Open Access
Add to BibSonomy
Abstract
Ultraviolet Imager (UVI) onboard Haiyang-1C/D (HY-1C/D) satellites has been providing ultraviolet (UV) data to detect marine oil spills since 2018. Although the scale effect of UV remote sensing has been preliminarily interpreted, the application characteristics of spaceborne UV sensors with medium spatial resolution in oil spill detection deserve further investigation, especially the role of sunglint in the process of detection. In this study, the performance of the UVI is thoroughly assessed by the following aspects: image features of oils under sunglint, sunglint requirement for spaceborne UV detection of oils, and the stability of the UVI signal. The results indicate that in UVI images, it is sunglint reflection that determines the image features of spilled oils, and the appearance of sunglint can strengthen the contrast between oils and seawater. Besides, the required sunglint strength in spaceborne UV detection has been deduced to be 10−3 - 10−4 sr-1, which is higher than that in the VNIR wavelengths. Moreover, uncertainties in the UVI signal can meet the demand to discriminate between oils and seawater. The above results can confirm the capability of the UVI and the critical role of sunglint in spaceborne UV detection of marine oil spills, and provide new reference for spaceborne UV remote sensing.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Haiyang-1C (HY-1C) and Haiyang-1D (HY-1D) are the first operational ocean color satellites of China. Both of them are equipped with three identical imaging sensors, including Coastal Zone Imager (CZI), Chinese Ocean Color and Temperature Scanner (COCTS), and Ultraviolet Imager (UVI). Particularly, the UVI onboard collects ultraviolet (UV) remote sensing data with relatively high signal-to-noise ratio (SNR) and large radiative dynamic range for global marine environment observation at 355 and 385 nm. Quality of UVI images has been proved reliable for the accuracy of both radiometric and geometric calibration [1–4]. These novel and valuable UV data can provide a wide range of applications for oceanic research, such as studying UV characteristics of water leaving radiance [5], estimating suspended sediment distribution [6], detecting marine oil spills [7], and monitoring harmful algae blooms (HABs) [8].
Oil spills in ocean, mainly including emulsified oil slicks and oil emulsions, pose a great threat to the marine and coastal environment [9]. Timely and accurate remote sensing data about spilled oils (e.g., oil types, oil concentration, oil volumes and oil slick thickness) is crucial for the emergency treatment and recovery. For optical remote sensing of spilled oils, sunglint reflection is important as it can strengthen the contrast between oils and seawater in VNIR wavelength range [10–12], and the role of sunglint in UV detection is also worth study. So far, the feasibility and scale effect of UV remote sensing of marine oil spills have been demonstrated from different platforms (Fig. 1) [7], [13–15]. Spilled oils show high UV reflectivity due to the enhancement-refection effect of the surface interference light, thus appearing positive contrast (brighter) with seawater in ground-based observation and airborne images with high spatial resolution (Figs. 1(a) and 1(b)). Moreover, thin oil slicks are notably sensitive to the UV radiance, and the detected spilled area is larger in the UV band than in the VNIR wavelengths [7], [13–15]. However, this is not recommended for spaceborne platforms. In UVI images with coarse spatial resolution, the slight difference caused by interference light cannot be discriminated, and the image features of oils are still affected by sunglint reflection (Fig. 1(c)). These have been verified by AVIRIS images of the Deepwater Horizon (DWH) oil spill in the Gulf of Mexico (GoM), and one oil spill incident near Indonesia detected by UVI [7].
Fig. 1. Scale effect of UV remote sensing of spilled oils [7]. (a) Experimentally obtained UV image of thin oil slicks (central wavelength: 385 nm). (b) AVIRIS UV images (central wavelength: 380.21 nm) of the spilled oils in GoM in 2010, with the respective RGB image (R: 638.2 nm; G: 550.3 nm; B: 472.5 nm) in the inset image. (c) HY-1C UVI image (355 nm) of the oil spill near Indonesia, synchronous CZI image (R: 650 nm; G: 560 nm; B: 460 nm) is in the inset image.
