The Advances in Targeted Observations for Tropical Cyclone Prediction Based on Conditional Nonlinear Optimal Perturbation (CNOP) Method - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 574

Table 24.1 The similarities among the energy distributions
obtained at 30, 60, and 120-km resolutions for TC Matsa (2005),
TC Meari (2004), and TC Mindulle (2004)
30 and 60 km

60 and 120 km

TC Matsa 0.70 0.78

TC Meari 0.55 0.75

TC Mindulle 0.49 0.72

Fig. 24.2 Verification areas for different designs. The solid rectangles in panel ( a ), ( b ), and ( c )
are the verification areas for schemes A, D, and G respectively, while the dashed rectangles are
for schemes B, E, and H respectively, the dotted rectangles are for C, F, and I respectively. The
observation tracks of the cyclone are also shown in the center of the domain (From Zhou and
Mu 2011 )

favorable for CNOP to display the nonlinear information, which play an important
role in the evolution of the initial perturbations, thus, as far as the computation
condition is permitted, using the CNOP method at a high resolution to identify the
sensitive areas may be more beneficial in targeted observations for tropical cyclone
predictions.

24.3.3 Sensitivity of CNOP Sensitive Areas with Respect

to the Verification Area Design

As indicated in Sect. 24.2.1 , the CNOPs also depend on the verification area design
(namely, different parameter P , and this has been studied in the paper of Zhou and

Mu ( 2011 ). The tropical cyclone Rananim, which occurred in the northwest Pacific
Ocean in 2004, is studied.
The design of the verification area is as follows. First, they defined a control
design in which the verification area includes the real cyclone tracks (the best storm
tracks) during the integral time period: scheme B (Fig. 24.2 a, dashed rectangle).
Second, the size of the verification area is kept constant as it is moved to other
places (schemes A, C, D, E, and F). Small positional variations are denoted as
schemes A and C (Fig. 24.2 a, solid and dotted rectangles, respectively). Large
position variations are denoted as schemes D, E, and F (Fig. 24.2 b, solid, dashed,
and dotted rectangles, respectively). Then both the size and the position variations

/