Hydro Genome Mapping: An Approach for the Diagnosis, Evaluation and Improving Prediction Capability of Hydro-Meteorological Models

摘要

Hydro-meteorological models form a cornerstone for critical societal decisions. The reliability of these models in terms of the trustworthiness of their outputs and the consistency of their performance across diverse hydro-meteorological conditions and time periods, has been widely recognized to be questionable. The major evaluation methods used to derive knowledge about hydro-meteorological models are open loop, calibration and data assimilation (DA). The open loop case tests the model output against observation; the calibration optimizes model output to observations; and the DA approach accounts for errors and corrects model trajectory to find an optimal merger between model and observation. These methods focus on reproducing past observation, which has been widely acknowledged to be insufficient in both model diagnosis and prediction capability for ultimate improvement in forecasts. Consequently, this paper makes the case for a definitive framework for the proposed hydro genome mapping based on the concept of biological genome mapping. The study outlines the procedure for the generation of genome-like data for hydro-meteorological models and subsequent hydro genomic mapping methods to construct a thorough understanding of the model, and its decision variables and outputs. The hydro genomic mapping has the capability to locate specific hydro markers in model decision space that are responsible for certain hydro-meteorological responses, leading to both diagnostic and predictive descriptions of the model. As an integrated framework, the hydro genome mapping approach is not a single specialized method but a combination of innovative methodologies from biological genome mapping, evolutionary computation and DA, designed to find answers to critical hydro-meteorological questions. Accordingly, the proposed hydro genomic mapping approach provides a framework towards a synthesis of both Newtonian and Darwinian approaches with opportunities for a greater insight into hydro-meteorological processes and phenomena.