Although spaceborne UV remote sensing has preliminarily shown its potential in marine oil spill detection, the application characteristics of it deserve further research, especially the role of sunglint in UV remote sensing. Therefore, the main objective of this work is to assess the performance of UVI in oil spill detection under different sunglint conditions. Besides, the minimum sunglint strength requirement for UVI detection is deduced based on Cox-Munk model [16]. Moreover, the uncertainties in detection caused by the sensor itself and atmospheric effect are counted and analyzed to evaluate the feasibility of UVI for the identification and quantitative estimation of spilled oils. This work is not to compare the advantages of spaceborne UV and VNIR remote sensing to detect spilled oils, but to deepen the understanding of sunglint in the UV band, and provide new reference for spaceborne UV remote sensing.
2. Data and methods
2.1 HY-1C/D UVI images and synchronous validation data
HY-1C/D satellites are operating in a sun-synchronous orbit. The three imaging sensors onboard HY-1C/D, including CZI, UVI and COCTS, cover the wavelength range from UV to VNIR band, ensuring a global coverage twice a day [7], [8], [17]. The designed data spatial resolution at the subastral point is ∼50 m for CZI, and ∼1100 m for UVI and COCTS. It should be noted that the three sensors work synchronously, and the synchronous data have identical observation geometry (including the solar/sensor zenith angle and the solar/sensor azimuth angle).
HY-1C/D satellites detected several oil spills in China Seas since 2018. Some groups of synchronous HY-1C/D images with identical observation geometry are presented in Fig. 2. Oil spills showed presence in all CZI and COCTS images, while in UVI images they were detected under certain condition. Specifically, Level-1 (spectral radiance; Lt; unit: mW·cm-2·µm-1·sr-1) and Level-2 data (Rayleigh-corrected reflectance, Rrc; dimensionless) of HY-1C/D were adopted for analysis in this study.
Fig. 2. HY-1C/D satellite images of oil spills. Oils were detected and annotated with red arrows in all CZI (R: 650 nm; G: 560 nm; B: 460 nm) and COCTS (R: 670 nm; G: 565 nm; B: 443 nm) RGB images. In UVI images (385 nm), the detected oils are also annotated with red arrows, while the location of the undetected oils are annotated with red boxes. Date of the images the images are annotated on the left, for example, 02/20/2019 refers to February 20, 2019.
2.2 Simulated LGN_oil using Cox-Munk model
Cox-Munk model, taking solar/viewing geometry and wind speed into consideration, was developed for interpreting the strength of sunglint [11], [12], [16], [18–20]. Based on this model, sunglint strength of the spilled area is defined as the normalized sunglint reflectance (LGN_oil, unit: sr-1), and can be calculated as
Another effective indicator to evaluate sunglint strength in satellite images is θm, referring to the angle between viewing direction and the direction of mirror reflection [10]
The potential critical angle range is between 12° - 13° for both UV and VNIR wavelengths in medium spatial resolutioimages, on either side of which oil slicks could appear positive (brighter) (θm ≤ 12°) or negative (darker) contrast (θm ≥ 14°) with seawater in images [7], [10], [11]. Moreover, oil emulsions would appear positive contrast with seawater under any sunglint condition [7,21].
2.3 Uncertainty evaluation
Uncertainties in detection of spilled oils will be studied using the spectral radiance data (avoiding the influence of atmospheric correction) to assess the ability of UVI to identify oil spills. In UVI images, the oil-containing pixels were extracted by threshold segmentation. Besides, 10 × 10 pixels of oil-free seawater were extracted to evaluate the background noise of the sensor. Here, the metd illustrated by Hu [22] was adopted
3. Results and discussn
3.1 Oil spill detection under sunglint in UVI images
Figure 3 shows 6 examples of HY-1C/D observation. Based on the spectral response characteristics and sunglint assessment, spilled oils were detected in all CZI images. Case 1 to 4 show cases where spilled oils were also detected in synchronous UVI images. The scale effect in UV remote sensing of oil spills has been proposed in previous studies [7]. For spaceborne UV sensors with coarse spatial resolution (i.e., UVI) which cannot discriminate the interference light, it is still the surface sunglint reflection that determines the image features of spilled oils (appearing brighter or darker than background seawater), in accordance with the VNIR wavelength corresponding θm values are annotated in each image, and all of them are larger than the potential critical angle range (12° - 13°). So, in these cases, spilled oils all appeared darker than background seawater under weak sunglint.
