p-Index From 2020 - 2025
0.444
P-Index
This Author published in this journals
All Journal Jurnal Natural
Claim Missing Document
Check
Articles

Found 2 Documents
Search

The Monitoring by satelitte image to are and surface runoff in surrounding of governor office of west papua province: Monitoring by satelitte image to are and surface runoff in surrounding of governor office of west papua province Erari, Ishak Semuel; Edi Kuncoro, Edi Kuncoro; Kardiputra, Kwasti K.; Muslimin, Abdul Muis
Jurnal Natural Vol. 20 No. 1 (2024): Jurnal Natural
Publisher : FMIPA Universitas Papua

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30862/jn.v20i1.266

Abstract

The development of Governor's Office of West Papua Province has cause change forest land area become using other land on the surrounding area, so that it has cause increasing the runoff surface wide and the coefficient of runoff . Objective of this study was for know development of the runoff surface wide and the average runoff coefficient with using satelitte image since 2010 to 2023. Results observation of satellite image that the runoff surface wide increase to 99.55 ha (56.25%) and average runoff coefficient increases from 0.32 to 0.56.
PREDIKSI CURAH HUJAN HARIAN KABUPATEN MANOKWARI BERBASIS LONG SHORT-TERM MEMORY Padang, Eohansen; Subgan, Aries A.; Kardiputra, Kwasti K.; Tukan, Tobias T
Jurnal Natural Vol. 21 No. 1 (2025): Jurnal Natural
Publisher : FMIPA Universitas Papua

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30862/jn.v21i1.293

Abstract

An accurate daily rainfall prediction in tropical regions such as Manokwari Regency, West Papua Province, is critically important yet challenging due to high rainfall variability and complex climatic influences, including ENSO, IOD, and orographic effects. This study develops a Long Short-Term Memory (LSTM) model for daily rainfall prediction in Manokwari Regency, utilizing bias-corrected historical data from BMKG (2020-2024) and reanalysis ERA5 ECMWF Data (2005-2024) through Random Forest correction. Four LSTM architectures (Vanilla LSTM, Stacked LSTM, Hybrid CNN-LSTM, and Bidirectional LSTM (BiLSTM)) were evaluated with optimized hyperparameters (window size 30/60 days, batch size 32/64, learning rate 0.001), assessed using root mean squared error (RMSE) and mean absolute error (MAE) metrics. Results demonstrate that the BiLSTM model with two layers (64-32 nodes) and 60-day window size achieved superior performance (RMSE 10.55 mm/day, MAE 6.49 mm/day) compared to other architectures. While the LSTM model effectively captured seasonal rainfall patterns, deviations occurred during extreme events, potentially due to limitations in modeling long-term rainfall dynamics. These findings suggest LSTM's strong potential for early warning systems and water resource management in high-variability rainfall regions.