Modeling Daily Rainfall Occurrence and Amount Conditional on Atmospheric Predictors: Improved Assessment of Rainfall Statistics Based on Climate Model Results
Event details
| Date | 29.09.2015 |
| Hour | 12:00 › 13:15 |
| Speaker | Dr Andreas Langousis, Department of Civil Engineering, University of Patras, Greece |
| Location | |
| Category | Conferences - Seminars |
Abstract:
Due to its intermittent and highly variable character, and the modeling parameterizations used, precipitation is one of the least well reproduced hydrologic variables by both Global Climate Models (GCMs) and Regional Climate Models (RCMs). This is especially the case at a regional level (where hydrologic risks are assessed) and at small temporal scales (e.g. daily) used to run hydrologic models.
To improve the level skill of GCMs and RCMs in reproducing the statistics of rainfall at a basin level and at hydrologically relevant temporal scales, two types of statistical approaches have been suggested. One is the use of distribution mapping (i.e. quantile-quantile, Q-Q, plots) to statistically correct climate model rainfall outputs based on historical series of precipitation. The other is the use of stochastic models of rainfall to conditionally simulate precipitation series, based on large-scale atmospheric predictors produced by climate models (e.g. geopotential height, relative vorticity, divergence, mean sea level pressure). The latter approach, usually referred to as statistical rainfall downscaling, aims at reproducing the statistical character of rainfall, while accounting for the effects of large-scale atmospheric circulation (and, therefore, climate forcing) on rainfall statistics.
While promising, statistical rainfall downscaling has not attracted much attention in recent years, since the suggested approaches involved complex (i.e. subjective or computationally intense) identification procedures of the local weather, in addition to demonstrating limited success in reproducing several statistical features of rainfall, such as seasonal variations, the distributions of dry and wet spell lengths, the distribution of the mean rainfall intensity inside wet periods, and the distribution of rainfall extremes.
In an effort to remedy those shortcomings, Langousis and Kaleris (2014, WRR, 50, doi:10.1002/2013WR014936) developed a statistical framework for simulation of daily rainfall intensities conditional on upper air indices. Here, we test the developed downscaling scheme using atmospheric data from the ERA-Interim archive (http://www.ecmwf.int/research/era/do/get/index) and daily rainfall measurements from western Greece, and find that it accurately reproduces several statistical properties of actual rainfall records, at both annual and seasonal levels, including: wet day fractions, the alternation of wet and dry intervals, the distributions of dry and wet spell lengths, the distribution of rainfall intensities in wet days, the distribution of yearly rainfall maxima, dependencies of rainfall statistics on the observation scale, and long-term climatic features of rainfall. This is done solely by conditioning rainfall simulation on a vector of atmospheric predictors, properly selected to reflect the relative influence of upper-air variables on ground-level rainfall statistics.
In a follow up application study, we assess the relative effectiveness of: a) the developed statistical downscaling scheme, and b) quantile-quantile (Q-Q) correction of climate model rainfall products (i.e. an approach commonly used in climate change impact studies) in reproducing the statistical structure of rainfall, as well as rainfall extremes, at a regional level, based on climate model results. This is done for an intermediate-sized catchment in Italy, i.e. the Flumendosa catchment, using climate model rainfall and atmospheric data from the ENSEMBLES project (http://ensembleseu.metoffice.com). In doing so, we split the historical rainfall record of mean areal precipitation (MAP) in 15-year calibration and 45-year validation periods, and compare the historical rainfall statistics to those obtained from: a) Q-Q corrected climate model rainfall products, and b) synthetic rainfall series generated by the suggested downscaling scheme. To our knowledge, this is the first time that a detailed statistical comparison of climate model rainfall, statistically downscaled precipitation, and catchment averaged MAP is performed at a daily resolution.
The obtained results are promising, since the proposed downscaling scheme is more accurate and robust in reproducing a number of historical rainfall statistics, independent of the climate model used and the length of the calibration period. This is particularly the case for the yearly rainfall maxima, where direct statistical correction of climate model rainfall outputs shows increased sensitivity to the length of the calibration period and the climate model used. The robustness of the suggested downscaling scheme in modeling rainfall extremes at a daily resolution, is a notable feature that can effectively be used to assess hydrologic risk at a regional level under changing climatic conditions.
