K. Georgakakos | University of California, San Diego (original) (raw)
Papers by K. Georgakakos
Water, 2022
This article is an open access article distributed under the terms and conditions of the Creative... more This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY
First-Year Progress Report. It also contains a description of the previous approaches for utilizi... more First-Year Progress Report. It also contains a description of the previous approaches for utilizing multiple sensors and their limitations, the proposed stochastic interpolation approaches, and the overall evaluation methodology which was implemented on this project.
Rainfall is often represented by a design storm with uniform intensity in urban hydrological mode... more Rainfall is often represented by a design storm with uniform intensity in urban hydrological models even though rainfall is a highly dynamic process across very small temporal and spatial scales. This study examines characteristics of highresolution radar data (5-minute temporal resolution, 1x1 km spatial resolution) over an area of 1824 km 2 covering the catchment of the river Wupper, North Rhine-Westphalia, Germany. Extreme events were sampled by a Peak Over Threshold method using several sampling strategies, all based on selecting an average of three events per year. A simple identificationand tracking algorithm for rain cells based on intensity threshold and fitting of ellipsoids, is developed for the study. Both hourly and daily extremes were analysed with respect to a set of 16 descriptive variables. The spatio-temporal properties of the extreme events are explored by means of principal component analysis, cluster analysis, and linear models for these 16 variables. The PCA indicated between 5 and 9 dimensions in the extreme event characteristic data. The cluster analyses identified four rainfall types: extreme convective, convective, convective events in front systems and front system events. The stepwise regression for each variable identified independent variables that correspond well with the correlation structure identified in the clusters. This indicates that the correlation structure may prove useful in setting up a weather generator. 1 Introduction Urban hydrological models of high quality are a required tool to make cities more resilient to pluvial flooding and pollution management. A key input parameter when modelling urban drainage systems is rainfall (Berndtsson and Niemczynowicz,
under a Creative Commons License. Observing extreme events in incomplete state spaces with applic... more under a Creative Commons License. Observing extreme events in incomplete state spaces with application to rainfall estimation from satellite images
Bulletin of the American Meteorological Society, 2021
ABSTRACT The occurrence of flash flooding is of concern in hydrologic and natural hazards science... more ABSTRACT The occurrence of flash flooding is of concern in hydrologic and natural hazards science due to the top ranking of such events among natural disasters in terms of both the number of people effected globally and the proportion of individual fatalities. A better understanding of such extreme event frequency over various time and spatial scales is sought in this research as an important aspect of climate and hydrologic science. The study characterizes flash flood occurrence frequency through interdisciplinary meteorological, hydrologic, and geomorphologic modeling in the Southern California region over the historical period from 1950 to 2005. A combined modeling approach is necessitated due to (a) infrequent reporting of flash flood occurrence with high spatial detail, and (b) the relative sparseness of observed precipitation records covering regions and the spatial scales of flash flood occurrence (e.g., typically over a few tens of km2) to derive climatology based on observations only. The modeling approach includes generation of historical orographically-driven precipitation over the region with 3km spatial resolution, hydrologic modeling of soil moisture, and estimation of flash flood occurrence via a hydrologic response threshold model for small-scale watershed on the order of 30km2. The precipitation modeling is driven by NCEP Reanalysis forcing, which subsequently provides input to the hydrologic and flash flood occurrence modeling. This research represents a regional look at the spatial variability in flash flood occurrence with relatively high spatial resolution. This paper will focus on a general description of the modeling methodology and on initial results of the spatial character of flash flood occurrence frequency over the mountain-to-foothill regions of the Transverse and Peninsular mountain ranges of Southern California.
