1.

INTRODUCTION

CyCLOPS consists of a comprehensive multi-parametric network spread across the government-controlled region of Cyprus, along with an Operation Centre. The permanent segment, illustrated in Figure 1, features six sites, each equipped with a Tier-1 GNSS continuously operating reference station (CORS) and two calibration-grade triangular trihedral corner reflectors (CRs) with an inner length of 1.5 meters. These are designed to support both the ascending and descending tracks of Synthetic Aperture Radar (SAR) satellite missions, such as ESA’s Sentinel-1 ¹. The site selection was guided by a GIS-based methodology developed by the CUT Laboratory of Geodesy ², adhering to stringent international standards ^3–5, thereby ensuring the network’s capability for precise crustal movement monitoring and enhancing SAR data.

Figure 1.

The CyCLOPS permanent network ². Green triangles denote the collocated GNSS CORS and CR sites.

For precise deformation monitoring using SAR acquisitions, external calibration is crucial. This involves the use of ground targets like corner reflectors (CRs), which are artificial persistent scatterers, with known Radar Cross Section (RCS), to ensure effective system calibration ^6,7. Radiometric calibration is achieved by comparing the observed backscatter signal intensity from CRs with their theoretical RCS (RCS_T) ⁸, thus allowing the determination of an absolute calibration factor ⁹. The goal is to establish an absolute calibration factor by associating SAR image pixel values with precise geophysical parameters, ensuring the imagery’s precision and utility for various applications ¹⁰.

Geolocation accuracy refers to the precision with which an Earth-observing remote sensing platform can determine the geographic positions of surface features in its images ¹¹. By comparing the surveyed in-situ coordinates of CRs to the actual location of its peak intensity within a specific image, one can determine an offset, accounting for system biases, and effects such as atmospheric path delay, plate tectonics, and solid Earth tide (SET) disturbances ¹². Concordantly, the collocation of CRs with GNSS CORS enhances the calibration process, evidently leading to greater accuracy in estimating interferometric phases and producing more reliable deformation products ¹³.

The objective of this paper is to present the initial results of assessing ESA Sentinel-1A (S1A) SAR performance using the response from CyCLOPS CRs. The structure is as follows: Sections 2, 3, and 4 outline the procedure used to assess CRs in terms of radiometric parameters, while Section 5 derives the Absolute Location Error (ALE) of SAR products.

This paper summarizes key findings and methodologies that are part of a more extensive work published in ¹⁴. For a more detailed analysis and comprehensive discussion, readers are encouraged to refer to the journal article.

2.

IMPULSE RESPONSE FUNCTION

The Impulse Response Function (IRF) in SAR imaging, also known as the point-spread function, depicts the two-dimensional luminosity pattern of a point scatterer such as a CR or transponder in focused images. The 3 dB width of the IRF determines the spatial resolution, while the sidelobes provide insights into the performance of the SAR instrument and processor. Key image quality metrics include spatial resolution (3 dB width of the peak lobe), Peak Side-Lobe Ratio (PSLR), and Integrated Side-Lobe Ratio (ISLR), all derived from the point target’s IRF ⁷. Table 1 lists standard S1 Interferometric Wide Swath (IW) Level-1 SLC product characteristics as defined in the ESA S1 product definition ¹⁵.

Table 1.

S1 IW SLC product performance parameters.

The actual image quality performance was assessed through a temporal evaluation of the CRs’ responses within the SLC images. A total of twenty-six Level-1 uncalibrated intensity images with linear vertical polarization (VV) from 29.06.2021 to 18.10.2023 were used to periodically monitor the consistency and stability of the quality parameters. Subsequently, the IRF parameters in both range and azimuth for each CR were calculated as defined in the CEOS standard definition, using the GAMMA software package ¹⁶. The IRF for each CR in all acquired images was generated by applying the integral method ¹⁷, keeping the clutter window size as large as the window used for extracting the point target to maintain result reliability. The peak of the point target is situated at the central pixel of the data segment, as identified by the image coordinates. The images were oversampled by a factor of 16, resulting in an oversampled point target image with dimensions of 256x256, originating from a 16x16 image. Consequently, the dimensions of the interpolated image are dictated by the magnitude of the oversampling factor. Finally, the temporal variation of the spatial resolution, PSLR, and ISLR from each CR is obtained.