Short biography:
Dr. Andreas Langousis was born in Athens in 1981. He is a Civil Engineer (National Technical University of Athens, NTUA, 2003), with Master of Science (MSc, 2005) and Doctor of Science (ScD, 2008) degrees from the Department of Civil and Environmental Engineering at the Massachusetts Institute of Technology (MIT). Currently he serves as an Assistant Professor in the Department of Civil Engineering at the University of Patras, Greece, in the area of Stochastic Processes and Hydrologic Risk Assessment. He has received numerous academic awards, including the Schoettler Fellowship of MIT, a 6-year scholarship from the Alexander S. Onassis Public Benefit Foundation, the 1st prize at the 1st International Summit on Hurricanes and Climate Change, and a 3-year Postdoctoral Fellowship from the General Secretariat of Research and Technology (Greece). In addition, he has more than 10 years of teaching experience in undergraduate and graduate courses related to engineering hydrology and hydraulics, environmental data analysis, risk modeling, and stochastic processes. He has participated in 8 research projects and has co-authored 19 research papers in well known scientific journals, more than 30 presentations in international conferences, 3 book chapters, 4 newspaper articles, and he has served as an active member of the organizing and scientific committees of more than 15 international conferences. In addition, he has given 11 invited talks in Greece and abroad, he serves as Reviewer in 16 international scientific journals, as Guest Editor in JoH (Journal of Hydrology, Jun. 2015- Jan. 2016), and as Associate Editor in WRR (Water Resources Research) and SERRA (Stochastic Environmental Research and Risk Assessment). During the period Jun. 2012 - Dec. 2013, he also served as Guest Associate Editor in HESS (Hydrology and Earth System Sciences). He is an active member of the Technical Chamber of Greece (TCG), the American Society of Civil Engineers (ASCE), the American Geophysical Union (AGU), the European Geosciences Union (EGU), the International Association of Hydrological Sciences (IAHS), the International Commission on Statistical Hydrology (ICSH-IAHS) and, since 2008, he serves as an active member of the scientific committee of the Precipitation and Climate Sub-Division of the European Geosciences Union (EGU). During the period 2010-2012, he was an active member of the Board of Directors of the Hellenic Chapter of ASCE, for the period 2011-2013 he served as the Treasurer of the Alexander S. Onassis Scholar’s Association and, since March 2014, he serves as a member of the examination committee of the Technical Chamber of Greece (TCG) for the conferment of professional rights to Civil Engineers. Dr. Andreas Langousis’ area of expertise is the development of stochastic models for hydrologic risk analysis, engineering and environmental design and prediction. His current research interests include (but are not limited to) environmental and health risk, hydrologic scaling, estimation of hydrologic extremes, statistical downscaling and forecasting.
Due to its intermittent and highly variable character, and the modeling parameterizations used, precipitation is one of the least well reproduced hydrologic variables by both Global Climate Models (GCMs) and Regional Climate Models (RCMs). This is especially the case at a regional level (where hydrologic risks are assessed) and at small temporal scales (e.g. daily) used to run hydrologic models.
To improve the level skill of GCMs and RCMs in reproducing the statistics of rainfall at a basin level and at hydrologically relevant temporal scales, two types of statistical approaches have been suggested. One is the use of distribution mapping (i.e. quantile-quantile, Q-Q, plots) to statistically correct climate model rainfall outputs based on historical series of precipitation. The other is the use of stochastic models of rainfall to conditionally simulate precipitation series, based on large-scale atmospheric predictors produced by climate models (e.g. geopotential height, relative vorticity, divergence, mean sea level pressure). The latter approach, usually referred to as statistical rainfall downscaling, aims at reproducing the statistical character of rainfall, while accounting for the effects of large-scale atmospheric circulation (and, therefore, climate forcing) on rainfall statistics.
While promising, statistical rainfall downscaling has not attracted much attention in recent years, since the suggested approaches involved complex (i.e. subjective or computationally intense) identification procedures of the local weather, in addition to demonstrating limited success in reproducing several statistical features of rainfall, such as seasonal variations, the distributions of dry and wet spell lengths, the distribution of the mean rainfall intensity inside wet periods, and the distribution of rainfall extremes.
In an effort to remedy those shortcomings, Langousis and Kaleris (2014, WRR, 50, doi:10.1002/2013WR014936) developed a statistical framework for simulation of daily rainfall intensities conditional on upper air indices. Here, we test the developed downscaling scheme using atmospheric data from the ERA-Interim archive (http://www.ecmwf.int/research/era/do/get/index) and daily rainfall measurements from western Greece, and find that it accurately reproduces several statistical properties of actual rainfall records, at both annual and seasonal levels, including: wet day fractions, the alternation of wet and dry intervals, the distributions of dry and wet spell lengths, the distribution of rainfall intensities in wet days, the distribution of yearly rainfall maxima, dependencies of rainfall statistics on the observation scale, and long-term climatic features of rainfall. This is done solely by conditioning rainfall simulation on a vector of atmospheric predictors, properly selected to reflect the relative influence of upper-air variables on ground-level rainfall statistics.