ABSTRACT The proximity of the Pacific Ocean to the Transverse and Peninsular Mountain Ranges of c... more ABSTRACT The proximity of the Pacific Ocean to the Transverse and Peninsular Mountain Ranges of coastal Southern California may lead to significant, orographically-enhanced precipitation in the region. With abundant moisture, such as evidenced in Pineapple Express events or atmospheric rivers, this precipitation may lead to other hydrologic hazards as flash flooding, landslides or debris flows. Available precipitation observation networks are relatively sparse in the mountainous regions and often do not capture the spatial variation of these events with high resolution. This study aims to simulate the topographically-driven precipitation over Southern California with high spatial resolution using a simplified orographic precipitation model. The model employs potential theory flow to estimate steady state three-dimensional wind fields for given free stream velocity forcing winds, atmospheric moisture advection, and cloud and precipitation microphysics proposed by Kessler (1969). The advantage of this modeling set-up is the computational efficiency as compared to regional mesoscale models such as the MM5. For this application, the Southern California region, comprised of the counties of Santa Barbara, Ventura, Los Angeles, Orange, and San Diego, and portions of San Bernardino and Riverside counties, are modeled at a 3-km resolution. The orographic precipitation model is forced by free stream wind velocities given by the 700mb winds from the NCEP Reanalysis I dataset. Atmospheric moisture initial conditions are defined also by the NCEP Reanalysis I dataset, and updated 4x- daily with the available 6-hourly NCEP Reanalysis forcing. This paper presents a comparison of the simulated precipitation to observations for over a variety of spatial scales and over the historical wet season periods from October 2000 to April 2005. The comparison is made over several performance measurements including (a) the occurrence/non-occurrence of precipitation, (b) overall bias and correlation, (c) bias and correlation for precipitation exceeding given thresholds, and (d) the frequency distributions of non-zero precipitation. The results of simulation performance are compared to reported results of other orographically-driven precipitation and regional mesoscale model studies within the Western U.S.
A new methodology is presented for development of macroscale hydrologic model percolation paramet... more A new methodology is presented for development of macroscale hydrologic model percolation parameters from a spatial database of soil properties. This approach is applied to three distinct catchments within California- the Kings, American, and lower Eel river basins. Each unique vertical soil profile in these catchments is divided into an upper and a lower layer based on permeability gradients. A one-dimensional numerical unsaturated flow model is applied to each profile to yield percolation from upper to lower layers as a function of lower layer moisture deficit. A Holtan-type power law relationship is postulated to adequately represent profile percolation, with parameters estimated by curve-fitting the numerical model results. These parameters are then aggregated from the scale of the observable soil profiles to the level of the macroscale hydrologic model elements in a mass-conserving manner. In this process, the power law relationship is considered a describing function approximation to the numerical model response. To estimate parameter and flux uncertainties, a Monte Carlo approach is employed in which the soil property values are sampled randomly within the uncertainty ranges provided in the soils database or established by previous studies in the literature. The resulting ensembles of soil profiles are used in the describing function method to generate ensembles of hydrologic parameter spatial distributions for the catchments of interest. These ensembles are used as measures of hydrologic parameter uncertainty. The relationship between aggregate moisture flux and soil parameter distributions is explored, as is the dependence of mean parameter and aggregate flux values and their respective uncertainties on spatial scale. The methodology presented provides a useful means of investigating model-process observability given present-day soils databases, and a robust method for determining physically based macro-scale hydrologic parameters for subsequent application in hydroclimatic modeling.
Within the context of predicting and mitigating flood hazards, this paper motivates the developme... more Within the context of predicting and mitigating flood hazards, this paper motivates the development of innovative ideas in three engineering research areas: (a) spatially-distributed quantitative rainfall prediction, (b) spatially-distributed flow prediction, and (c) long-term flow prediction for water resources management. Design, operational implementation, real-time testing, and training of field personnel should be the cornerstone of such development for realizing the potential benefits in these fields.
Bulletin of the American Meteorological Society
At the beginning of the 21st Century a research-to-operations program was initiated to design and... more At the beginning of the 21st Century a research-to-operations program was initiated to design and develop operational systems to support local forecasters in their challenging task to provide advance warning for flash floods worldwide. Twenty some years later, the Flash Flood Guidance System with global coverage provides real-time assessment and guidance products to more than 60 countries, serving nearly 3 billion people. The implementation domains cover a wide range of hydroclimatological, geomorphological and land-use regimes worldwide. This flexible and evolving system combines meteorology and hydrology data and concepts as well as supports product utility for flash-flood disaster mitigation on very large scales with high spatial and temporal resolution. Through quality control procedures, it integrates remotely-sensed data of land-surface precipitation and of land-surface properties from geostationary and polar orbiter satellite platforms, reflectivity data from a variety of wea...
