Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/171444Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.advisor | Nurdiati, Sri | - |
| dc.contributor.advisor | Budiarti, Retno | - |
| dc.contributor.author | Rafhida, Syukri Arif | - |
| dc.date.accessioned | 2025-11-03T23:30:02Z | - |
| dc.date.available | 2025-11-03T23:30:02Z | - |
| dc.date.issued | 2025 | - |
| dc.identifier.uri | http://repository.ipb.ac.id/handle/123456789/171444 | - |
| dc.description.abstract | Perubahan iklim telah menyebabkan meningkatnya ketidakpastian dalam pola curah hujan di berbagai wilayah di Indonesia, termasuk kawasan Danau Toba yang merupakan daerah penting secara ekologis, sosial, dan ekonomi. Pemanfaatan data proyeksi iklim dari model global seperti Coupled Model Intercomparison Project Phase 6 (CMIP6) semakin krusial dalam mendukung kebijakan adaptasi terhadap perubahan iklim. Namun, data hasil keluaran model iklim tersebut umumnya mengandung bias sistematis yang perlu dikoreksi agar dapat merepresentasikan kondisi iklim lokal secara lebih akurat. Oleh karena itu, penelitian ini dilakukan untuk mengidentifikasi sebaran probabilistik yang sesuai, menerapkan metode koreksi bias statistik, mengevaluasi efektivitas koreksi bias, serta menyusun proyeksi curah hujan masa depan yang lebih realistis untuk wilayah Danau Toba. Penelitian ini menggunakan data observasi curah hujan bulanan dari tahun 1973 hingga 2014 yang diperoleh dari tujuh stasiun di sekitar Danau Toba, serta data model iklim CMIP6 historis dan dua skenario proyeksi (SSP2-4.5 dan SSP5-8.5) untuk periode 1950 hingga 2050. Proses koreksi bias dilakukan menggunakan metode Quantile Delta Mapping (QDM), yang bekerja dengan menyesuaikan fungsi distribusi kumulatif antara data observasi dan model, serta mempertahankan perubahan relatif antar waktu. Dua pendekatan digunakan dalam proses koreksi: pendekatan pertama berdasarkan distribusi per bulan (QDM1), dan pendekatan kedua menggunakan distribusi keseluruhan data (QDM2). Identifikasi sebaran probabilistik dilakukan menggunakan sepuluh jenis distribusi dan estimasi parameter dengan metode Maximum Likelihood Estimation (MLE), sementara evaluasi kinerja dilakukan menggunakan Mean Absolute Error (MAE) dan uji Kolmogorov–Smirnov. Hasil penelitian menunjukkan bahwa distribusi generalized extreme value (GEV) merupakan sebaran terbaik untuk sebagian besar bulan dalam menggambarkan karakteristik curah hujan wilayah Danau Toba, terutama untuk periode dengan curah hujan ekstrem. Sebaran Weibull dan logistik juga menunjukkan kecocokan yang signifikan pada bulan-bulan tertentu. Metode koreksi bias QDM berhasil memperbaiki kesesuaian antara data model dan data observasi, dengan pendekatan QDM1 memberikan hasil koreksi yang lebih stabil dibandingkan QDM2. Penurunan nilai MAE dan nilai statistik KS menunjukkan bahwa model CMIP6 setelah koreksi bias memiliki performa yang lebih baik, terutama dalam mempertahankan karakteristik musiman dan pola distribusi curah hujan. Meskipun demikian, efektivitas koreksi bias cenderung menurun pada bulan-bulan dengan intensitas curah hujan ekstrem di awal dan akhir tahun. Proyeksi curah hujan masa depan yang telah dikoreksi menunjukkan adanya fluktuasi musiman yang lebih realistis, terutama pada musim DJF dan MAM. Sebelum koreksi, model CMIP6 menunjukkan kecenderungan overestimate yang signifikan, yang berhasil dikurangi setelah penerapan koreksi bias QDM. Skema QDM1 mampu mempertahankan dinamika perubahan kuantil dan menghasilkan proyeksi yang mendekati pola historis. Hasil ini memberikan gambaran yang lebih dapat diandalkan mengenai kondisi curah hujan di masa depan, yang dapat dimanfaatkan sebagai dasar dalam penyusunan kebijakan adaptasi perubahan iklim berbasis bukti di wilayah Danau Toba. Penelitian ini memberikan kontribusi penting dalam pengembangan metode koreksi bias berbasis distribusi probabilistik untuk data iklim global, serta menunjukkan pentingnya pemilihan sebaran statistik yang tepat dalam proses koreksi. Temuan ini juga menegaskan bahwa pendekatan kuantil dinamis seperti QDM lebih unggul dalam mempertahankan karakteristik statistik data dibanding metode konvensional. Implikasi dari hasil ini sangat relevan bagi perencanaan sektor sumber daya air, pertanian, dan pengurangan risiko bencana di kawasan Danau Toba, serta membuka ruang pengembangan metode yang lebih adaptif dan berbasis data lokal di masa mendatang. | - |
