Nadee Peiris
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Nadee Peiris has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Chamal Perera
University of Moratuwa (Moratuwa, Sri Lanka)
4 shared publications - Lalith Rajapakse
UNESCO-Madanjeet Singh Centre for South Asia Water Management (UMCSAWM), Department of Civil Engineering, University of Moratuwa, Moratuwa 10400, Sri Lanka
4 shared publications Ajith Wijemannage
Department of Meteorology, Bauddhaloka Mawatha, Colombo-07, Sri Lanka
3 shared publicationsNimal Wijayaratna
UNESCO-Madanjeet Singh Centre for South Asia Water Management (UMCSAWM), University of Moratuwa, Moratuwa 10400, Sri Lanka
3 shared publications- Binal Chatura Dissanayake
Department of Civil Engineering, Faculty of Engineering, University of Moratuwa, Sri Lanka
1 shared publication - Kasun De Silva
Department of Civil Engineering, University of Moratuwa, Moratuwa, Sri Lanka
1 shared publication
Latest Publications
- Open Access
- 15 April 2026
- Published by Copernicus GmbH
- Vol. 2026, 1-37
Comment on egusphere-2025-6551
Ground-based rain gauges remain the benchmark for accurate precipitation measurement; however, their sparse spatial distribution limits the representation of rainfall heterogeneity. Satellite-based Precipitation Products (SPPs) provide consistent spatial coverage but are often affected by retrieval errors and regional biases, restricting their direct use in local-scale hydrological applications. To overcome these limitations, Precipitation Data Merging (PDM) techniques integrating gauge and satellite observations have gained prominence. This study introduces a novel Machine Learning (ML) framework, GraphIDW, which combines Graph Neural Networks (GNNs) with Inverse Distance Weighting (IDW) interpolation to explicitly incorporate spatial autocorrelation into the merging process, addressing a major limitation of traditional ML-based PDM approaches. The framework was evaluated across the Wet Zone of Sri Lanka from 2001 to 2015 using two state-of-the-art SPPs (IMERG and CHIRPS) together with ground observations. IMERG data (0.1°) were first downscaled to 0.05° using CHIRPS, after which the downscaled product was merged with gauge observations through GraphIDW. A total of 60 gauges (70 %) were used for training and 28 (30 %) for validation. Results show that GraphIDW outperforms conventional ML algorithms, including Random Forest, Artificial Neural Network, Support Vector Regression, and XGBoost. It achieved the highest probability of detection (0.97) and reduced root mean square error (RMSE) and mean absolute error (MAE) by 13 %–41 % and 9 %–36 %, respectively, compared with the original SPPs. The results demonstrate that explicitly accounting for spatial dependence through graph-based learning significantly improves precipitation estimation, particularly in regions characterized by strong spatial heterogeneity. By embedding spatial autocorrelation directly into the merging process, GraphIDW provides a robust and computationally efficient framework for generating high-resolution rainfall datasets that are better suited for hydrological analysis in complex climatic and topographic settings.
