Combination of stereo SAR and InSAR for DEM generation using TanDEM-X spotlight data

Figures & data

Figure 1. Upper left image: TSX image of 9 March 2015 in Table 1 with the selected GCPs (G1–G7). The other images are zoomed around each of the encircled GCPs.

Figure 2. UTM geocoded staring spotlight image, 10 March 2015 in Table 1. Pixel spacing is 1.8 m for the Northing (vertical) and Easting (horizontal). Scene centre: 59 º 59ʹ 10ʺ N, 11º 3ʹ 0ʺ E.

Figure 3. UTM geocoded StInSAR-derived DEM, 10 March 2015 in Table 1. Pixel spacing is 1.8 m for the Northing and Easting.

Figure 4. Sub-images of Figures 2 and 3 showing some buildings (B1–B5). B4 is shown in Figure 3.

Figure 5. Staring spotlight image over Nikel in Russia on 4 February 2015 in Table 2. G1 is the GCP and A1–A3 are small regions for the estimation of height.

Figure 6. Interferogram from the same time and area as in Figure 5.

Figure 7. UTM geocoded DEM with a pixel spacing of 1.8 m derived from the interferogram in Figure 6. Scene centre: 69 º 24ʹ 36ʺ N, 30º 12ʹ 11ʺ E.

Figure 8. (a) UTM geocoded DEM of 22 December at incidence angle 47º in an area around the A1 point shown in Figure 5. (b) Average of six DEMs at 47º.

Table 1. Ascending CoSSC interferometric pairs from test site Kjeller (scene centre: 59 º 59ʹ 10ʺ N, 11º 3ʹ 0ʺ E). The size of staring spotlight scenes is around 3 km × 6 km. The pixel size in azimuth is 0.17 m and in range 0.79 m.

	Date	Amb (m)	Inc (º)	Coherence
Staring spotlight mono-static	22 December 2014	−26.6	49.622	0.59
Staring spotlight mono-static	24 January 2015	−29.0	49.609	0.57
Staring spotlight mono-static	9 March 2015	16.8	49.612	0.57
Staring spotlight mono-static	10 March 2015	14.8	32.415	0.67
Strip map bi-static	1 November 2015	−62.1	42.672	0.73

Table 2. Ascending CoSSC mono-static staring spotlight interferometric pairs from test site Nikel (scene centre: 69 º 24ʹ 36ʺ N, 30º 12ʹ 11ʺ E). The size of staring spotlight scenes is around 3 km × 6 km. The pixel size in azimuth is 0.17 m and in range 0.79 m.

Date	Amb (m)	Inc (º)	Coherence
29 December 2014	−11.5	27.602	0.77
9 January 2015	−10.0	27.600	0.79
20 January 2015	−19.9	27.587	0.86
11 February 2015	13.4	27.604	0.80
22 February 2015	9.9	27.587	0.77
Date	Amb (m)	Inc (º)	Coherence
23 December 2014	−17.6	34.968	0.83
3 January 2015	−11.8	34.987	0.77
14 January 2015	−13.3	34.986	0.78
16 February 2015	11.7	34.980	0.76
27 February 2015	11.1	34.989	0.71
10 March 2015	17.5	34.983	0.81
Date	Amb (m)	Inc (º)	Coherence
17 December 2014	−28.9	41.430	0.81
28 December 2014	−14.0	41.453	0.76
8 January 2015	−11.4	41.452	0.72
19 January 2015	−20.0	41.453	0.78
20 February 2015	17.1	41.462	0.70
21 February 2015	−11.4	41.452	0.75
Date	Amb (m)	Inc (º)	Coherence
22 December 2014	−22.4	47.027	0.75
2 January 2015	−14.1	47.039	0.67
13 January 2015	−14.7	47.041	0.71
4 January 2015	35.1	47.041	0.75
15 January 2015	14.4	47.036	0.67
9 March 2015	18.5	47.042	0.65
Time	Amb (m)	Inc (º)	Coherence
16 December 2014	−40.2	51.837	0.65
27 December 2014	−17.0	51.868	0.67
7 January 2015	−13.4	51.866	0.62
9 February 2015	22.0	51.867	0.63
20 February 2015	13.9	51.872	0.60
9 March 2015	15.3	51.864	0.54

Table 3. Ascending CoSSC interferometric pairs from test site Åraksbø (scene centre: 58º 56ʹ 20ʺ N, 7º 47ʹ 2ʺ E). The size of spotlight scenes is around 5 km × 10 km. The pixel size in azimuth is 0.86 m and in range 0.79 m.

