Graduate Studies, UNL

 

Dissertations and Doctoral Documents from University of Nebraska-Lincoln, 2023–

First Advisor

Ayse Kılıç

Degree Name

Doctor of Philosophy (Ph.D.)

Committee Members

David Admiraal, Francisco Munoz-Arriola, Yusong Li

Department

Civil Engineering

Date of this Version

2025

Document Type

Dissertation

Citation

A dissertation presented to the faculty of the Graduate College of the University of Nebraska in partial fulfillment of requirements for the degree Doctor of Philosophy (Ph.D.)

Major: Civil Engineering

Under the supervision of Professor

Lincoln, Nebraska, December 2025

Comments

Copyright 2025, the author. Used by permission

Abstract

This dissertation explores three rainfall-runoff models in the humid tropical regions of Indonesia using satellite-based precipitation products (SBPPs) and develops integrated machine-learning modeling frameworks. Several ground-based observations from BMKG (Badan Meteorologi, Klimatologi, dan Geofisika (also known as the Indonesian Agency for Meteorology, Climatology, and Geophysics)) stations across Indonesia (133−165 stations) are compared and evaluated against satellite products, indicating that GPM performs well, with R-squared values ranging from 0.54 to 0.76 and correlation coefficients ranging from 0.45 to 0.69, respectively.

In the Martapura Watershed, South Kalimantan, due to a lack of observational discharge data, streamflow was generated using the FJ Mock and the NRECA methods, which successfully estimated monthly streamflow from satellite precipitation data. Further, a complete eligibility test was conducted for 56 GPM grid cells within the watershed, identifying 14 reliable grids for hydrological analysis. Rainfall-runoff models, namely HEC-HMS, HEC-RAS, and the Rainfall-Runoff-Inundation (RRI) model, were employed to simulate four major flood events during (2014−2023), with the RRI model showing the best performance in capturing expected flood hydrograph characteristics.

Enhancing the accuracy of streamflow generation, a machine learning approach using Support Vector Machine (SVM) regression, applying eight different scenarios. It was acknowledged that Scenario 6, combining the NRECA-generated streamflow with soil moisture and uncorrected GPM precipitation, achieved optimal performance (highest R2 and NSE, and lowest RMSE). This research provides a representative model for satellite-based flood modeling in ungauged tropical basins, particularly by improving early warning systems and water resources management in developing countries with limited data.

Advisor: Ayse Kılıç

Download

Share

COinS