Open Access
Add to BibSonomy
Abstract
The use of thermal remote sensing for marine green tide monitoring has not been clearly demonstrated due to the lack of high-resolution spaceborne thermal observation data. This problem has been effectively solved using high-spatial resolution thermal and optical images collected from the sensors onboard the Ziyuan-1 02E (ZY01-02E) satellite of China. The characteristics and principles of spaceborne thermal remote sensing of green tides were investigated in this study. Spaceborne thermal cameras can capture marine green tides depending on the brightness temperature difference (BTD) between green tides and background seawater, which shows a positive or negative BTD contrast between them in the daytime or nighttime. There is a significant difference between thermal and optical remote sensing in the ability to detect green tides; compared with optical remote sensing, pixels containing less algae are not easily distinguishable in thermal images. However, there is a good linear statistical relationship between the BTD and the optical parameter (scaled algae index of virtual baseline height of floating macroalgae, SAI(VB)) of green tides, which indicates that the BTD can be used to quantify the green tide coverage area in a pixel or biomass per area. Then, the uncertainty in thermal quantitative remote sensing of green tides was clarified according to the pixel-to-pixel relationship between optical and thermal images. In a mixed pixel, green tide coverage and algal thickness have different thermal signal responses, which results in this uncertainty. In future research, more thermally remotely sensed images with high spatial resolution are needed to increase the observation frequency in the daytime and nighttime for the dynamic monitoring of green tides.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Floating macroalgae (e.g., Ulva prolifera or Sargassum) found globally on ocean surfaces are important targets of ocean color remote sensing [1–3]. Green tides, caused by Ulva (U.) prolifera, have occurred annually in the Yellow Sea of China since 2008 [4–6]. In addition, human activities (such as floating raft farming in the coastal zone of Jiangsu Province) and eutrophication in oceans caused by terrigenous nutrients (such as nitrogen or phosphorus) result in the outbreak of green tide algal blooms [7]. Satellite remote sensing of green tides is of great significance to marine environment monitoring, floating raft farming management, disaster prediction, and warning as it can provide accurate, long-time series, and wide-range green tide data [5]. Timely detection and quantification is key in the research on green tide monitoring, and the resulting response.
Many remote sensing techniques have been used for this purpose, including optical remote sensing, synthetic aperture radar, and thermal remote sensing [8–10]. Among these techniques, optical remote sensing is the most frequently used owing to its high revisit cycles and quantification capability, however it is still subject to weather conditions. Multi-source remote sensing is expected to meet the application requirements for real-time dynamic monitoring of green tides. Thermal remote sensing is considered to be another effective technique for green tide monitoring, however the thermal remote sensing mechanism for the detection and quantification of green tides has not been clarified to date, owing to the lack of high-spatial-resolution spaceborne thermal images.
The Ziyuan01-02E (ZY01-02E) satellite of China was launched on December 26, 2021, and is equipped with a high spatial resolution long-wavelength infrared scanner (IRS), a visible near-infrared camera (VNIC), and an advanced hyperspectral imager (AHSI). The additional parameters of these sensors are listed in Table 1. ZY01-02E was designed to explore natural resources, provide alert for environmental disasters, and monitor water quality over an eight-year lifespan in Earth’s orbit. Thermal and optical sensors onboard the satellite can obtain simultaneous thermal and optical images with a spatial resolution of 16 m and 10 m, respectively. These high-spatial-resolution images can effectively reduce the scale effect in the coarse-resolution remote sensing of green tides [11–13], so as to investigate the satellite thermal remote sensing of green tides.
Table 1. Parameters of the ZY01-02E various sensors
In this study, two sets of IRS and VNIC daytime images and a nighttime thermal image of the YS near China were collected for June 18 and 24, 2022 to investigate thermal remote sensing of green tides. The rational and method applied in this study refers to previous studies of thermal remote sensing of oil spills [14–15]. A scaled algae index (SAI) was employed to successfully detect algal pixels from thermal and optical images respectively [16–18]. The detection differences between the optical and thermal remote sensing images show that green tides which cannot be detected by IRS but are detected by VNIC during the daytime are mainly affected by the mixing of algae and seawater. The brightness temperature difference (BTD) between algal and algae-free seawater has a linear statistical relationship with the optical parameter derived from the VNIC, which includes the biomass per area (BPA). This demonstrates that thermal remote sensing can be used to quantify green tides during the daytime. The different abilities of thermal remote sensing of green tides in the daytime and nighttime have also been verified. Based on the results, spaceborne thermal remote sensing of green tides was used to detect and quantify green tides during the day and night.