In a follow up application study, we assess the relative effectiveness of: a) the developed statistical downscaling scheme, and b) quantile-quantile (Q-Q) correction of climate model rainfall products (i.e. an approach commonly used in climate change impact studies) in reproducing the statistical structure of rainfall, as well as rainfall extremes, at a regional level, based on climate model results. This is done for an intermediate-sized catchment in Italy, i.e. the Flumendosa catchment, using climate model rainfall and atmospheric data from the ENSEMBLES project (http://ensembleseu.metoffice.com). In doing so, we split the historical rainfall record of mean areal precipitation (MAP) in 15-year calibration and 45-year validation periods, and compare the historical rainfall statistics to those obtained from: a) Q-Q corrected climate model rainfall products, and b) synthetic rainfall series generated by the suggested downscaling scheme. To our knowledge, this is the first time that a detailed statistical comparison of climate model rainfall, statistically downscaled precipitation, and catchment averaged MAP is performed at a daily resolution.
The obtained results are promising, since the proposed downscaling scheme is more accurate and robust in reproducing a number of historical rainfall statistics, independent of the climate model used and the length of the calibration period. This is particularly the case for the yearly rainfall maxima, where direct statistical correction of climate model rainfall outputs shows increased sensitivity to the length of the calibration period and the climate model used. The robustness of the suggested downscaling scheme in modeling rainfall extremes at a daily resolution, is a notable feature that can effectively be used to assess hydrologic risk at a regional level under changing climatic conditions.
Short biography:
Dr. Andreas Langousis was born in Athens in 1981. He is a Civil Engineer (National Technical University of Athens, NTUA, 2003), with Master of Science (MSc, 2005) and Doctor of Science (ScD, 2008) degrees from the Department of Civil and Environmental Engineering at the Massachusetts Institute of Technology (MIT). Currently he serves as an Assistant Professor in the Department of Civil Engineering at the University of Patras, Greece, in the area of Stochastic Processes and Hydrologic Risk Assessment. He has received numerous academic awards, including the Schoettler Fellowship of MIT, a 6-year scholarship from the Alexander S. Onassis Public Benefit Foundation, the 1st prize at the 1st International Summit on Hurricanes and Climate Change, and a 3-year Postdoctoral Fellowship from the General Secretariat of Research and Technology (Greece). In addition, he has more than 10 years of teaching experience in undergraduate and graduate courses related to engineering hydrology and hydraulics, environmental data analysis, risk modeling, and stochastic processes. He has participated in 8 research projects and has co-authored 19 research papers in well known scientific journals, more than 30 presentations in international conferences, 3 book chapters, 4 newspaper articles, and he has served as an active member of the organizing and scientific committees of more than 15 international conferences. In addition, he has given 11 invited talks in Greece and abroad, he serves as Reviewer in 16 international scientific journals, as Guest Editor in JoH (Journal of Hydrology, Jun. 2015- Jan. 2016), and as Associate Editor in WRR (Water Resources Research) and SERRA (Stochastic Environmental Research and Risk Assessment). During the period Jun. 2012 - Dec. 2013, he also served as Guest Associate Editor in HESS (Hydrology and Earth System Sciences). He is an active member of the Technical Chamber of Greece (TCG), the American Society of Civil Engineers (ASCE), the American Geophysical Union (AGU), the European Geosciences Union (EGU), the International Association of Hydrological Sciences (IAHS), the International Commission on Statistical Hydrology (ICSH-IAHS) and, since 2008, he serves as an active member of the scientific committee of the Precipitation and Climate Sub-Division of the European Geosciences Union (EGU). During the period 2010-2012, he was an active member of the Board of Directors of the Hellenic Chapter of ASCE, for the period 2011-2013 he served as the Treasurer of the Alexander S. Onassis Scholar’s Association and, since March 2014, he serves as a member of the examination committee of the Technical Chamber of Greece (TCG) for the conferment of professional rights to Civil Engineers. Dr. Andreas Langousis’ area of expertise is the development of stochastic models for hydrologic risk analysis, engineering and environmental design and prediction. His current research interests include (but are not limited to) environmental and health risk, hydrologic scaling, estimation of hydrologic extremes, statistical downscaling and forecasting.
Practical information
- General public
- Free
- This event is internal
Organizer
- EESS - IIE
Contact
- Prof. Alexis Berne, LTE