Tropical cyclone forecasting with model‐constrained 3D‐Var. I: Description

Abstract

AbstractA disadvantage of three‐dimensional variational data assimilation (3D‐Var) technique is that it cannot make use of complicated constraints such as the dynamics and physics in a numerical model, which are used in the four‐dimensional variational data assimilation (4D‐Var) technique. On the other hand, using a numerical model and its adjoint in the 4D‐Var technique requires a large amount of computer resources, thus limits its practical application. In this paper, a new 3D‐Var method is proposed by adding a numerical model as a constraint. This method minimizes the distance between observation and model variables and time tendency of model variables, which makes the optimized initial conditions (ICs) that not only fit the observations but also satisfy the constraints of full dynamics and physics in the numerical model. The forward and adjoint models used in this method are as same as those in the 4D‐Var method but are only integrated one time step to calculate the time tendency. Because observations are only used at one time slice and are being constrained by the model, it is called the model‐constrained 3D‐Var (MC‐3DVar) technique.A set of ideal experiments based on a shallow‐water equation model indicates that the model constraints used in MC‐3DVar can spread the observation information spatially and balance the model variables. Part II employs this method to improve tropical cyclone track forecasting using AMSU‐A, QuikSCAT and cloud‐drift wind data. The study shows that the assimilation of these data with MC‐3DVar improves tropical cyclone forecasts and that more satellite data give better performances. Copyright © 2007 Royal Meteorological Society