Figure 2 illustrates a representative example of the TROU01 CR impulse response. Among the three parameters mentioned, spatial resolution, which refers to the minimum distance at which two distinct objects on the ground surface can be discerned as separate entities in an image, is the most critical in assessing image quality ¹⁸.

Figure 2.

The TROU01 CR response function, after interpolation, in the S1A - IW3 - VV image of 23/09/2023 ascending pass. (a) and (b) plots represent a cut through the peak in the slant range and azimuth direction, respectively, while (c) illustrates the point target relative power in a 3D plot.

Spatial Resolution in SAR Imaging

Spatial resolution in range and azimuth directions is defined by the width of two points at the –3 dB mainlobe ⁷. As shown in Figures 2a and 2b, for instance, a relative power of –3 dB (y-axis) represents a width of 3.5 m and 22.4 m (x-axis) in range and azimuth, respectively. Figure 3 illustrates the temporal variation of spatial resolution from the TROU01 CR.

Figure 3.

The temporal variation of spatial resolution from TROU01 CR, for the monitoring period.

Temporal analysis indicates consistent and reliable spatial resolution, aligning with S1A’s IW3 swath specifications. Assessments across IW1, IW2, and IW3 swaths (see Table 2) reveal that CRs’ responses maintain consistency with theoretical slant range and azimuth resolutions.

Table 2.

Statistics of Spatial Resolution Quality Assessment from all the twelve (12) CRs.

	Range [m]	Azimuth [m]
IW1
Mean	2.7	21.9
St. Deviation	0.09	0.5
Max	2.8	22.8
Min	2.5	20.8
IW2
Mean	3.1	22.1
St. Deviation	0.09	0.6
Max	3.3	23.5
Min	2.9	21.2
IW3
Mean	3.5	21.8
St. Deviation	0.09	0.5
Max	3.7	22.8
Min	3.4	20.8

Discrepancies exceeding theoretical values suggest potential processing issues, such as inaccurate orbit data ¹⁹.

3.

RCS AND SCR ESTIMATION

To ensure sufficient visibility of point targets above the surrounding background clutter, it’s essential to evaluate the target size as observed in SAR images. This evaluation is achieved through the estimation of the RCS of the CRs, which measures the amount of energy backscattered by a target. Additionally, the visibility of CRs above the clutter is determined by the Signal to Clutter Ratio (SCR), calculated as the ratio of the CR’s RCS to the clutter RCS. RCS estimation can be performed using either the integral method ¹⁷ or the peak method ²¹. The peak method relies on the resolution parameters of the point targets, making its accuracy dependent on the quality of the point target image, such as clarity and focus, for precise calibration factor determination. In contrast, the integral method is independent of point target parameters ²². The RCS of each CR was measured in every SAR image using both methods and compared to the theoretical value to calculate the mean RCS for evaluation.

The theoretical RCS at C-band for a triangular trihedral CR of 1.5 m is 38.38 dBm²; therefore, the estimated RCS values should not exceed this theoretical value ²³. Generally, for SAR images, the RCS of a triangular trihedral CR of 1.5 m suitable for radiometric calibration at C-band should range between 34 and 38 dBm^{2 24}. Regarding SCR requirements, a CR should provide high-intensity radar reflections in SAR images while maintaining a low level of surrounding scatterers’ signal reflectivity, ensuring a high SCR. Thus, a high-intensity and temporally stable backscatter response is crucial for the correct identification of a CR in SAR images ²⁵. For a 1.5 m CR operating at C-band, the minimum suitable SCR for radiometric calibration should be at least 20 dB ^24,26. Even if the surrounding clutter is low, it should be estimated during the analysis, typically in the corners of the image chip, and subtracted from the signal estimate. Estimating the SCR also involves assessing factors such as phase stability.

A total of 195 Ground Range Detected (GRD) image products (IW swath, VV polarization), including both ascending and descending paths, from 17.06.2021 to 26.07.2023, were processed for RCS and SCR estimation using the integral method. The integral method can be used in multilook images. The responses of the CRs were calculated using CoRAL software, and the methodology followed is described in ²⁷. The mean estimated RCS is 37.58 ± 0.05 (1σ) dBm², which is a 0.8 dBm² difference from the RCST. The estimated mean SCR, assuming spatial ergodicity, is 24.23 ± 0.08 (1σ) dB.

