Methods of verification and evaluation of forecasts and their inter-comparison, including statistical significance, with a focus on high resolution applications and object based or process based methodologies.

This session will feature applications of AI and machine learning models for analyzing and understanding our climate.

Over the last several decades, substantial progress has been achieved in probabilistic hydrometeorological forecasting theories and applications. However, significant challenges still exist in assessing the uncertainty of complex hydrometeorological processes and improving hydrometeorological predictions, especially extreme hydrometeorological events. This session solicits papers that focus on, but not limit to, (1) addressing uncertainty in hydrometeorological forecasting from a number of sources in both offline and couple systems, and (2) innovative methods in hydrometeorological ensemble forecasting. The former includes uncertainties in forcing data (e.g., quantitative precipitation estimation and meteorological forcing data), initial conditions (e.g., soil moisture and snow status), parameters (e.g., land use and soil texture), model structure (e.g., assumptions, formulations and numerical solutions), and calibration (e.g., single-objective optimization and multi-objective optimizations).  The latter emphasizes integrated ensemble methods to improve hydrometeorological forecasting, verification methods to evaluate probabilistic forecasting, and statistical postprocessing techniques to generate hydrometeorological data products.

Over the last several decades, substantial progress has been achieved in probabilistic hydrometeorological forecasting theories and applications. However, significant challenges still exist in assessing the uncertainty of complex hydrometeorological processes and improving hydrometeorological predictions, especially extreme hydrometeorological events. This session solicits papers that focus on, but not limit to, (1) addressing uncertainty in hydrometeorological forecasting from a number of sources in both offline and couple systems, and (2) innovative methods in hydrometeorological ensemble forecasting. The former includes uncertainties in forcing data (e.g., quantitative precipitation estimation and meteorological forcing data), initial conditions (e.g., soil moisture and snow status), parameters (e.g., land use and soil texture), model structure (e.g., assumptions, formulations and numerical solutions), and calibration (e.g., single-objective optimization and multi-objective optimizations).  The latter emphasizes integrated ensemble methods to improve hydrometeorological forecasting, verification methods to evaluate probabilistic forecasting, and statistical postprocessing techniques to generate hydrometeorological data products.

Description: All observing and assimilation systems contain a wide-variety of parameters.  These parameters could control the quality of retrievals from observing systems.  They might be used to represent background/observation error variances or even the structure of ensemble localization/inflation in assimilation systems.  In all cases these parameters are typically estimated from training sets using statistical methods.  This session intends to provide a broad overview of both theory and practice in the application of statistical parameter estimation for integrated observing and assimilation systems for the atmosphere, oceans, and land surface.  Subject areas of particular interest include, but are not limited to, the following: statistical estimation methods for observation operator development; estimation methods for observation and background error covariance matrices (e.g. Desroziers, Höllingsworth-Lonnberg, etc.); estimation methods for prior/posterior inflation as well as ensemble covariance localization parameters; as well as the application of artificial intelligence/machine learning methods to the tuning/configuration/diagnosis of observing and assimilation systems.

Over the last several decades, substantial progress has been achieved in probabilistic hydrometeorological forecasting theories and applications. However, significant challenges still exist in assessing the uncertainty of complex hydrometeorological processes and improving hydrometeorological predictions, especially extreme hydrometeorological events. This session solicits papers that focus on, but not limit to, (1) addressing uncertainty in hydrometeorological forecasting from a number of sources in both offline and couple systems, and (2) innovative methods in hydrometeorological ensemble forecasting. The former includes uncertainties in forcing data (e.g., quantitative precipitation estimation and meteorological forcing data), initial conditions (e.g., soil moisture and snow status), parameters (e.g., land use and soil texture), model structure (e.g., assumptions, formulations and numerical solutions), and calibration (e.g., single-objective optimization and multi-objective optimizations).  The latter emphasizes integrated ensemble methods to improve hydrometeorological forecasting, verification methods to evaluate probabilistic forecasting, and statistical postprocessing techniques to generate hydrometeorological data products.

Physical interpretability in machine learning