainfall Measuring Mission (TRMM) satellite covering Indonesia regions. The running time required decreases significantly compared to the previous model. The three methods in the current model still underestimate the monthly rainfall. However, the methods are better in predicting rainfall over monsoon type region (Java, Bali, Nusa Tenggara, and part of Sumatera) than over equatorial and local regions.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.