Fig. 3. Cases of HY-1C/D observation of oil spills. (a) Classification maps based on CZI observations. (b) CZI Rayleigh-corrected reflectance spectra (Rrc). (c) and (d) UVI images (355 and 385 nm) with spectral radiance (Lt) legend overlaid. Location of the undetected oils are annotated with red boxes.
The appearance of spilled oils in UVI images can be proved by the statistical anomaly of Lt values derived from the profile lines over oil pixels in Fig. 3, and is defined as ΔLt
Fig. 4. ΔLt values derived along the profile lines in Fig. 2, where oil spills are indicated. The spilled oil was undetected at 355 nm on February 20, 2019.
3.2 Sunglint requirement for spaceborne ultraviolet detection of spilled oils
In case 5 and 6 of Fig. 3, the spilled oils were only detected by CZI and COCTS, but undetected by UVI. UVI and COCTS have quite close spatial resolution and identical observation geometry, so it can be considered that the length and area of spilled oils in these cases also satisfy the minimum requirement of spatial resolution for UVI, and the sunglint strength in COCTS images is the same as that in UVI images. The undetectability can be associated with sunglint reflection. Here, two problems need to be solved: (1) can spilled oils still be detected in UVI images under weak sunglint; (2) if can, what is and how to determine the minimum sunglint strength for oil spill detection in UVI images.
HY-1C/D satellites have detected 57 oil spill incidents from 2018 to 2021. 35 of them can be detected by COCTS (including the cases in Fig. 2), however, only 13 of them can also be detected in synchronous UVI images. Here, we counted the sunglint reflection state of the COCTS-detected oils, and the statistical results are shown in Fig. 5. It is clear that data can be divided into two groups, and the LGN_oil or θm values to separate the groups can be regarded as the boundary threshold for detection. In this study, the LGN_oil threshold is 10−3 - 10−4 sr-1 for UVI and 10−4 - 10−5 sr-1 for COCTS. Sunglint requirement of the UV band is apparently higher than the VNIR wavelengths. Below the lower bound for UVI, around 90% of the spilled oils cannot be detected in UVI images, which is an interesting boundary condition, especially given that the SNR of UVI (1000) is higher than COCTS (400). This issue may result from the strong atmospheric effect to UV radiance, scattering is strong and transmittance is low in the UV band. For UVI images, the absence of oil spills can only be demonstrated when oils are undetected and LGN_oil is above the threshold, and images with LGN_oil below the threshold should not be regarded as valid data for oil spill detection. Moreover, the θm threshold is 26° for UVI and 39° for COCTS. It should be noted that the above results are deduced for UVI images, but not applicable for UV images with high spatial resolution where the surface interference light can be discriminated.
Fig. 5. (a) Statistics of sunglint strength (LGN_oil) where the spilled oils exist and can or cannot be detected in UVI images. (b) Statistics of θm where the spilled oils exist and can or cannot be detected in UVI images. The red and blue dotted lines indicate the LGN_oil or θm thresholds for UVI and COCTS, respectively. The black dotted line indicates the potential critical angle.
3.3 Uncertainties in HY-1C/D UVI images
Uncertainties of UVI-measured spectral radiance are shown in Fig. 6. The error bars inside represent the standard deviation (Γ) in Eq. (6), indicating the statistical distribution around the mean spectral radiance (Lt_mean). It is clear that the Γ values of seawater are noticeably low, and the minimum Lt values of seawater are larger than the maximum Lt values of spilled oils in all UVI bands. The signal of UVI is hence relatively stable, and UVI can effectively discriminate between oil spills and seawater. However, the existence of sunglint would strengthen the oil-water contrast but also increases the uncertainties of both oil and seawater pixels. Uncertainties (γ) caused by background noise and atmospheric effect can make a contribution of at least 30% on variation of oils’ signal measured by UVI under any sunglint condition. For pixels of oils, the numerical fluctuation may result from the variation of oil concentration and oil slick thickness, but the high uncertainty level indicates that UVI can hardly be applied for quantitative estimation of oil spills.
