Modelling and correcting azimuthal anisotropy in Sentinel-1 backscatter data

Figures & data

Figure 1. Geometric dependency of the incidence radar beam: Local incidence angle (vertical) angle and azimuth (horizontal) angle .

Table 1. Percentage frequency of the absolute value of the normalized backscatter difference over the area of Western Europe (Figure 2). Original dataset and after azimuth-correction of the dataset (October 2014 – December 2016).

	normalized backscatter difference
	mean	>0.25dB	>0.5dB	>0.75dB	>1dB	>1.5dB
Original	0.07dB	58.5%	30.1%	16.3%	8.7%	2.4%
Azimuth-corrected	0.0dB	6.8%	4.0%	2.8%	2.1%	1.2%

Figure 2. Normalized backscatter difference over Western Europe (October 2014 – December 2016).

Figure 3. Austrian Alps: Backscatter shift due to different slope orientations: (a) Calculated from Sentinel-1 backscatter (equation (2)), and (b) modelled from SRTM-3 DEM using equation (4). (c),(d) Detail example of the Zillertal in the Tyrolean Alps.

Figure 4. (a) Illustration of different azimuth angles for ascending and descending overpasses observing a predominant slope, and (b) modelled backscatter difference depending on the slope orientation . (using equation (4), with , , s = 1).

Figure 5. (a) Incidence and azimuth angle normalization workflow of the Sentinel-1 backscatter data, and (b) an example of the normalization steps: (i) 2 years of Sentinel-1 backscatter observed in VV-polarisation separated by orbit number, (ii) incidence angle normalized backscatter, and (iii) normalized and azimuth-corrected backscatter.