- 15 April 2026
- Published by Copernicus GmbH
- Vol. 2026, 1-37
Open AccessComment on egusphere-2025-6551
Ground-based rain gauges remain the benchmark for accurate precipitation measurement; however, their sparse spatial distribution limits the representation of rainfall heterogeneity. Satellite-based Precipitation Products (SPPs) provide consistent spatial coverage but are often affected by retrieval errors and regional biases, restricting their direct use in local-scale hydrological applications. To overcome these limitations, Precipitation Data Merging (PDM) techniques integrating gauge and satellite observations have gained prominence. This study introduces a novel Machine Learning (ML) framework, GraphIDW, which combines Graph Neural Networks (GNNs) with Inverse Distance Weighting (IDW) interpolation to explicitly incorporate spatial autocorrelation into the merging process, addressing a major limitation of traditional ML-based PDM approaches. The framework was evaluated across the Wet Zone of Sri Lanka from 2001 to 2015 using two state-of-the-art SPPs (IMERG and CHIRPS) together with ground observations. IMERG data (0.1°) were first downscaled to 0.05° using CHIRPS, after which the downscaled product was merged with gauge observations through GraphIDW. A total of 60 gauges (70 %) were used for training and 28 (30 %) for validation. Results show that GraphIDW outperforms conventional ML algorithms, including Random Forest, Artificial Neural Network, Support Vector Regression, and XGBoost. It achieved the highest probability of detection (0.97) and reduced root mean square error (RMSE) and mean absolute error (MAE) by 13 %–41 % and 9 %–36 %, respectively, compared with the original SPPs. The results demonstrate that explicitly accounting for spatial dependence through graph-based learning significantly improves precipitation estimation, particularly in regions characterized by strong spatial heterogeneity. By embedding spatial autocorrelation directly into the merging process, GraphIDW provides a robust and computationally efficient framework for generating high-resolution rainfall datasets that are better suited for hydrological analysis in complex climatic and topographic settings.
- Open Access
- 11 March 2026
- Published by Copernicus GmbH
- Vol. 2026, 1-37
Comment on egusphere-2025-6551
Ground-based rain gauges remain the benchmark for accurate precipitation measurement; however, their sparse spatial distribution limits the representation of rainfall heterogeneity. Satellite-based Precipitation Products (SPPs) provide consistent spatial coverage but are often affected by retrieval errors and regional biases, restricting their direct use in local-scale hydrological applications. To overcome these limitations, Precipitation Data Merging (PDM) techniques integrating gauge and satellite observations have gained prominence. This study introduces a novel Machine Learning (ML) framework, GraphIDW, which combines Graph Neural Networks (GNNs) with Inverse Distance Weighting (IDW) interpolation to explicitly incorporate spatial autocorrelation into the merging process, addressing a major limitation of traditional ML-based PDM approaches. The framework was evaluated across the Wet Zone of Sri Lanka from 2001 to 2015 using two state-of-the-art SPPs (IMERG and CHIRPS) together with ground observations. IMERG data (0.1°) were first downscaled to 0.05° using CHIRPS, after which the downscaled product was merged with gauge observations through GraphIDW. A total of 60 gauges (70 %) were used for training and 28 (30 %) for validation. Results show that GraphIDW outperforms conventional ML algorithms, including Random Forest, Artificial Neural Network, Support Vector Regression, and XGBoost. It achieved the highest probability of detection (0.97) and reduced root mean square error (RMSE) and mean absolute error (MAE) by 13 %–41 % and 9 %–36 %, respectively, compared with the original SPPs. The results demonstrate that explicitly accounting for spatial dependence through graph-based learning significantly improves precipitation estimation, particularly in regions characterized by strong spatial heterogeneity. By embedding spatial autocorrelation directly into the merging process, GraphIDW provides a robust and computationally efficient framework for generating high-resolution rainfall datasets that are better suited for hydrological analysis in complex climatic and topographic settings.