2. Data and methods
2.1 High-spatial remote sensing data and its processing
2.1.1 VB derived from ZY01-02E VNIC image
The ZY01-02E Level-1 VNIC data have eight bands with central wavelengths located at 425 nm, 485 nm, 555 nm, 605 nm, 660 nm, 725 nm, 830 nm, and 950 nm. The spatial resolution of the VNIC image was 10 m, and they were resampled to 16 m to make them consistent with the spatial resolution of the ZY01-02E thermal image (Fig. 1). All the data were obtained from the China Centre for Resource Satellite Data and Application, and then processed to Top-of-Atmosphere (TOA) reflectance using the following steps (Eqs. (1) and (2)). Data preprocessing and atmospheric correction are based on the preprocessing methods of other Chinese satellites [18–20] as follows: (1) The calibrated total radiance (Ltλ, units: W·m-2·sr-1·µm-1) was derived from the digital number (DN) for Band-1 to 8. (2) TOA reflectance (R, dimensionless) was determined using Eq. (2).
where Gain and Bias are provided in the ZY01-02E metadata, F0 is the extra-terrestrial solar irradiance (units: W·m-2·µm-1), and θ0 is the solar zenith angle. The virtual baseline (VB) height of the floating macroalgae spectra method (Eq. (3)) was employed to enhance the difference between green tides and seawater in the Yellow Sea of China among the ZY01-02E VNIC measurements, especially for satellite sensors with no short-wave infrared bands [10,16].2.1.2 BT derived from ZY01-02E thermal image
The thermal data of the ZY01-02E thermal camera were processed to generate Brightness Temperature (BT, units: K) using Eq. (4), according to Planck’s law (refer to Ref. [15]).
2.2 Scaled algae index
The SAI algorithm was used to detect algal pixels from the VB and BT images. The SAI is an image processing technique in which the high variability of an image is removed [17]. This is done by subtracting the VB or BT index value of a local pixel (pink pixel in Fig. 2) from each pixel in the VB or BT images using the kernel (yellow pixel in Fig. 2) of an odd-numbered pixel square region (201 × 201 pixels in this study, blue solid and dotted boxes in Fig. 2). The index value of the median pixel (blue pixel in Fig. 2) within the square region is computed and subtracted from the index value of the central pixel. This kernel is iterated through all pixels in the BT or VB image, except for cloud and land pixels, to generate the SAI. The SAI equations can be written as Eqs. (5) and (6).
where f(x, y) is the index value of the VB or BT pixel, g(x, y) is the VB or BT value of the median pixel, and SAI(x, y) is the VB or BTD value generated from the SAI images [16].3. Results and discussion
3.1 Spaceborne thermal images capture green tides
Using on the SAI algorithm, the algal pixels within the ZY01-02E thermal and optical images could be identified from the background seawater and clouds. The optical sensor detecting green tides can be used for the simultaneous verification of thermal remote sensing of green tides, and the values of SAI(VB) and BTD of green tides on June 18 and 24 2022 are displayed in Fig. 3. The BTD differences of green tides are not significantly different, however BTDs reached approximately 4°C. This indicates that the floating algae had an obvious warming phenomenon during the daytime (Figs. 3(a) and 3(c)). When compared with the algal pixels detected by the optical remote sensing images (Figs. 3(b) and 3(d)), large quantities of floating algae could not be detected by thermal remote sensing. The main reasons for the difference between thermal and optical remote sensing are as follows. 1) Optical remote sensing of green tides with small SAI(VB) values cannot be detected by thermal remote sensing (Figs. 1 and 3). 2) Green tides in thermal images can also be affected by low temperature clouds, which may not affect the optical remote sensing of green tides (Figs. 1 and 3).
Fig. 1. The thermal and optical RGB images collected from the IRS and VNIC of the ZY01-02E satellite. (a) and (c) BT images over green tides in the YS near to Jiangsu Province on June 18 and 24, 2022, respectively. (b) and (d) the RGB images (R:830 nm, G:660 nm, B:555 nm) corresponding to (a) and (c). The image insets have been zoomed in for comparative analysis. Note that the green tides in thermal image is affected by low temperature clouds (a2 and a3) which did not affect the optical remote sensing of green tides (b2 and b3).