Fig. 6. Mean spectral radiance (Lt_mean) derived from UVI images in Fig. 3. Here H and L refer to the high and low dynamic range respectively. Uncertainties in discriminating between spilled oils and seawater are shown in the inset images. Note that the data of 355 nm on February 20, 2019 are vacant, as spilled oil was undetected in this band on that day.
4. Conclusion
The remote detection and quantification of oil spills is a key research direction in the monitoring of marine environment, and spaceborne UV remote sensing has shown its potential in oil spill detection. However, the application characteristics of it deserve further investigation, and the performance of HY-1C/D UVI, which provides the novel spaceborne UV data, has yet to be thoroughly studied. This work assessed the performance of UVI from the following aspects: (1) features of spilled oils in UVI images under different sunglint conditions; (2) sunglint requirement for detecting oil spills in UVI images; (3) uncertainties in UVI images. The results indicate that sunglint reflection determines the features of spilled oils in UVI images with coarse spatial resolution, so oils and seawater appear negative contrast under weak sunglint, and the contrast is larger with stronger sunglint. In addition, the required LGN_oil threshold for oil spill detection for UVI is 10−3 - 10−4 sr-1 (higher than the VNIR wavelengths), which is an important boundary condition for the interpretation of sunglint in spaceborne UV images. Moreover, uncertainty evaluation based on pixels of oils and oil-free seawater indicates that the signal of UVI is stable and reliable for identification of oil spills. Therefore, UVI has been proved to be applicable for oil spill detection, and the sunglint reflection plays a critical role in it, which can facilitate the performance of spaceborne UV sensor by influencing the detectability and oil-water contrast. Last but not least, these results would provide reference for the design and development of the new generation spaceborne UV sensors, especially the ones with medium spatial resolution. Considering the importance of sunglint (associated with the observation geometry) in spaceborne UV detection of spilled oils and the less sensitivity of UV band to sunglint comparing with the VNIR wavelengths, the future spaceborne UV sensor should further improve its SNR, and the atmospheric effect to the UV band deserves further study. We hope our work can stimulate further investigation by our community in the application of spaceborne UV remote sensing.
Funding
National Key Research and Development Program of China (2021YFC2803301); National Natural Science Foundation of China (42071387); Strategic Priority Research Program of the Chinese Academy of Sciences (XDA19090123, XDA19090140); Dragon 5 Cooperation Programme (59193).
Acknowledgments
We thank NSOAS (http://www.nsoas.org.cn/) for providing HY-1C/D data.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. H. Xu, L. Zhang, W. Huang, W. Xu, X. Si, H. Chen, X. Li, and Q. Song, “Onboard absolute radiometric calibration and validation of the satellite calibration spectrometer on HY-1C,” Opt. Express 28(20), 30015–30034 (2020). [CrossRef]
2. Q. Song, C. Ma, J. Liu, X. Wang, Y. Huang, G. Liu, and Z. Li, “Time series analysis-based long-term onboard radiometric calibration coefficient correction and validation for the HY-1C Satellite Calibration Spectrometer,” Remote Sen. 14(19), 4811 (2022). [CrossRef]
3. J. Cao, Z. Zhang, and S. Jin, “Sensor orientation of HaiYang-1C ultraviolet imager based on a piecewise rational function model,” J. Appl. Rem. Sens. 14(01), 1 (2020). [CrossRef]
4. J. Cao, Z. Zhang, S. Jin, and X. Chang, “Geometric stitching of a HaiYang-1C ultra violet imager with a distorted virtual camera,” Opt. Express 28(9), 14109–14126 (2020). [CrossRef]
