Comparison of the methods

Here a simple test is described that was performed to see the differences between the methods. A unwrapped interferogram was obtained of the Veluwe (Holland) by processing the bottom half of the Tandem images 3512 (ERS2) and 23185 (ERS1). The interferogram was multilooked by 40x8, resulting in 367 lines and 610 pixels. There were about 3 fringes.

Baseline:
$B = 185 m, \alpha = 3\ensuremath{^\circ}$
${B_{\parallel}}= -62, {B_{\perp}}= -173$
${B_{h}}= -184, {B_{v}}= -2$

The processing was done with a debugger version of the Doris software, so the cpu times are not really representative for the performance of Doris. table 32.1 shows some processing parameters.

Table 32.1: Processing with the methods

Method	cpu	options	remarks
Ambiguity	60	-	geocoding as well
Schwabisch	12	1000 pnts, 1d=2, 2d=5	-
Rodriguez	4	-	-

Figure 32.2 and 32.3 show plots for these three methods. Figure 32.4 shows a comparison between schwabisch and ambiguity method.

Figure 32.2: Comparison methods in azimuth direction for line 250.

Figure 32.3: Comparison methods in range direction for pixel 110.

Figure 32.4: Comparison ambiguity, schwabisch methods for total image.

It can be seen there is a trend between schwabisch and ambiguity. And there is an offset with rodriguez.

Schwabisch method is always higher then ambiguity, suggests error in computation of baseline parameters?

Some other tests also showed that the method Schwabisch, as implemented in Doris, seems to yield a (more or less) scaled version of the height of the ambiguity method. (The schwabisch being higher).

After rescaling (with a factor 0.86) of the heights obtained by schwabisch method to the level of the ambiguity method, the differences between both methods were only a few meters.