Space-Time Rainfall: Scaling, Extremes and Prediction

Massachusetts Institute Of Technology, Cambridge MA

Investigators

Abstract

0228835 Veneziano The objective of the project is to study the scaling of rainfall in space and time and develop models for hydrologic application including extremes and short-term forecasting. This is an extension of work done recently by the PI and co-workers on temporal rainfall. The main issues to be addressed are: Scaling invariance: Determine whether and in what form the limited multifractal scaling properties of temporal rainfall (at small scales and within synoptic events) extend to space-time. Modeling: A new type of renewal models with multifractal pulse-based rainfall events has proved successful in modeling temporal rainfall series. Extend these models to space-time in a way that reproduces on/off statistics, scaling invariance, spatial and temporal dependencies, extremes, and other key properties of rainfall. Extremes: Derive theoretical scaling laws for the Area-Intensity-Duration-Frequency curves based on properties of space-time rainfall. Use these theoretical laws to extrapolate beyond the observation range. Forecasting: Develop short-term forecasting methods based on the scaling properties of time-space rainfall. These issues are of great theoretical and practical importance to hydrology. They will be pursued through a combination of statistical modeling, theoretical scaling analysis, numerical simulation and validation, and analysis of space-time rainfall records (from the GATE and TOGA-COARE experiments). In the past, rainfall models have emphasized either the conceptual pulse structure of meso-scale atmospheric circulation or the multifractal cascade structure of turbulence. The models we shall develop combine both aspects and use a parsimonious parameterization. Extremes are of obvious importance for hydrologic design and risk analysis. Theoretical determination of their scaling behavior will allow more confident extrapolation beyond the observation range as well as linkage to climatic parameters. Of equal importance is real time forecasting, for which we intend to develop flexible and practical methods that take advantage of the scaling properties of space-time rainfall. The developments under this project are directly relevant also to space-time rainfall downscaling and disaster prevention through short-term rainfall prediction and long-term flood risk assessment.