Hasselmann (1980) derived a 'signal-image-Fourier-transform' (SIFT) algorithm that contracts synthetic aperture radar (SAR) image formation and Fourier analysis into a single quadratic transform of the raw return signal, delivering the two-dimensional variance spectrum of the SAR image without forming the image - motivated by on-board data compression for global satellite wave spectroscopy. The paper's deliverable was expressly the SAR image spectrum, and it presented global surface-wave spectral retrieval as the possibility this data compression opens; inferring the actual ocean surface-wave spectrum requires inverting the strongly nonlinear sea-surface-to-SAR-image mapping, which the paper did not provide. Its associated operational questions (the relative merits of SAR-SIFT versus a two-dimensional dual-frequency scatterometer, real-time on-board implementation, and the validity of the contracted formulation in geometrically distorted regimes) were addressed or bypassed by the 1981-1984 follow-up literature and lie outside the open core. Later literature derived the closed nonlinear mapping and its Gaussian azimuthal cutoff (Hasselmann and Hasselmann 1991), simplified it (Krogstad 1992), removed the 180-degree propagation-direction ambiguity via image cross-spectra (Engen and Johnsen 1995), improved iterative first-guess inversions (Hasselmann et al. 1996; Schulz-Stellenfleth et al. 2005), and continues today with deep-learning retrievals (Wu and Li 2023; Fang et al. 2025, 2026); the instrument-design and digital-implementation questions were substantially addressed in 1982-1984 (Alpers and Hasselmann; LaHaie et al.). No field-accepted inversion recovers the azimuth-cutoff-suppressed azimuth-traveling components of the wave spectrum from a single SAR acquisition in the strongly nonlinear regime - the nearest attempts are recent deep-learning approximations of the inverse mapping, trained on databases generated by the forward model and reported at preprint and conference level - and the azimuth-cutoff information loss has neither been inverted away for a single SAR acquisition nor proven intrinsic (published mitigation adds simultaneous viewing geometries and achieves only partial compensation). The problem remains open.

Background

A synthetic aperture radar (SAR) is a coherent side-looking radar, flown on a satellite or aircraft, that achieves fine along-track (azimuth) resolution by exploiting the Doppler phase history of the backscattered signal as the platform moves, rather than by carrying a physically long antenna; the across-track coordinate is called range. When a SAR images the ocean, the return is dominated by Bragg scattering - resonant backscatter from centimetre-scale surface waves - so that longer ocean waves imprint themselves as modulations of the returned power. The central statistical description of the sea state is the two-dimensional directional ocean-wave spectrum: the variance spectrum of the sea-surface elevation as a function of the wave-number vector. Recovering it globally from space has been a goal of satellite oceanography since the 1970s; the first civilian spaceborne SAR, Seasat (1978), generated such high data rates that the unprocessed signal had to be transmitted in real time to line-of-sight ground stations or recorded on board.

Hasselmann (1980) attacked the data problem at its root. If the scientific product is not an image but the two-dimensional variance spectrum of the SAR image, then the two-step chain - form the image by matched filtering (correlating the return with the transmitted chirped signal), then Fourier-transform the image - can be contracted algebraically into a single quadratic transform of the raw return signal. The resulting signal-image-Fourier-transform (SIFT) algorithm computes the image spectrum as a twofold integral over the product of the complex return signal and its time-lagged complex conjugate, the two lags being proportional to the azimuth and range wave-number components. Because this reduces the image to its statistical variance spectrum - a massive data compression - the author suggested it could run in real time on board a satellite, opening global wave spectroscopy without a costly telemetry and ground-station network. Physically, the SIFT processor was interpreted as the SAR acting as a continuous two-dimensional dual-frequency (Δk\Delta k) scatterometer: an instrument that probes a prescribed ocean wave number by mixing two transmitted frequencies, whereas the chirped SAR pulse synthesizes the full spectrum of difference frequencies within a single pulse.

The paper itself, however, was explicit that this delivers the SAR image spectrum, not the true surface-wave spectrum. The mapping from the moving sea surface into a SAR image is nonlinear: long gravity waves modulate the Bragg-wave backscatter through tilt and hydrodynamic (wave-current interaction) modulation transfer functions - frequency-dependent proportionality factors between wave slope and image modulation - and, dominantly, through velocity bunching: the orbital velocities of the long waves Doppler-shift the return, displacing scatterers in azimuth by amounts that can reach a significant fraction of the wavelength. Hasselmann (1980) therefore adopted the working assumption that the image variance spectrum carries the useful information and did not address the inversion problem - how to infer the actual wave spectrum when the mapping is strongly nonlinear. The paper also raised, as matters needing further technical analysis, the relative operational merits of the SAR-SIFT route versus a purpose-built two-dimensional Δk\Delta k-scatterometer, the technical realization of real-time on-board processing, and the validity of the contracted formulation in geometrically distorted regimes (image scales smaller than the pulse footprint, or slant-to-horizontal-range distortion that is not weak, as for aircraft SAR at small incidence angles); these operational questions were addressed or bypassed by the 1981-1984 follow-up literature and lie outside the open core.

