Pengembangan Model Jaringan Pos Hujan Berbasis Integrasi Data In Situ dan Satelit di Wilayah Kerja Balai Besar Wilayah Sungai Serayu Opak (BBWS SO) untuk Mendukung Pengelolaan Sumber Daya Air Pertanian

Wahyu Duwi Santoso, Hanggar Ganara Mawandha, S.T., M.Eng., Ph.D. ; Muhamad Khoiru Zaki, S.P., M.P., Ph.D., IPM.

2025 | Skripsi | TEKNIK PERTANIAN

Pengelolaan sumber daya air yang akurat memerlukan data curah hujan yang representatif secara spasial dan temporal. Penelitian ini bertujuan mengidentifikasi variabilitas curah hujan menggunakan data in situ dari 10 DAS wilayah kerja BBWS Serayu Opak (1998–2020) dan data satelit harian GPM-IMERG resolusi 0,1° (1998–2020). Hasil menunjukkan bahwa data satelit memiliki korelasi spasial dan temporal lebih tinggi dibanding in situ, dengan nilai Pearson mencapai 0,91. Namun, uji statistik mengindikasikan bahwa data satelit belum sepenuhnya layak untuk analisis rasionalisasi. Evaluasi jaringan pos hujan eksisting menunjukkan keterbatasan, seperti di DAS Bogowonto yang hanya memiliki 4 pos—jauh di bawah standar WMO untuk wilayah pegunungan tropis (1 pos per 100–250 km²). Analisis dilakukan melalui autokorelasi spasial menggunakan Indeks Moran, interpolasi dengan metode Ordinary Kriging (model semivariogram spherical, exponential, dan gaussian) dengan model gaussian menghasilkan galat terkecil (MAE 1,44 mm untuk satelit), serta rasionalisasi jaringan menggunakan metode Kagan-Rodda. Model LSTM digunakan untuk prediksi curah hujan berbasis data historis. Distribusi pos paling optimal dihasilkan oleh pendekatan segitiga sama sisi (Rekomendasi I) pada semua DAS. Penelitian ini merekomendasikan penempatan pos pada simpul spasial optimal dan wilayah dengan galat tinggi. Integrasi data satelit dan in situ belum efektif untuk merancang jaringan pemantauan curah hujan yang efisien. Tindak lanjut mencakup relokasi, penambahan, dan pengurangan pos berdasarkan standar WMO, Ordinary Kriging, Moran, Kagan-Rodda, dan LSTM.

Accurate water resource management requires rainfall data that is spatially and temporally representative. This study aims to analyze rainfall variability using in situ data from ten watersheds within the jurisdiction of the Serayu Opak River Basin Organization (BBWS Serayu Opak) from 1998 to 2020, alongside daily GPM-IMERG satellite data at a 0.1° spatial resolution for the same period. Findings show that satellite data demonstrates higher spatial and temporal correlation than in situ data, with Pearson correlation values reaching 0.91. However, statistical tests indicate that satellite data is not yet fully reliable for rationalization analysis. Evaluation of the existing rain gauge network reveals significant limitations; for instance, the Bogowonto watershed has only four rain gauges—far below the World Meteorological Organization (WMO) standard for tropical mountainous areas, which requires one station per 100–250 km². Spatial analysis was conducted using Moran’s I for spatial autocorrelation and interpolation using the Ordinary Kriging method with spherical, exponential, and Gaussian semivariogram models, where the Gaussian model produced the lowest error (MAE of 1.44 mm for satellite data). The Kagan-Rodda method was applied to rationalize the rain gauge network, and Long Short-Term Memory (LSTM) modeling was used to predict rainfall based on historical data. The optimal configuration of rain gauge stations was obtained using the equilateral triangle method (Recommendation I), which identified optimal spatial nodes and high-error regions. This study concludes that integrating satellite and in situ data is not yet effective for designing efficient rainfall monitoring networks, recommending future efforts focused on station relocation, addition, or reduction guided by WMO standards, geostatistical analysis, and deep learning models.

Kata Kunci : Hidrologi, rasionalisasi, model, hujan, in situ, satelit, kagan-rodda, ordinary kriging, WMO, indeks moran, LSTM