About this Research Topic
Quantitative precipitation forecasting, or nowcasting, at temporal lead times of minutes to a few days is crucial for river management, including operations of dams and reservoirs, emergency preparedness, such as advance flash flood warning and advisories, and power production, for example hydro and nuclear power plants. However, spatially distributed precipitation nowcasting remains a challenge with limited improvements over the years. Numerical weather prediction and cloud resolving model refreshed with data assimilation, meteorologists applying judgmental factors to the weather predictions, and data-driven methods including statistics and image processing, often fail to improve consistently over advection and optical flow or even persistence, and systematically miss the extremes. While recent development in machine learning seem to offer hope for progress, considerable ground needs to be covered to improve forecasts in a way that makes them useful to and usable by end users and stakeholders.
Recent progress in machine learning and computer vision, including hybrid approaches combined with process understanding as well as relatively physics-free approaches, suggest the tantalizing possibility of step jumps in precipitation nowcasting. However, there have been limited studies that take such improvements all the way to river, emergency, or power plant managers. Furthermore, neither physics-based numerical models nor machine learning or related data-driven methods nor even their hybrid combinations exhibit skills in predicting the extremes of precipitation which leads to flash floods and loss of lives and service. Suggestions have been made that climate and land use intensify precipitation extremes while additional exposure and vulnerability cause even more damage. The ability to develop nowcasting methods that measure not just statistical improvements but directly address metrics and end use of relevance to stakeholders is critical under these scenarios.
This Research Topic welcomes original articles, case studies, reviews, and commentaries that address the precipitation nowcasting problem from the perspectives of fundamental understanding in hydrometeorology, novel developments in data-driven methods and physics-based models, hybrid physics-data combinations, uncertainty quantification and decision mining, metrics and methods that characterize and translate improvements in nowcasting to end user needs, improved use of forecasts for river management including emergency management and power production, and pragmatic case studies and approaches to inform stakeholders and end users. Interdisciplinary ideas that can lead to actionable and risk-informed insights, whether from methodological or science advances, or from meta-analysis of case studies, are particularly encouraged.
Recent progress in machine learning and computer vision, including hybrid approaches combined with process understanding as well as relatively physics-free approaches, suggest the tantalizing possibility of step jumps in precipitation nowcasting. However, there have been limited studies that take such improvements all the way to river, emergency, or power plant managers. Furthermore, neither physics-based numerical models nor machine learning or related data-driven methods nor even their hybrid combinations exhibit skills in predicting the extremes of precipitation which leads to flash floods and loss of lives and service. Suggestions have been made that climate and land use intensify precipitation extremes while additional exposure and vulnerability cause even more damage. The ability to develop nowcasting methods that measure not just statistical improvements but directly address metrics and end use of relevance to stakeholders is critical under these scenarios.
This Research Topic welcomes original articles, case studies, reviews, and commentaries that address the precipitation nowcasting problem from the perspectives of fundamental understanding in hydrometeorology, novel developments in data-driven methods and physics-based models, hybrid physics-data combinations, uncertainty quantification and decision mining, metrics and methods that characterize and translate improvements in nowcasting to end user needs, improved use of forecasts for river management including emergency management and power production, and pragmatic case studies and approaches to inform stakeholders and end users. Interdisciplinary ideas that can lead to actionable and risk-informed insights, whether from methodological or science advances, or from meta-analysis of case studies, are particularly encouraged.
Keywords: Nowcasting, precipitation, weather forecasting, predictive modeling, spatiotemporal data, river management, power production, emergency management, infrastructure operations
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.