Bulletin of the American Meteorological Society
Journal of Hydroinformatics
Corporate technology transfer by US non-governmental organizations with the substantial involveme... more Corporate technology transfer by US non-governmental organizations with the substantial involvement of university faculty is a new activity in hydrometeorology. The issues involved in such US corporate technology transfers are discussed by way of two examples selected from the activities of the Hydrologic Research Center, a non-profit-making public-benefit research and technology transfer corporation in San Diego, California, USA. The projects discussed are: (a) the development and implementation of a robust state estimator for national use within the US National Weather Service River Forecast System, and (b) the development and implementation of a prototype multi-sensor rainfall forecasting system for the Panama Canal Authority. The issues covered include technical ones associated with improving theoretical formulations for robust operational performance, those associated with the necessary reciprocal education between modellers and field personnel, and the accommodation of the educational objectives of participating postdoctoral associates.
Open-File Report
An integrated forecast-control system was designed to allow the profitable use of ensemble foreca... more An integrated forecast-control system was designed to allow the profitable use of ensemble forecasts for the operational management of multipurpose reservoirs. The system ingests large-scale climate model monthly precipitation through the adjustment of the marginal distribution of reservoir-catchment precipitation to reflect occurrence of monthly climate precipitation amounts in the extreme terciles of their distribution. Generation of ensemble reservoir inflow forecasts is then accomplished with due account for atmospheric-forcing and hydrologic-model uncertainties. These ensemble forecasts are ingested by the decision component of the integrated system, which generates non-inferior trade-off surfaces and, given management preferences, estimates of reservoir-management benefits over given periods. In collaboration with the Bureau of Reclamation and the California Nevada River Forecast Center, the integrated system is applied to Folsom Lake in California to evaluate the benefits for flood control, hydroelectric energy production, and low flow augmentation. In addition to retrospective studies involving the historical period 1964-1993, system simulations were performed for the future period 2001-2030, under a control (constant future greenhouse-gas concentrations assumed at the present levels) and a greenhouse-gas-increase (1-% per annum increase assumed) scenario. The present paper presents and validates ensemble 30-day reservoir-inflow forecasts under a variety of situations. Corresponding reservoir management results are presented in Yao and Georgakakos, A., this issue. Principle conclusions of this paper are that the integrated system provides reliable ensemble inflow volume forecasts at the 5-% confidence level for the majority of the deciles of forecast frequency, and that the use of climate model simulations is beneficial mainly during high flow periods. It is also found that, for future periods with potential sharp climatic increases of precipitation amount and to maintain good reliability levels, operational ensemble inflow forecasting should involve atmospheric forcing from appropriate climatic periods.
Water, 2022
This article is an open access article distributed under the terms and conditions of the Creative... more This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY
First-Year Progress Report. It also contains a description of the previous approaches for utilizi... more First-Year Progress Report. It also contains a description of the previous approaches for utilizing multiple sensors and their limitations, the proposed stochastic interpolation approaches, and the overall evaluation methodology which was implemented on this project.
Rainfall is often represented by a design storm with uniform intensity in urban hydrological mode... more Rainfall is often represented by a design storm with uniform intensity in urban hydrological models even though rainfall is a highly dynamic process across very small temporal and spatial scales. This study examines characteristics of highresolution radar data (5-minute temporal resolution, 1x1 km spatial resolution) over an area of 1824 km 2 covering the catchment of the river Wupper, North Rhine-Westphalia, Germany. Extreme events were sampled by a Peak Over Threshold method using several sampling strategies, all based on selecting an average of three events per year. A simple identificationand tracking algorithm for rain cells based on intensity threshold and fitting of ellipsoids, is developed for the study. Both hourly and daily extremes were analysed with respect to a set of 16 descriptive variables. The spatio-temporal properties of the extreme events are explored by means of principal component analysis, cluster analysis, and linear models for these 16 variables. The PCA indicated between 5 and 9 dimensions in the extreme event characteristic data. The cluster analyses identified four rainfall types: extreme convective, convective, convective events in front systems and front system events. The stepwise regression for each variable identified independent variables that correspond well with the correlation structure identified in the clusters. This indicates that the correlation structure may prove useful in setting up a weather generator. 1 Introduction Urban hydrological models of high quality are a required tool to make cities more resilient to pluvial flooding and pollution management. A key input parameter when modelling urban drainage systems is rainfall (Berndtsson and Niemczynowicz,
under a Creative Commons License. Observing extreme events in incomplete state spaces with applic... more under a Creative Commons License. Observing extreme events in incomplete state spaces with application to rainfall estimation from satellite images
Bulletin of the American Meteorological Society, 2021