- 11 March 2026
- Published by Copernicus GmbH
- Vol. 2026, 1-37
Open AccessComment on egusphere-2025-6551
Ground-based rain gauges remain the benchmark for accurate precipitation measurement; however, their sparse spatial distribution limits the representation of rainfall heterogeneity. Satellite-based Precipitation Products (SPPs) provide consistent spatial coverage but are often affected by retrieval errors and regional biases, restricting their direct use in local-scale hydrological applications. To overcome these limitations, Precipitation Data Merging (PDM) techniques integrating gauge and satellite observations have gained prominence. This study introduces a novel Machine Learning (ML) framework, GraphIDW, which combines Graph Neural Networks (GNNs) with Inverse Distance Weighting (IDW) interpolation to explicitly incorporate spatial autocorrelation into the merging process, addressing a major limitation of traditional ML-based PDM approaches. The framework was evaluated across the Wet Zone of Sri Lanka from 2001 to 2015 using two state-of-the-art SPPs (IMERG and CHIRPS) together with ground observations. IMERG data (0.1°) were first downscaled to 0.05° using CHIRPS, after which the downscaled product was merged with gauge observations through GraphIDW. A total of 60 gauges (70 %) were used for training and 28 (30 %) for validation. Results show that GraphIDW outperforms conventional ML algorithms, including Random Forest, Artificial Neural Network, Support Vector Regression, and XGBoost. It achieved the highest probability of detection (0.97) and reduced root mean square error (RMSE) and mean absolute error (MAE) by 13 %–41 % and 9 %–36 %, respectively, compared with the original SPPs. The results demonstrate that explicitly accounting for spatial dependence through graph-based learning significantly improves precipitation estimation, particularly in regions characterized by strong spatial heterogeneity. By embedding spatial autocorrelation directly into the merging process, GraphIDW provides a robust and computationally efficient framework for generating high-resolution rainfall datasets that are better suited for hydrological analysis in complex climatic and topographic settings.
- Open Access
- 13 February 2026
- Published by Copernicus GmbH in EGUsphere
- Vol. 2026, 1-37
GraphIDW: Incorporating spatial autocorrelation in satellite–gauge precipitation merging using graph neural networks over a tropical region
Ground-based rain gauges remain the benchmark for accurate precipitation measurement; however, their sparse spatial distribution limits the representation of rainfall heterogeneity. Satellite-based Precipitation Products (SPPs) provide consistent spatial coverage but are often affected by retrieval errors and regional biases, restricting their direct use in local-scale hydrological applications. To overcome these limitations, Precipitation Data Merging (PDM) techniques integrating gauge and satellite observations have gained prominence. This study introduces a novel Machine Learning (ML) framework, GraphIDW, which combines Graph Neural Networks (GNNs) with Inverse Distance Weighting (IDW) interpolation to explicitly incorporate spatial autocorrelation into the merging process, addressing a major limitation of traditional ML-based PDM approaches. The framework was evaluated across the Wet Zone of Sri Lanka from 2001 to 2015 using two state-of-the-art SPPs (IMERG and CHIRPS) together with ground observations. IMERG data (0.1°) were first downscaled to 0.05° using CHIRPS, after which the downscaled product was merged with gauge observations through GraphIDW. A total of 60 gauges (70 %) were used for training and 28 (30 %) for validation. Results show that GraphIDW outperforms conventional ML algorithms, including Random Forest, Artificial Neural Network, Support Vector Regression, and XGBoost. It achieved the highest probability of detection (0.97) and reduced root mean square error (RMSE) and mean absolute error (MAE) by 13 %–41 % and 9 %–36 %, respectively, compared with the original SPPs. The results demonstrate that explicitly accounting for spatial dependence through graph-based learning significantly improves precipitation estimation, particularly in regions characterized by strong spatial heterogeneity. By embedding spatial autocorrelation directly into the merging process, GraphIDW provides a robust and computationally efficient framework for generating high-resolution rainfall datasets that are better suited for hydrological analysis in complex climatic and topographic settings.
