Institutional-Repository, University of Moratuwa
Welcome to the University of Moratuwa Digital Repository, which houses postgraduate theses and dissertations, research articles presented at conferences by faculties and departments, university-published journal articles and research publications authored by academic staff. This online repository stores, preserves and distributes the University's scholarly work. This service allows University members to share their research with a larger audience.
 Research Publications |  Thesis & Dissertation |  E- Books |
 UoM Journal Publications |  UoM Conference Proceedings |  Articles published in Scimago's Q1 journals |  UoM Research Reports |  Other Articles authored by UoM staff |
Recent Submissions
item: Conference-Abstract
Development of a harmonized long-term global river discharge database for basin water resources management
(2026) Senavirathne, A; Leh, M; Matheswaran, K; Prabhath, PT; Velpuri, NM; Seid, A; Rajapakse, RLHL; Prasath, RA
Reliable long-term river discharge observations are fundamental for understanding surface-water variability and supporting effective river-basin management, yet global streamflow records remain fragmented across national and international repositories. This study developed a global daily river discharge database covering the period 1800–2026. Data was harmonized using PostgreSQL for systematic data integration, preprocessing, and quality control. Station metadata and daily discharge observations were compiled from multiple hydrological databases through application programming interfaces, direct downloads, and requestbased access. Spatial preprocessing removed stations with missing, invalid, or geographically erroneous coordinates. Stations sharing identical coordinates were identified as potential duplicates and evaluated using station names, record start and end dates, temporal overlap, common observation dates, and ischargeseries similarity. The preprocessing workflow generated 20,209 pairwise comparisons, including 908 pairs with accessible discharge observations from both sources. Duplicate, continuation, and overlapping records were reconciled by retaining the most complete record or merging complementary observations after removing repeated dates. Overall, 18,393 source records were consolidated during harmonization. The final database contains 65,804 unique physical stations, including 50,806 stations (77.2%) with daily discharge observations and 14,998 metadata only stations (22.8%). The harmonized database provides a consistent foundation for global and basin scale assessments of streamflow, water availability, hydrological variability, flood and drought conditions, and long-term changes monitoring, supporting evidence-based water resource planning and sustainable river basin management.
item: Conference-Abstract
Data quality as part of the model performance: benchmarking missing-value imputation for tropical streamflow prediction
(2026) Aqthaf, AHZ; Rajapakse, RLHL; Rajapakse, RLHL; Prasath, RA
Globally, reliable hydrological prediction depends not only on model architecture but also on how missing hydro-meteorological observations are reconstructed before model training. This issue is critical in datascarce tropical basins, where missing rainfall, temperature, evaporation, and streamflow observations are common, and imputation methods are often selected by convention rather than by evidence. This research benchmarks twelve missing-value imputation methods across four Sri Lankan river basins, four variable classes, and five synthetic missingness levels (5–40%) generated as short (1–3-day) and occasional longer (20–30-day) dropouts mimicking sensor failure, yielding 960 completed method–basin–variable– missingness evaluations. The methods span linear interpolation, Kalman filtering, k-nearest neighbours, multiple imputation by chained equations, missForest, Gaussian-copula and gradient-boosted approaches, together with self-attention and bidirectional recurrent deep imputers. A Friedman test demonstrates significant performance differences among methods (p<0.001), with SAITS and BRITS achieving the strongest ranks. However, linear interpolation remains statistically competitive with several established methods, matches SAITS for streamflow reconstruction, and requires approximately 18,000 times less computation. Performance depends strongly on variable and basin characteristics; notably, an STL-plusspline approach fails severely for bursty tropical rainfall. When reconstructed inputs are propagated through a downstream XGBoost streamflow model, the mean spread in validation Nash–Sutcliffe efficiency across imputation methods is only 0.091, while 74.5% of all method–basin–missingness combinations exceed NSE values of 0.80. These results show that imputation accuracy does not translate into proportionate gains in hydrological prediction. The study therefore proposes a performance- and application-based decision framework in which imputation sophistication is matched to downstream sensitivity, data characteristics, and computational constraints.
