03942nas a2200373 a 4500 20260817161747.0 260114s20259999th u ms t 000 eng d .b12470545 AIT Thesis no.WM-25-10 Pandey, Bikram Advancing flood forecasting for Northern Thailand using RRI model and data assimilation Pathum Thani, Thailand : Asian Institute of Technology, 2025 148 leaves : ill.+ 1 online Thesis; no. WM-25-10 A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Water Engineering and Management Thesis (M. Eng.) - Asian Institute of Technology, 2025 Flooding in Northern Thailand{u2019}s Nan River Basin poses significant risks during the monsoon season due to its mountainous terrain and exposure to the monsoon trough. This study applies the physically based Rainfall Runoff Inundation (RRI) model and Particle Filter Data Assimilation (PF-DA) technique to enhance flood forecasting in the Upper Nan River Basin. The study begins by calibrating and validating the RRI model using interpolated HII rain gauge data and observed discharge at stations N1 and N64. The model performed reliably in both the 2024 calibration and the 2022 validation period, producing realistic runoff responses with positive efficiency scores, low RMSE, and acceptable bias. A grid-wise, monthly Quantile Mapping bias correction was applied to the 0.25° daily GFS rainfall to reduce systematic biases. It effectively lowered the raw GFS overestimation and improved simulated discharge, but also dampened extremes, causing underestimated flood peaks.This study applied a daily RRI-Particle Filter data assimilation (PF-DA) system using 16 particles and Gaussian system noise, with rainfall boundary coordinates as state variables and an RMSE-based likelihood. Applying PF-DA during the 2022 and 2024 monsoon seasons yielded consistent gains across a 1-10 day lead time. Daily assimilation of observed water levels refined the model states, reduced false peaks from raw GFS, and softened the strong underestimation produced by bias-corrected GFS, improving hydrograph timing and stabilizing both flooding and recession periods. Short lead gains were notable for raw GFS, where KGE exceeded 0.5 for lead days 1-3 instead of only 1-2 without assimilation. Overall improvements were substantial, as KGE rose from 0.392 to 0.494 (~ 11%) and bias dropped from +0.543 m to +0.129 m in 2022. In 2024, KGE increased from 0.331 to 0.421 (~28%) with bias reduced from +0.568 m to +0.218 m. While PF-DA significantly improved short and medium-range forecasts, long-lead performance still depends on better rainfall forcing and maintaining sufficient ensemble spread. While QM improves overall stability, it also reduces extreme rainfall, leading to underestimated peak flows, showing the need for improved bias correction methods that better preserve extremes. Flood forecasting Thailand, Northern Rain and rainfall Mathematical models Thailand, Northern Natthachet Tangdamrongsub, Chairperson Shrestha, Sangam, Examination Committee Shanmugam, Mohana Sundaram, Examination Committee SET DEAN Scholarships, Scholarship Donor AIT Scholarship, Scholarship Donor Asian Institute of Technology. Thesis; no. WM-25-10 Full-Text http://203.159.5.9/ait-thesis/detail.php?q=B23299 .b12470545 mnarc a 260115 0 260115 m h a 0 mnarc 67 Barcode : - CREATED : 2026-01-14 RECORD # : i13566465 LPATRON : 0 LCHKIN : - # RENEWALS : 0 # OVERDUE : 0 IUSE3 : 0 TOT CHKOUT : 0 TOT RENEW : 0 2561 2561 0 0 0 0 MAIN MAIN mnarc 2026-08-17 0 AIT Thesis no.WM-25-10 2026-08-17 16:17:47 1 2026-08-17 67