	Date	Amb (m)	Inc (º)	Coherence
Spotlight Mono-static	22 December 2014	−20.3	35.454	0.58
Spotlight Mono-static	24 January 2015	−24.4	35.457	0.69
Spotlight Mono-static	9 March 2015	14.2	35.458	0.47
Spotlight Mono-static	27 December 2014	−16.1	44.651	0.56
Spotlight Mono-static	18 January 2015	−15.8	44.659	0.52
Spotlight Mono-static	9 February 2015	24.8	44.650	0.55
Spotlight Mono-static	21 December 2014	−30.5	52.000	0.56
Spotlight Mono-static	23 January 2015	−26.0	52.000	0.54
Spotlight Mono-static	14 February 2015	18.7	51.994	0.48
Spotlight Bi-static	9 December 2015	47.9	35.373	0.67
Spotlight Bi-static	20 December 2015	−48.2	35.384	0.69
Spotlight Bi-static	3 December 2015	62.0	44.593	0.68
Spotlight Bi-static	14 December 2015	62.2	44.576	0.65
Spotlight Bi-static	05 November 2015	−80.0	51.963	0.64
Spotlight Bi-static	16 November 2015	−77.2	51.977	0.65

Table 4. Stereo-estimated heights and DGPS heights of the ground control points (GCPs) and reference points in test site Kjeller.

	Height stereo (m)	DGPS (m)	Type	Reference point	GCP
G1	112.47		Light pole	No	Yes
G2	115.30		Traffic sign	No	Yes
G3	106.20		Fence	No	Yes
G4	103.40	103.60	CR	Yes	Yes
G5	153.60		Wall	No	Yes
G6	104.22	104.00	CR	Yes	Yes
G7	103.10	102.90	CR	Yes	Yes

Table 5. Mean error (ME) and root mean square error (RMSE) heights estimated over four GCPs (G1, G2, G3, and G5) in the four staring spotlight scenes in Table 1.

Reference height	ME (m)	RMSE (m)
G4	−0.24	0.38
G6	0.22	0.32
G7	0.10	0.41

Table 6. ME and RMSE estimated over 14 trigonometric points in the four staring spotlight scenes in Table 1 for each of the GCPs.

	G1	G2	G3	G4	G5	G6	G7
ME (m)	−0.06	0.26	−0.24	0.30	−0.26	0.38	0.13
RMSE (m)	0.73	0.79	0.71	0.71	0.64	0.84	0.69

Table 7. Heights of buildings B1–B7 in Figures 3 and 4 obtained with three different methods.

	StInSAR DEM (m)	Laser pen (m)	Google Earth (m)
B1	7.5	8.4	8
B2	9.6	10.0	10
B3	14.6	14.8	16
B4	23.2	23.5	24
B5	8.4	10.8	10
B6	14.6	15.3	15

Figure 9. Averaged time series of DEMs at different incidence angles in a small region around the A2 and A3 points shown in Figure 5. The dates of the DEMs for each incidence angle are: (a) Incidence angle = 27°. Dates: 29 December, 9 January, 20 January, 11 February, 22 February. (b) Incidence angle = 35°. Dates: 23 December, 3 January, 14 January, 16 February, 27 February, 10 March. (c) Incidence angle = 41°. Dates: 17 December, 28 December, 8 January, 19 January, 10 February, 21 February. (d) Incidence angle = 47°. Dates: 22 December, 2 January, 13 January, 4 February, 15 February, 9 March. (e) Incidence angle = 52°. Dates: 16 December, 27 December, 7 January, 9 February, 20 February, 3 March.

Figure 10. Photograph taken by Johnny Foss over Åraksbø with powered paragliding on 19 July 2012 with licence from Norwegian airsport federation.

Figure 11. Bi-static spotlight image over Åraksbø on 16 November 2015 in Table 3.

Figure 12. (a) Coherence of the bi-static image pair on 5 November 2015 in Table 3. Azimuth time difference of TSX and TDX is 0.03 s. (b) Coherence of the mono-static image pair of 21 December 2014 in Table 3. Azimuth time difference of TSX and TDX is 10 s.

Figure 13. UTM geocoded DEM derived from the bi-static interferometric image pair of 16 November 2015 in Table 3. Pixel spacing is 2.0 m for the Northing and Easting. Scene centre: 58º 56ʹ 20ʺ N, 7º 47ʹ 2ʺ E.

Figure 14. Difference of the DEM in Figure 13 and the DTM from NMA.

Table 8. Mean error (ME) and root mean square error (RMSE) in the four mono-static scenes from Table 3 using four stereo points and five trigonometric points as references.

	G1	G2	G3	G4
ME (m)	−3.05	−1.78	−1.91	−1.81
RMSE (m)	4.28	3.59	3.18	2.61

Table 9. Mean error (ME) and root mean square error (RMSE) in three bi-static scenes from Table 3 using four stereo points and five trigonometric points as references.

	G1	G2	G3	G4
ME (m)	−3.04	−0.25	−3.08	−2.08
RMSE (m)	4.58	3.20	4.36	3.63

Figure 15. (a) Road upon the high hydropower dam. (b) SAR image showing low and high hydropower dam at Hovatn. (c) Low and high dam with the measuring stick. Photographs: Knut Eldhuset, October 2015.