5. R. Bai, X. He, Y. Bai, T. Li, and F. Gong, “Characteristics of water leaving reflectance at ultraviolet wavelengths: radiative transfer simulations,” Opt. Express 28(20), 29714–29729 (2020). [CrossRef]
6. L. Cai, S. Chen, X. Yan, Y. Bai, and J. Bu, “Study on high-resolution suspended sediment distribution under the influence of coastal zone engineering in the Yangtze River Mouth, China,” Remote Sen. 14(3), 486 (2022). [CrossRef]
7. Z. Suo, Y. Lu, J. Liu, J. Ding, D. Yin, F. Xu, and J. Jiao, “Ultraviolet remote sensing of marine oil spills: a new approach of Haiyang-1C satellite,” Opt. Express 29(9), 13486–13495 (2021). [CrossRef]
8. Z. Suo, Y. Lu, J. Liu, J. Ding, Q. Xing, D. Yin, F. Xu, and J. Liu, “HY-1C ultraviolet imager captures algae blooms floating on water surface,” Harmful Algae 114, 102218 (2022). [CrossRef]
9. I. Leifer, W. J. Lehr, D. S. Beatty, E. Bradley, R. Clark, P. Dennison, Y. Hu, S. Matheson, C. E. Jones, B. Holt, M. Reif, D. A. Roberts, J. Svejkovsky, G. Swayze, and J. Wozencraft, “State of the art satellite and airborne marine oil spill remote sensing: Application to the BP Deepwater Horizon oil spill,” Remote Sens Environ. 124(9), 185–209 (2012). [CrossRef]
10. C. Hu, X. Li, W. G. Pichel, and F. E. Muller-Karger, “Detection of natural oil slicks in the NW Gulf of Mexico using MODIS imagery,” Geophys. Res. Lett. 36(1), L01604 (2009). [CrossRef]
11. Y. Wen, M. Wang, Y. Lu, S. Sun, M. Zhang, Z. Mao, J. Shi, and Y. Liu, “An alternative approach to determine critical angle of contrast reversal and surface roughness of oil slicks under sunglint,” Int. J. Digit. Earth. 11(9), 972–979 (2018). [CrossRef]
12. S. Sun and C. Hu, “Sun glint requirement for the remote detection of surface oil films,” Geophys. Res. Lett. 43(1), 309–316 (2016). [CrossRef]
13. S. Fang, X. Huang, D. Yin, C. Xu, X. Feng, and Q. Feng, “Research of the ultraviolet reflectivity characteristic of simulative targets of oil spill on the ocean,” J. Spectrosc. Spectral Anal. 30(3), 738–742 (2010). [CrossRef]
14. D. Yin, X. Huang, W. Qian, Q. Feng, C. R. Bostater, S. P. Mertikas, X. Neyt, and M. Velez-Reyes, “Airborne validation of a new-style ultraviolet push-broom camera for ocean oil spill pollution surveillance,” Proc. SPIE 7825, 78250I (2010). [CrossRef]
15. P. Wagner, T. Hengstermann, and O. Zielinski, “MEDUSA: An airborne multispectral oil spill detection and characterization system,” Proc. SPIE 4130, 610–620 (2000). [CrossRef]
16. C. Cox and W. Munk, “Measurement of the roughness of the sea surface from photographs of the sun’s glitter,” J. Opt. Soc. Am. 44(11), 838–840 (1954). [CrossRef]
17. X. Ye, J. Liu, J. Ding, and B. Song, “Global ocean Chlorophyll-a concentrations derived from COCTS onboard the HY-1C satellite and their pre-liminary evaluation,” IEEE Trans. Geosci. Remote Sensing 59(12), 9914–9926 (2021). [CrossRef]
18. M. Wang and S. W. Bailey, “Correction of sunglint contamination on the SeaWiFS ocean and atmosphere products,” Appl. Opt. 40(27), 4790–4798 (2001). [CrossRef]
19. C. R. Jackson and W. Alpers, “The role of the critical angle in brightness reversals on sunglint images of the sea surface,” J. Geophys. Res. 115(C9), C09019 (2010). [CrossRef]
20. H. Zhang and M. Wang, “Evaluation of sun glint models using MODIS measurements,” J. Quant. Spectrosc. Radiative Transfer 111(3), 492–506 (2010). [CrossRef]
21. Y. Lu, J. Shi, Y. Wen, C. Hu, Y. Zhou, S. Sun, M. Zhang, Z. Mao, and Y. Liu, “Optical interpretation of oil emulsions in the ocean - Part I: Laboratory measurements and proof-of-concept with AVIRIS observations,” Remote Sens Environ. 230, 111183 (2019). [CrossRef]
22. C. Hu, L. Feng, and Z. Lee, “Uncertainties of SeaWiFS and MODIS remote sensing reflectance: implications from clear water measurements,” Remote Sens Environ. 133, 168–182 (2013). [CrossRef]