item: Conference-Abstract
Performance evaluation of global ensemble forecast system (GEFS) based reservoir inflow forecasting for almatti reservoir, India
(2026) Saravanan, S; Arivoli, E; Joshi, S; Chandrasekar, K; Rajapakse, RLHL; Prasath, RA
Accurate reservoir inflow forecasting is essential for effective reservoir operation and water resources management. This study evaluates the performance of the Global Ensemble Forecast System (GEFS) coupled with the Variable Infiltration Capacity (VIC) hydrological model for forecasting inflows to the Almatti Reservoir during a selected high-flow event from 15 to 30 June 2021. During the event, the observed reservoir inflow peaked at 4,390 m3/s. Three inflow hydrographs were considered: observed reservoir inflow, VIC-simulated inflow using observed rainfall from the India Meteorological Department (IMD), and forecast inflow generated using GEFS rainfall. The GEFS–VIC forecasts were evaluated at a 3-day forecast lead time by comparing the forecast and observed inflows. Forecast deviation was quantified as the percentage difference between GEFS–VIC forecast and observed inflows, with positive and negative values indicating overestimation and underestimation, respectively. The results show substantial deviation during the initial forecast period, with an average deviation of about 243%. Forecast performance improved during subsequent periods, with average deviations reducing to −18% and −32%. The forecast errors also varied with the stage of the observed inflow hydrograph. Overall, the GEFS–VIC forecasts captured the general temporal pattern, including the rapid rising and subsequent recession phases, although underestimation was evident during several later forecast periods. Comparison with VIC simulations driven by observed IMD rainfall provides a reference for understanding the influence of rainfall forecast uncertainty. The results demonstrate the potential of the GEFS–VIC framework for short-range operational reservoir inflow forecasting and supporting reservoir management decisions.
item: Conference-Abstract
Development of a national-scale SMHI HYPE model for water resources assessment in Sri Lanka under data-scarce conditions
(2026) Gangabadage, K; Amaguchi, H; Imamura, Y; Rajapakse, RLHL; Prasath, RA
Sri Lanka exhibits pronounced spatial and temporal variability in water availability, resulting in contrasting hydrological risks ranging from recurrent floods in wet regions to water scarcity in seasonally dry regions. National-scale hydrological assessment is therefore important for integrated water resources management; however, its application remain constrained by the limited spatial coverage, accessibility, and continuity of in-situ hydro-meteorological observations. Although globally available Earth observation and reanalysis datasets offer a potential alternative, their suitability for hydrological modelling under Sri Lanka’s contrasting hydroclimatic conditions require systematic evaluation. This study develops a national-scale semi-distributed hydrological modelling framework for Sri Lanka using the process-based SMHI HYPE model, integrating publicly available JAXA GSMaP precipitation, ERA5 reanalysis temperature, and SMAP L4 root-zone soil moisture, with in-situ observed streamflow. A pilot application is conducted in the upstream Kelani River Basin, selected for its relatively greater availability of hydrological observations, to evaluate the modelling framework prior to its broader national-scale application. Model calibration incorporates streamflow and root-zone soil moisture through a multi-variable calibration approach using Monte Carlo simulations, while multi-source validation considers streamflow, soil moisture, and reservoir storage variations. Initial pilot basin model application identified the need for further evaluation and local bias correction of public hydro-meteorological datasets before robust model calibration and national-scale implementation. These results guide the ongoing refinement of the modelling framework towards its application in data-scarce and poorly gauged regions of Sri Lanka.
item: Conference-Abstract
Development of a modified SCS curve number method to estimate runoff in small catchments in Sri Lanka
(2026) Manjitha, HHU; Perera, GMCA; Gunawardhana, HGLN; Rajapakse, RLHL; Prasath, RA
The conventional SCS Curve Number (SCS-CN) method is the most widely used approach for runoff estimation in ungauged catchments, but its static, table-based parameterization poorly represents the dynamic runoff response of tropical catchments. This study develops a modified SCS-CN method for 10+ small catchments in the wet, intermediate, and dry zones of Sri Lanka by integrating causal discovery, structural equation modelling (SEM), physics-informed neural networks (PINNs), transfer learning, and uncertainty quantification within a single framework. Causal discovery first identifies the rainfall, soil, landuse, and catchment characteristics that drive runoff generation, and SEM quantifies their direct and combined effects, replacing the static tabulated curve number with a dynamic, catchment-specific CN function. This dynamic CN is then embedded as a physical constraint in a PINN trained on observed rainfallrunoff data, while transfer learning extends the calibrated framework to data-scarce catchments and uncertainty quantification provides prediction intervals for each runoff estimate. The framework is evaluated through split-sample testing and benchmarked against the standard SCS-CN method and conventional datadriven models using Nash-Sutcliffe efficiency (NSE), percent bias (PBIAS), and root-mean-square error (RMSE). By combining physically interpretable structure with data-driven adaptability, the modified method is designed to reduce the parameter uncertainty of the conventional approach and to provide reliability-quantified runoff estimates for water resources planning, flood design, and agricultural water management in ungauged tropical catchments.