The spatial distribution of green tides derived from the thermal and optical images were determined to show the detection difference among them (Fig. 4). Using the green tides detected by the VNIC as the verification value, the comparative analysis of the green tide spatial distribution shows that thermal remote sensing can only identify a portion of the algal pixels. For example, the number of pixels detected by the thermal camera was 25.5% and 52.2% of the algal pixels detected by the VNIC on June 18 and 24, 2022, respectively. The pixels of green tides detected by thermal and optical images were mainly concentrated in areas with a high SAI(VB) value which implies a larger coverage area and BPA of green tides. This is also demonstrated by the statistical histogram of the algal pixels (Fig. 5). With an increase in the SAI(VB) value derived from optical data, aggregated green tides are readily detected by thermal cameras. Another important factor of the thermal remote sensing of green tides is the environmental background temperature difference between green tides and seawater. Green tides were readily detected by thermal remote sensing on June 24 than on June 18. The SAI(VB) value at the intersection of the statistical histogram curves of the thermal and optically detected algal pixels on June 24 was 0.08, which was less than that on June 18 (0.11).
Fig. 4. Green tides detection differences between the thermal and optical images on June 18 and 24, 2022
Fig. 5. Comparison between green tide detection between thermal camera and VNIC during the daytimes of June 18 (a) and 24 (b)2022, respectively. Their corresponding footprints are shown in Fig. 4.
3.2 Statistical relationship between BTD and SAI(VB) of green tides
Thermal remote sensing can detect green tides, which has been preliminarily demonstrated; however, its use to quantify green tide coverage area or BPA is questioned [11–13]. Although there is a lack of in-situ measured data of green tides to investigate this, optical remote sensing data with the same spatial resolution provides the best cross-validation data. This is because optical remotely sensed data derived parameters, such as VB or floating algae index (FAI), have been widely used to estimate the coverage area or BPA of green tides in the YS of China.
In this study, the VNIC-derived SAI(VB) value was employed as a quantification parameter for green tide coverage area and BPA [16]. To reduce the uncertainty of the regional statistics of BTD and SAI(VB) of green tides, a grid statistical method was used to calculate the average BTD and SAI(VB) of both sets of algal pixels in the thermal and optical images. The area of the statistical grid is 1 km2 and is composed of 100 × 100 pixels in the thermal or optical images. The good linear statistical relationships between the BTD and SAI(VB) of the green tide pixels indicated that spaceborne thermal remote sensing can be used to quantify green tides with high coverage areas or BPA (Fig. 6). The slopes of the linear functions varied (15.24 and 11.59 June 18 and 24, 2022, respectively), which indicates that the ability of thermal remote sensing to quantify green tides also depends on the temperature different between green tides and environmental background.
Fig. 6. Statistical relationship between BTD and SAI(VB) of both green tides from thermal and optical images on June 18 and 24, 2022, respectively. Please note that the slopes of the linear functions are different, larger BTD between green tides and background environment on June 24 (blue line) have high slope than that on June 18 (red line).
The BTD of green tides and seawater also changes with time due to solar radiation during the daytime. This means that the green tide could be detected by the thermal camera during the nighttime, and the green tide would be cooler than the background seawater owing to this negative temperature contrast. The ZY01-02E thermal camera captures the green tide during night time (GMT 14:12 is equal to local time 22:12), which is cooler than that of the background seawater (see the inset thermal image in Fig. 7). Therefore, a schematic diagram of thermal remote sensing of green tides during the day time could be determined. During the positive and negative contrast switching of BTD between the green tide and background seawater at sunrise or sunset, it is difficult to detect green tides by thermal remote sensing (Fig. 7).
Fig. 7. Schematic diagram of profiles of the BT difference as a function of the time of day at a given thicknesses green tides. Please note that the inset thermal image at t0, t1and t2 were collected from thermal camera. The green tide is hotter that the background seawater in the daytime (t0 and t2), and is cooler than the background in the nighttime (t1).
3.3 Uncertainly in thermal quantification of green tides
Although there is a good linear statistical relationship between the BTD and SAI(VB) of green tides, the uncertainty in the BTD of green tides has not been clarified due to the mixed thermal pixels which are affected by the coverage and thickness of green tides. The spatial scale effect and uncertainty of optical remote sensing of green tides has been clarified [11], and similar considerations could be applied to discuss the uncertainty in the thermal quantification of the green tides and seawater mixed pixels. The spatial distribution relationship of pixel-to-pixel between SAI(VB) and BTD of green tides on June 18 and 24, 2022, were determined to show the effect of coverage and thickness on the mixed BTD pixel (Fig. 8). A nearly convex triangular quadrangle can be found in this pixel-to-pixel spatial distribution, particularly in the region of high pixel density (green area in Fig. 8, and were marked with white dotted lines). This indicates that in thermal remote sensing of green tides, mixed thermal pixels are affected by both algal thickness and coverage.