The subsequent literature built the now-standard theory of this nonlinear mapping (the 'Hasselmann transform'): the SAR image spectrum equals the quasi-linear, tilt- and hydrodynamic-modulated spectrum multiplied by a Gaussian azimuthal cutoff factor whose scale is set by the root-mean-square azimuthal displacement of scatterers, a factor that suppresses waves traveling near the azimuth direction with wavelengths below the cutoff. Iterative inversions were constructed that start from a first-guess wave spectrum (typically from a numerical wave model) and adjust it until the forward-mapped spectrum matches the observed SAR image spectrum, or the image cross-spectrum - the cross-spectrum between two SAR 'looks' (sub-aperture images formed from separate parts of the antenna aperture, whose non-independent speckle cancels in the cross-product while the wave-induced modulation retains a phase related to wave propagation), which resolves the 180-degree propagation-direction ambiguity inherent to power spectra. Coherent SAR images of the sea are additionally corrupted by speckle: the multiplicative fading noise caused by interference of the many scatterers within one resolution cell. Despite these developments, the inversion in the strongly nonlinear regime remains first-guess-dependent, and the azimuth-cutoff suppression has never been inverted away or rigorously proven intrinsic.

Problem Statement

Given SAR measurements of the ocean surface - the raw return signal or the formed image - determine the actual two-dimensional ocean surface-wave spectrum, not merely the SAR image variance spectrum, when the sea-surface-to-SAR-image mapping is strongly nonlinear.

This is the open core motivated by Hasselmann (1980), whose SIFT algorithm addressed only the first stage (direct extraction of the SAR image spectrum from the return signal) and presented global surface-wave spectral retrieval as the possibility this compression opens; the required inversion of the nonlinear mapping was not provided. The objective is to recover the true two-dimensional ocean surface-wave spectrum, including the azimuth-cutoff-suppressed azimuth-traveling components, from a single SAR acquisition alone, with no independent wave-model first guess and under an explicit speckle and thermal-noise model, for sea states in which the sea-surface-to-SAR-image mapping is strongly nonlinear.

The relative-instrument question (SAR-SIFT versus a purpose-built two-dimensional Δk\Delta k-scatterometer), the real-time on-board implementation question raised in the same paper, and the validity of the contracted signal-domain formulation in geometrically distorted regimes were substantially addressed or bypassed by the 1982-1984 follow-up literature (Alpers and Hasselmann; LaHaie et al.) and by later on-board recording technology; these questions lie outside the open core.

