<![CDATA[Abstrak.Data curah hujan yang akurat, reliabel, dan mendekati waktu nyata adalah faktor penting dalam analisis peramalan dan mitigasi bencara alam hidro klimatologi (banjir, tanah longsor, topan, dan curah hujan ekstrim), pemodelan hidrologi, prakiraan cuaca, perencanaan pertanian, manajemen ekologi, dan manajemen sumber daya air. Observasi curah hujan stasiun menghadapi kendala di Provinsi Bali, terutama pengukuran jarang ditemui di daerah terpencil dan pegunungan. Oleh karena itu, perlu mencari sumber data hujan yang dapat diandalkan seperti produk hujan berbasis satelit, yang menyediakan data dalam waktu mendekati waktu nyata (near real-time), deretan waktu hujan yang tidak terputus dengan resolusi spasial tinggi. Penelitian ini mengevaluasi kinerja produk hujan satelit global yang mendekati waktu nyata dengan 43 stasiun di Provinsi Bali. Produk curah hujan satelit yang dianalisis adalah Integrated Multi-satellitE Retrievals for Global Precipitation Measurement-Early Run (IMERG-ER) dan The Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks - Dynamic Infrared Rain Rate near real-time (PDIR-Now). Selanjutnya, kedua data curah hujan berbasis satelit tersebut dikoreksi menggunakan tiga pendekatan, yaitu koreksi rasio bias, koreksi rata-rata deviasi, dan koreksi nilai fungsi distribusi probabilitas. Metode tradisional berbasis titik ke piksel bersama dengan pengukuran statistik kontinu, metrik kategoris, serta indeks volumetrik diimplementasikan untuk mengevaluasi kinerja produk satelit. Studi ini menunjukkan bahwa meskipun kedua dataset memiliki kelebihan masing-masing, IMERG-ER cenderung lebih konsisten dan andal dalam berbagai kondisi dibandingkan PDIR-Now, terutama setelah koreksi dilakukan. Koreksi nilai fungsi distribusi probabilitas menunjukkan peningkatan kinerja paling signifikan dibandingkan dengan metode koreksi yang lainnya. Hasil studi ini juga mempertegas bahwa koreksi kesalahan perlu dilakukan sebelum data curah hujan berbasis satelit diaplikasikan dan berbagai bidang.Abstract. Accurate, reliable, and near-real-time rainfall data are critical factors for forecasting and mitigating hydro-meteorological natural disasters (such as floods, landslides, typhoons, and extreme rainfall), hydrological modeling, weather forecasting, agricultural planning, ecological management, and water resource management. Rainfall observations from station measurements face challenges in Bali Province, particularly due to the scarcity of measurements in remote and mountainous areas. Therefore, it is necessary to seek reliable sources of rainfall data, such as satellite-based rainfall products, which provide near real-time data, uninterrupted rainfall time series, and high spatial resolution. This research evaluates the performance of global near real-time satellite rainfall products with data from 43 stations across Bali Province. The satellite rainfall products analyzed include the Integrated Multi-satellite Retrievals for Global Precipitation Measurement-Early Run (IMERG-ER) and the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks - Dynamic Infrared Rain Rate near real-time (PDIR-Now). Subsequently, the satellite-based rainfall data were corrected using three approaches: bias ratio correction, mean deviation correction, and probability distribution function value correction. Traditional point-to-pixel methods, along with continuous statistical measurements, categorical metrics, and volumetric indices, were implemented to evaluate the performance of satellite products. The study reveals that although both datasets have their respective strengths, IMERG tends to be more consistent and reliable under various conditions compared to PERSIANN, especially after corrections are applied. The probability distribution function value correction demonstrated the most significant performance improvement compared to the other correction methods. The findings of this study also emphasize the necessity of error correction before satellite-based rainfall data is applied across various fields. Submitted: 2024-09-14 Revisions: 2025-03-06 Accepted: 2024-09-11 Published: 2025-03-14]]>