For the peak calibration method, 297 SLC image products (IW swath, VV polarization), including both ascending and descending paths, from 22.06.2021 to 29.10.2023, were processed. The RCS and SCR estimation, assuming temporal ergodicity of the clutter, were extracted using GECORIS software, and the methodology followed is described in ²⁸. The mean RCS is 37.6 ± 0.03 (1σ) dBm², which is a 0.78 dBm² difference from the RCS_T, while the mean SCR is 27± 0.8 dB.

As shown in Figure 4, the mean estimated RCS difference between the two methods is 0.02 dBm², with the peak method’s standard 1σ error being smaller than that of the integral method. These findings align with existing literature, indicating that RMS errors associated with the peak method are consistently smaller than or at least equal to those of the integral method in well-focused systems ²¹.

Figure 4.

Estimated RCS difference from RCS_T for all the CRs using both methods.

4.

SLANT DISTANCE ERROR IN LOS

The pixel value containing the CR results from a complex sum of the backscatter signal contributed by the CR, the dominant scatterer, and the distributed individual scatterers within the pixel. This clutter contribution results in an uncorrelated signal, and the probability density function for the phase error (φ) magnitude of a CR can be estimated by its SCR ²⁴. For accurate SAR image calibration using a CR, it is crucial to maintain the phase standard deviation below 0.25 radians to ensure that phase residuals are normally distributed, and the SCR phase variance estimate remains unbiased ^{24–26,29,30}. By using λ, the φ angle’s radians can be converted into a Line of Sight (LoS) slant distance error. For an SCR of 20 dB, the theoretical dispersion threshold for the LoS displacement error in C-band is approximately 0.31 mm. Figure 5 illustrates the average LoS distance error for each CR, where phase stability ranges from 0.04 to 0.05 radians, and all values lie below the aforementioned threshold, varying between 0.17 mm and 0.21 mm.

Figure 5.

Mean displacement error in LoS.

5.

GEOLOCATION ANALYSIS

Another critical aspect of assessing SAR image quality is geometric calibration, where pixel coordinates are compared to well-known locations of reference point targets, such as CRs, whose positions have been surveyed using adequate GNSS receivers. Due to accurate time and precise orbit determination, SAR images demonstrate sufficient geometric accuracy. However, these measurements are susceptible to factors such as variable atmospheric conditions, Earth dynamics, and approximations made during SAR processing. These influences can introduce apparent displacement shifts, occasionally reaching several meters. Mitigating these influences requires several post-processing steps and external data for accurate correction ³¹.

For each CR, a fixed reference point (RP) below the CR apex (phase center) was surveyed using GNSS static measurements, considering the height offsets between the RP and the apex. Their 3D coordinate positions were corrected with respect to the coordinates of the co-located CyCLOPS GNSS/CORS at each site, in ITRF 2014. Using GECORIS software, ALE for each CR was extracted. A representative example of the timing corrections can be seen in Figure 6, where the mean ALE in TROU01 is –0.029 ± 0.046 m, and –0.198 ± 0.287 m in range and azimuth, respectively.

Figure 6.

An example of the ALE estimation in TROU01 after mitigating the propagation influences. Blue triangles indicate the initial positions whereas black triangles are the final positions, in each acquisition.

6.

CONCLUSIONS

The temporal analysis of IRF parameters, including spatial resolution, ISLR, indicated consistent and reliable performance of Sentinel-1A. The spatial resolution of CR responses aligned closely with the theoretical expectations for the IW swath, with minor discrepancies attributed to potential processing issues such as inaccurate orbit data.

Both the integral and peak methods for RCS estimation provided results close to the theoretical values, with mean estimated RCS values slightly below the RCS_T for a 1.5 m CR at C-band. The SCR estimates confirmed the high-intensity and stable backscatter response required for accurate CR identification in SAR imagery. Furthermore, the analysis of phase stability and corresponding slant distance errors in LoS ensured that the SCR phase variance remained unbiased, with average displacement errors in LoS well within acceptable limits.

Geometric calibration through comparison of pixel coordinates with surveyed CR positions revealed that geometric accuracy in SAR images is influenced from several factors such as atmospheric conditions and Earth dynamics which introduced potential displacement shifts, necessitating post-processing corrections. The ALE analysis has proven effective in mitigating atmospheric and dynamic Earth influences, ensuring localization accuracy.

Concordantly, the CyCLOPS infrastructure is a state-of-the-art, reliable unit for radiometric calibration and validation of SAR products, which will contribute significantly to the precision and reliability of SAR imaging, crucial for various applications such as crustal movement monitoring.