ABSTRACT The occurrence of flash flooding is of concern in hydrologic and natural hazards science... more ABSTRACT The occurrence of flash flooding is of concern in hydrologic and natural hazards science due to the top ranking of such events among natural disasters in terms of both the number of people effected globally and the proportion of individual fatalities. A better understanding of such extreme event frequency over various time and spatial scales is sought in this research as an important aspect of climate and hydrologic science. The study characterizes flash flood occurrence frequency through interdisciplinary meteorological, hydrologic, and geomorphologic modeling in the Southern California region over the historical period from 1950 to 2005. A combined modeling approach is necessitated due to (a) infrequent reporting of flash flood occurrence with high spatial detail, and (b) the relative sparseness of observed precipitation records covering regions and the spatial scales of flash flood occurrence (e.g., typically over a few tens of km2) to derive climatology based on observations only. The modeling approach includes generation of historical orographically-driven precipitation over the region with 3km spatial resolution, hydrologic modeling of soil moisture, and estimation of flash flood occurrence via a hydrologic response threshold model for small-scale watershed on the order of 30km2. The precipitation modeling is driven by NCEP Reanalysis forcing, which subsequently provides input to the hydrologic and flash flood occurrence modeling. This research represents a regional look at the spatial variability in flash flood occurrence with relatively high spatial resolution. This paper will focus on a general description of the modeling methodology and on initial results of the spatial character of flash flood occurrence frequency over the mountain-to-foothill regions of the Transverse and Peninsular mountain ranges of Southern California.
ABSTRACT The proximity of the Pacific Ocean to the Transverse and Peninsular Mountain Ranges of c... more ABSTRACT The proximity of the Pacific Ocean to the Transverse and Peninsular Mountain Ranges of coastal Southern California may lead to significant, orographically-enhanced precipitation in the region. With abundant moisture, such as evidenced in Pineapple Express events or atmospheric rivers, this precipitation may lead to other hydrologic hazards as flash flooding, landslides or debris flows. Available precipitation observation networks are relatively sparse in the mountainous regions and often do not capture the spatial variation of these events with high resolution. This study aims to simulate the topographically-driven precipitation over Southern California with high spatial resolution using a simplified orographic precipitation model. The model employs potential theory flow to estimate steady state three-dimensional wind fields for given free stream velocity forcing winds, atmospheric moisture advection, and cloud and precipitation microphysics proposed by Kessler (1969). The advantage of this modeling set-up is the computational efficiency as compared to regional mesoscale models such as the MM5. For this application, the Southern California region, comprised of the counties of Santa Barbara, Ventura, Los Angeles, Orange, and San Diego, and portions of San Bernardino and Riverside counties, are modeled at a 3-km resolution. The orographic precipitation model is forced by free stream wind velocities given by the 700mb winds from the NCEP Reanalysis I dataset. Atmospheric moisture initial conditions are defined also by the NCEP Reanalysis I dataset, and updated 4x- daily with the available 6-hourly NCEP Reanalysis forcing. This paper presents a comparison of the simulated precipitation to observations for over a variety of spatial scales and over the historical wet season periods from October 2000 to April 2005. The comparison is made over several performance measurements including (a) the occurrence/non-occurrence of precipitation, (b) overall bias and correlation, (c) bias and correlation for precipitation exceeding given thresholds, and (d) the frequency distributions of non-zero precipitation. The results of simulation performance are compared to reported results of other orographically-driven precipitation and regional mesoscale model studies within the Western U.S.
A new methodology is presented for development of macroscale hydrologic model percolation paramet... more A new methodology is presented for development of macroscale hydrologic model percolation parameters from a spatial database of soil properties. This approach is applied to three distinct catchments within California- the Kings, American, and lower Eel river basins. Each unique vertical soil profile in these catchments is divided into an upper and a lower layer based on permeability gradients. A one-dimensional numerical unsaturated flow model is applied to each profile to yield percolation from upper to lower layers as a function of lower layer moisture deficit. A Holtan-type power law relationship is postulated to adequately represent profile percolation, with parameters estimated by curve-fitting the numerical model results. These parameters are then aggregated from the scale of the observable soil profiles to the level of the macroscale hydrologic model elements in a mass-conserving manner. In this process, the power law relationship is considered a describing function approximation to the numerical model response. To estimate parameter and flux uncertainties, a Monte Carlo approach is employed in which the soil property values are sampled randomly within the uncertainty ranges provided in the soils database or established by previous studies in the literature. The resulting ensembles of soil profiles are used in the describing function method to generate ensembles of hydrologic parameter spatial distributions for the catchments of interest. These ensembles are used as measures of hydrologic parameter uncertainty. The relationship between aggregate moisture flux and soil parameter distributions is explored, as is the dependence of mean parameter and aggregate flux values and their respective uncertainties on spatial scale. The methodology presented provides a useful means of investigating model-process observability given present-day soils databases, and a robust method for determining physically based macro-scale hydrologic parameters for subsequent application in hydroclimatic modeling.