- 13 February 2026
- Published by Copernicus GmbH in EGUsphere
- Vol. 2026, 1-37
Open AccessGraphIDW: Incorporating spatial autocorrelation in satellite–gauge precipitation merging using graph neural networks over a tropical region
Ground-based rain gauges remain the benchmark for accurate precipitation measurement; however, their sparse spatial distribution limits the representation of rainfall heterogeneity. Satellite-based Precipitation Products (SPPs) provide consistent spatial coverage but are often affected by retrieval errors and regional biases, restricting their direct use in local-scale hydrological applications. To overcome these limitations, Precipitation Data Merging (PDM) techniques integrating gauge and satellite observations have gained prominence. This study introduces a novel Machine Learning (ML) framework, GraphIDW, which combines Graph Neural Networks (GNNs) with Inverse Distance Weighting (IDW) interpolation to explicitly incorporate spatial autocorrelation into the merging process, addressing a major limitation of traditional ML-based PDM approaches. The framework was evaluated across the Wet Zone of Sri Lanka from 2001 to 2015 using two state-of-the-art SPPs (IMERG and CHIRPS) together with ground observations. IMERG data (0.1°) were first downscaled to 0.05° using CHIRPS, after which the downscaled product was merged with gauge observations through GraphIDW. A total of 60 gauges (70 %) were used for training and 28 (30 %) for validation. Results show that GraphIDW outperforms conventional ML algorithms, including Random Forest, Artificial Neural Network, Support Vector Regression, and XGBoost. It achieved the highest probability of detection (0.97) and reduced root mean square error (RMSE) and mean absolute error (MAE) by 13 %–41 % and 9 %–36 %, respectively, compared with the original SPPs. The results demonstrate that explicitly accounting for spatial dependence through graph-based learning significantly improves precipitation estimation, particularly in regions characterized by strong spatial heterogeneity. By embedding spatial autocorrelation directly into the merging process, GraphIDW provides a robust and computationally efficient framework for generating high-resolution rainfall datasets that are better suited for hydrological analysis in complex climatic and topographic settings.
- Open Access
Designing for the Past in a Nonstationary Climate: Evidence from Cyclone Ditwah’s Extreme Rainfall in Sri Lanka
The November 2025 extreme rainfall event associated with Tropical Cyclone Ditwah caused catastrophic flooding and landslides across Sri Lanka. This study presents a national-scale statistical and Intensity–Duration–Frequency (IDF)-based assessment of the event using long-term rain gauge observations, extreme value analysis, and climate scenario-based projections. The 24-h rainfall data from 46 stations were analyzed for 1-, 2-, and 3-day durations. Historical annual maximum series were extracted and compared with the 2025 event to identify record-breaking extremes. Rainfall volumes were also estimated and compared with the island’s Average Annual Rainfall (AAR) and volumes from major flood events in 2010 and 2016. The November 2025 event exceeded historical maxima at 14 stations, with estimated return periods frequently surpassing 1000 years. The cumulative rainfall volume from 26–28 November accounted for 15.8% of Sri Lanka’s AAR. Updated IDF curves incorporating the event showed marked upward shifts, with intensities at some locations matching or exceeding projections under high-emission climate scenarios. The results highlight the inadequacy of existing design standards in capturing emerging extremes and the need for urgent updates to Sri Lanka’s national IDF relationships to support climate-resilient flood risk management and infrastructure planning.
Open AccessDesigning for the Past in a Nonstationary Climate: Evidence from Cyclone Ditwah’s Extreme Rainfall in Sri Lanka
The November 2025 extreme rainfall event associated with Tropical Cyclone Ditwah caused catastrophic flooding and landslides across Sri Lanka. This study presents a national-scale statistical and Intensity–Duration–Frequency (IDF)-based assessment of the event using long-term rain gauge observations, extreme value analysis, and climate scenario-based projections. The 24-h rainfall data from 46 stations were analyzed for 1-, 2-, and 3-day durations. Historical annual maximum series were extracted and compared with the 2025 event to identify record-breaking extremes. Rainfall volumes were also estimated and compared with the island’s Average Annual Rainfall (AAR) and volumes from major flood events in 2010 and 2016. The November 2025 event exceeded historical maxima at 14 stations, with estimated return periods frequently surpassing 1000 years. The cumulative rainfall volume from 26–28 November accounted for 15.8% of Sri Lanka’s AAR. Updated IDF curves incorporating the event showed marked upward shifts, with intensities at some locations matching or exceeding projections under high-emission climate scenarios. The results highlight the inadequacy of existing design standards in capturing emerging extremes and the need for urgent updates to Sri Lanka’s national IDF relationships to support climate-resilient flood risk management and infrastructure planning.