References

  1. Hasselmann, K. (1980). A simple algorithm for the direct extraction of the two-dimensional surface image spectrum from the return signal of a synthetic aperture radar. International Journal of Remote Sensing 1(3), 219-240. DOI: 10.1080/01431168008948234. https://doi.org/10.1080/01431168008948234
  2. Martin, Philippe J. (1981). Direct Determination of the Two-Dimensional Image Spectrum from Raw Synthetic Aperture Radar Data. IEEE Transactions on Geoscience and Remote Sensing GE-19(4), 194-203. DOI: 10.1109/TGRS.1981.350372. https://doi.org/10.1109/TGRS.1981.350372
  3. Alpers, Werner; Hasselmann, Klaus (1982). Spectral signal to clutter and thermal noise properties of ocean wave imaging synthetic aperture radars. International Journal of Remote Sensing 3(4), 423-446. DOI: 10.1080/01431168208948413. https://doi.org/10.1080/01431168208948413
  4. LaHaie, I.; Dias, A.; Darling, G. (1984). Digital processing considerations for extraction of ocean wave image spectra from raw synthetic aperture radar data. IEEE Journal of Oceanic Engineering 9(2), 114-120. DOI: 10.1109/JOE.1984.1145607. https://doi.org/10.1109/JOE.1984.1145607
  5. Lyzenga, David R. (1988). An analytic representation of the synthetic aperture radar image spectrum for ocean waves. Journal of Geophysical Research: Oceans 93(C11), 13859-13865. DOI: 10.1029/JC093IC11P13859. https://doi.org/10.1029/JC093IC11P13859
  6. Brüning, Claus; Alpers, Werner; Hasselmann, Klaus (1990). Monte-Carlo simulation studies of the nonlinear imaging of a two dimensional surface wave field by a synthetic aperture radar. International Journal of Remote Sensing 11(10), 1695-1727. DOI: 10.1080/01431169008955125. https://doi.org/10.1080/01431169008955125
  7. Hasselmann, Klaus; Hasselmann, Susanne (1991). On the nonlinear mapping of an ocean wave spectrum into a synthetic aperture radar image spectrum and its inversion. Journal of Geophysical Research: Oceans 96(C6), 10713-10729. DOI: 10.1029/91JC00302. https://doi.org/10.1029/91JC00302
  8. Krogstad, Harald E. (1992). A simple derivation of Hasselmann's nonlinear ocean-synthetic aperture radar transform. Journal of Geophysical Research: Oceans 97(C2), 2421-2425. DOI: 10.1029/91JC03010. https://doi.org/10.1029/91JC03010
  9. Engen, G.; Johnsen, H. (1995). SAR-ocean wave inversion using image cross spectra. IEEE Transactions on Geoscience and Remote Sensing 33(4), 1047-1056. DOI: 10.1109/36.406690. https://doi.org/10.1109/36.406690
  10. Szeto, A. M. K. (1995). On the direct determination of reflectivity spectrum from synthetic aperture radar data. IEEE Transactions on Geoscience and Remote Sensing 33(1), 224-228. DOI: 10.1109/36.368204. https://doi.org/10.1109/36.368204
  11. Hasselmann, S.; Brüning, C.; Hasselmann, K.; Heimbach, P. (1996). An improved algorithm for the retrieval of ocean wave spectra from synthetic aperture radar image spectra. Journal of Geophysical Research: Oceans 101(C7), 16615-16629. DOI: 10.1029/96JC00798. https://doi.org/10.1029/96JC00798
  12. Heimbach, Patrick; Hasselmann, Klaus (2000). Development and application of satellite retrievals of ocean wave spectra. In: Satellites, Oceanography and Society (ed. D. Halpern). Elsevier Oceanography Series, Vol. 63, pp. 5-33. DOI: 10.1016/S0422-9894(00)80003-3. https://doi.org/10.1016/S0422-9894(00)80003-3
  13. Schulz-Stellenfleth, J.; Lehner, S.; Hoja, D. (2005). A parametric scheme for the retrieval of two-dimensional ocean wave spectra from synthetic aperture radar look cross spectra. Journal of Geophysical Research: Oceans 110(C5), C05004. DOI: 10.1029/2004JC002822. https://doi.org/10.1029/2004JC002822
  14. Grelier, T.; Amiot, T.; Tison, C.; Delaye, L.; Hauser, D.; Castillan, P. (2016). The SWIM instrument, a wave scatterometer on CFOSAT mission. 2016 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 5793-5796. DOI: 10.1109/IGARSS.2016.7730513. https://doi.org/10.1109/IGARSS.2016.7730513
  15. Wan, Yong; Zhang, Xiaoyu; Dai, Yongshou; Li, Ligang; Qu, Xiaojun (2020). Azimuth Cutoff Compensation Method for SAR Wave Observation Based on Multiview Wave Spectrum Data Fusion. IEEE Access 8, 120923-120935. DOI: 10.1109/ACCESS.2020.3006244. https://doi.org/10.1109/ACCESS.2020.3006244
  16. Wu, Ke; Li, Xiao-Ming (2023). A Deep Learning-Based Paradigm for Retrieving Ocean Wave Spectra from Spaceborne SAR. 2023 SAR in Big Data Era (BIGSARDATA), 1-4. DOI: 10.1109/BIGSARDATA59007.2023.10294734. https://doi.org/10.1109/BIGSARDATA59007.2023.10294734
  17. Fang, Yuxin; Zhang, Xi; Cao, Rui; Zhang, Jie; Yan, Qiushuang (2025). Physics-Guided Deep Learning for Two-Dimensional Wave Spectrum Retrieval From Synthetic Aperture Radar. SSRN preprint. DOI: 10.2139/ssrn.5155782. https://doi.org/10.2139/ssrn.5155782
  18. Fang, Yuxin; Cao, Rui; Jiang, Haoyu; Meng, Junmin; Zhang, Jie; Yan, Qiushuang (2025). On the Inverse Nonlinear Mapping of a SAR Image Spectrum into an Ocean Wave Spectrum Approximated by the Data-Driven Strategy. TechRxiv preprint. DOI: 10.36227/techrxiv.175303739.96364089/v1. https://doi.org/10.36227/techrxiv.175303739.96364089/v1
  19. Fang, Yuxin; Fan, Chenqing; Meng, Junmin; Wang, Xiaoqing; Zhang, Jie; Yan, Qiushuang (2026). Near-real-time two-dimensional wave spectrum retrieval from Gaofen-3 wave mode imagettes. Fourth International Conference on Environmental Remote Sensing and Geographic Information Technology (ERSGIT 2025), Proceedings of SPIE. DOI: 10.1117/12.3098807. https://doi.org/10.1117/12.3098807