| dc.description.abstract | Climate change has led to increasing uncertainty in rainfall patterns across various regions in Indonesia, including the Lake Toba area, which holds significant ecological, social, and economic importance. The utilization of climate projection data from global models such as the Coupled Model Intercomparison Project Phase 6 (CMIP6) is becoming increasingly crucial in supporting adaptation policies in response to climate change. However, outputs from these climate models typically contain systematic biases that must be corrected to more accurately represent local climate conditions. Therefore, this study was conducted to identify suitable probabilistic distributions, apply statistical bias correction methods, evaluate their effectiveness, and develop more realistic future rainfall projections for the Lake Toba region. This research utilized monthly observed rainfall data from 1973 to 2014 collected from seven meteorological stations around Lake Toba, along with historical CMIP6 climate model data and two future projection scenarios (SSP2-4.5 and SSP5-8.5) for the period 1950 to 2050. The bias correction process was performed using the Quantile Delta Mapping (QDM) method, which adjusts the cumulative distribution function between observed and model data while preserving relative changes over time. Two approaches were employed in the correction process: the first based on monthly distributions (QDM1), and the second using the distribution of the entire dataset (QDM2). Probabilistic distribution identification was conducted using ten types of distributions, with parameter estimation performed via the Maximum Likelihood Estimation (MLE) method. Model performance was evaluated using the Mean Absolute Error (MAE) and the Kolmogorov–Smirnov test. The results indicate that the generalized extreme value (GEV) distribution is the most suitable for representing rainfall characteristics across most months in the Lake Toba region, especially during periods of extreme rainfall. The Weibull and logistic distributions also showed good fit for certain months. The QDM bias correction method successfully improved the alignment between model outputs and observed data, with the QDM1 approach producing more stable correction results compared to QDM2. Reductions in MAE values and improved Kolmogorov–Smirnov statistics demonstrate that the corrected CMIP6 models performed better, particularly in preserving seasonal characteristics and rainfall distribution patterns. Nonetheless, the effectiveness of bias correction tends to decline during months with extreme rainfall intensities at the beginning and end of the year. Corrected future rainfall projections showed more realistic seasonal fluctuations, particularly during the DJF and MAM seasons. Before correction, CMIP6 models exhibited a significant tendency to overestimate rainfall, which was substantially reduced following the application of QDM. The QDM1 scheme effectively preserved the dynamics of quantile changes and produced projections that more closely mirrored historical patterns. These results offer more reliable insights into future rainfall conditions, which can serve as a solid basis for developing evidence-based climate adaptation policies in the Lake Toba region. This study contributes significantly to the development of probabilistic distribution-based bias correction methods for global climate data and highlights the importance of selecting appropriate statistical distributions in the correction process. The findings also underscore that dynamic quantile-based approaches such as QDM are superior in maintaining the statistical characteristics of climate data compared to conventional methods. The implications of these results are highly relevant for planning in the water resources, agriculture, and disaster risk reduction sectors in the Lake Toba region, while also paving the way for the development of more adaptive and locally informed methodologies in the future. | - |
| dc.description.sponsorship | null | - |
| dc.language.iso | id | - |
| dc.publisher | IPB University | id |
| dc.title | Koreksi Bias Data Curah Hujan CMIP6 pada Kawasan Danau Toba Menggunakan Quantile Delta Mapping | id |
| dc.title.alternative | Bias Correction of Rainfall Data from CMIP6 for the Lake Toba Region Using Quantile Delta Mapping | - |
| dc.type | Tesis | - |
| dc.subject.keyword | curah hujan | id |
| dc.subject.keyword | CMIP6 | id |
| dc.subject.keyword | Danau Toba | id |
| dc.subject.keyword | koreksi bias | id |
| dc.subject.keyword | quantile delta mapping | id |
| Appears in Collections: | MT - School of Data Science, Mathematic and Informatics | |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| cover_M0502241011_f5f9752c3fd14c77886fca97227237fc.pdf | Cover | 637.87 kB | Adobe PDF | View/Open |
| fulltext_M0502241011_c430c5343892467cb46e6a9a29daabbc.pdf Restricted Access | Fulltext | 3.49 MB | Adobe PDF | View/Open |
| lampiran_M0502241011_dbc203db3c074a8d8faefeee3f8a21cf.pdf Restricted Access | Lampiran | 652.18 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.