Within the context of predicting and mitigating flood hazards, this paper motivates the developme... more Within the context of predicting and mitigating flood hazards, this paper motivates the development of innovative ideas in three engineering research areas: (a) spatially-distributed quantitative rainfall prediction, (b) spatially-distributed flow prediction, and (c) long-term flow prediction for water resources management. Design, operational implementation, real-time testing, and training of field personnel should be the cornerstone of such development for realizing the potential benefits in these fields.
Bulletin of the American Meteorological Society
At the beginning of the 21st Century a research-to-operations program was initiated to design and... more At the beginning of the 21st Century a research-to-operations program was initiated to design and develop operational systems to support local forecasters in their challenging task to provide advance warning for flash floods worldwide. Twenty some years later, the Flash Flood Guidance System with global coverage provides real-time assessment and guidance products to more than 60 countries, serving nearly 3 billion people. The implementation domains cover a wide range of hydroclimatological, geomorphological and land-use regimes worldwide. This flexible and evolving system combines meteorology and hydrology data and concepts as well as supports product utility for flash-flood disaster mitigation on very large scales with high spatial and temporal resolution. Through quality control procedures, it integrates remotely-sensed data of land-surface precipitation and of land-surface properties from geostationary and polar orbiter satellite platforms, reflectivity data from a variety of wea...
Bulletin of the American Meteorological Society
Journal of Hydroinformatics
Corporate technology transfer by US non-governmental organizations with the substantial involveme... more Corporate technology transfer by US non-governmental organizations with the substantial involvement of university faculty is a new activity in hydrometeorology. The issues involved in such US corporate technology transfers are discussed by way of two examples selected from the activities of the Hydrologic Research Center, a non-profit-making public-benefit research and technology transfer corporation in San Diego, California, USA. The projects discussed are: (a) the development and implementation of a robust state estimator for national use within the US National Weather Service River Forecast System, and (b) the development and implementation of a prototype multi-sensor rainfall forecasting system for the Panama Canal Authority. The issues covered include technical ones associated with improving theoretical formulations for robust operational performance, those associated with the necessary reciprocal education between modellers and field personnel, and the accommodation of the educational objectives of participating postdoctoral associates.
Open-File Report
An integrated forecast-control system was designed to allow the profitable use of ensemble foreca... more An integrated forecast-control system was designed to allow the profitable use of ensemble forecasts for the operational management of multipurpose reservoirs. The system ingests large-scale climate model monthly precipitation through the adjustment of the marginal distribution of reservoir-catchment precipitation to reflect occurrence of monthly climate precipitation amounts in the extreme terciles of their distribution. Generation of ensemble reservoir inflow forecasts is then accomplished with due account for atmospheric-forcing and hydrologic-model uncertainties. These ensemble forecasts are ingested by the decision component of the integrated system, which generates non-inferior trade-off surfaces and, given management preferences, estimates of reservoir-management benefits over given periods. In collaboration with the Bureau of Reclamation and the California Nevada River Forecast Center, the integrated system is applied to Folsom Lake in California to evaluate the benefits for flood control, hydroelectric energy production, and low flow augmentation. In addition to retrospective studies involving the historical period 1964-1993, system simulations were performed for the future period 2001-2030, under a control (constant future greenhouse-gas concentrations assumed at the present levels) and a greenhouse-gas-increase (1-% per annum increase assumed) scenario. The present paper presents and validates ensemble 30-day reservoir-inflow forecasts under a variety of situations. Corresponding reservoir management results are presented in Yao and Georgakakos, A., this issue. Principle conclusions of this paper are that the integrated system provides reliable ensemble inflow volume forecasts at the 5-% confidence level for the majority of the deciles of forecast frequency, and that the use of climate model simulations is beneficial mainly during high flow periods. It is also found that, for future periods with potential sharp climatic increases of precipitation amount and to maintain good reliability levels, operational ensemble inflow forecasting should involve atmospheric forcing from appropriate climatic periods.