next up previous [pdf]

Next: Error analysis using the Up: Comparison of accuracy between Previous: Comparison of accuracy between

A simple illustration of the accuracy

We first use a simple 1-D example shown in Figure 4 to demonstrate the accuracy of the SGL method and SGLFD method when they are used to calculate the partial derivatives in equation 9. The velocity increases linearly from 1000 to 2275 m/s. The rank is 2 for lowrank decomposition, assuming 1 ms time step. The exact $ kx$ -space operator $ \partial /\partial ^+x$ in equation 9 is shown in Figure 4a. Figures 4b4c and 4d display errors of approximation operators of SGL, SGLFD and conventional staggered grid finite-difference (SGFD), respectively. Figure 5 shows the middle column of the error matrix. The errors of SGL and SGLFD are significantly smaller than that of SGFD.

Mexact Mlrerr Mapperr Mfd10err
Mexact,Mlrerr,Mapperr,Mfd10err
Figure 4.
(a) $ kx$ -space operator $ \partial /\partial ^+x$ for 1-D linearly increasing velocity model. (b) Error of SGL operator. (c) Error of 8th order SGLFD operator. (d) Error of 8th order SGFD operator.
[pdf] [pdf] [pdf] [pdf] [png] [png] [png] [png] [scons]

slicel
slicel
Figure 5.
Middle column of the error matrix. Blue dashed line: SGL operator. Green dotted line: the 8th order SGLFD operator. Red solid line: the 8th order SGFD operator.
[pdf] [png] [scons]


next up previous [pdf]

Next: Error analysis using the Up: Comparison of accuracy between Previous: Comparison of accuracy between

2014-06-02