Fig. 8. The spatial distribution relationship of pixel-to-pixel relationship between SAI(VB) and BTD of green tides on June, 18 and 24, 2022 to show the uncertainty in thermal remote sensing of them. Please note that white dotted lines were figured out to show the uncertainly caused by coverage area or BPA of green tides.
The uncertainty in the thermal mixed pixels affected by green tide thickness and coverage is displayed in Fig. 9. For any thermal pixel (16 m×16 m) with a green tide, if it is not completely covered by the green tides, coverage is used to represent the mixing ratio between the algal area and total area in a pixel. The linear relationship between the BTD and SAI(VB) is shown by the blue line in Fig. 9. In addition, the thermal BTD increases with an increase in the area ratio of green tides in the pixel, which also leads to an increase in SAI(VB) in the daytime. If the thermal pixel is completely covered by the green tide, but the green tides vary in thickness, then the thermal BTD will increase with the increase in the thickness of green tides in the pixel. This also leads to an increase in the SAI(VB) in the daytime (green line in Fig. 9). In the above two cases, it should be noted that slope of the linear relationship is different for the BTD with SAI(VB) (green and blue lines in Fig. 8). In fact, both of the above conditions exist together in the spaceborne thermal and optical images, which creates a mixed effect and leads to uncertainty in the thermal quantification of green tides (the red line in Fig. 9, which represents mixed pixels, and the distance between the red line and green or blue line implies uncertainty).
Fig. 9. Schematic graph showing the uncertainty in thermal remote estimation of green tides. Note that the SAI(VB) was used to quantify green tide coverage area or BPA.
4. Conclusion
Thermal remote sensing can provide technical support for detecting green tides over the ocean surface, however the capability and principle of thermal remote sensing of green tides have not been clarified to date due to the lack of high resolution spaceborne thermal images. In this study, simultaneous high-spatial resolution thermal and optical data collected from the ZY01-02E satellite were used to investigate the ability of the thermal camera to monitor and estimate green tides. These data were processed to generate thermal BT and optical VB images. the SAI method was then employed to detect green tide pixels from them to obtain the BTD and SAI(VB) of green tides. Comparative analysis of thermal BTD and optical SAI(VB) of green tides can reveal the ability of thermal remote sensing of green tides. Thermal cameras can effectively monitor green tides, especially for green tides with high SAI(VB) values during the daytime. In addition it can also monitor the green tide at night according to the negative BTD contrast between green tides and seawater. The linear statistical relationship between BTD and SAI(VB) of green tides during the daytime indicates that thermal remote sensing can be used to quantify green tides, and the difference in the linear function slope implies that the thermal quantification ability also depends on the temperature difference between green tides and background seawater. Similar to the optical estimation of green tides, the spaceborne thermal quantification of green tides is also affected by the uncertainty caused by pixel mixing. Pixel-to-pixel green tides between BTD and SAI(VB) images further clarify that the uncertainty in the thermal quantification of green tides mainly comes from both the coverage and thickness of green tides in a pixel. Lastly, although the principle of thermal remote sensing of green tides has been clarified, the effective utilization of thermal remote sensing to improve green tides monitoring, which requires more studies.
Funding
National Natural Science Foundation of China (42176183).
Acknowledgments
We thank China Center for Resources Satellite Date and Application for providing 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. C. Hu, D. Johnson, M. Wang, J.-P. Maréchal, D. Goodwin, and B. Lapointe, “Sargassum Watch Warns of Incoming Seaweed,” EOS Trans. Am. Geophys Union 97(22), 10–15 (2016). [CrossRef]
2. J. Gower, E. Young, and S. King, “Satellite images suggest a new Sargassum source region in 2011,” Remote Sens. Lett. 4(8), 764–773 (2013). [CrossRef]
3. C. Hu and M. He, “Origin and offshore extent of floating algae in Olympic sailing area,” EOS Am. Geophys. Union Trans. 89(33), 302–303 (2008). [CrossRef]
4. D. Liu, J. K. Keesing, P. He, Z. Wang, Y. Shi, and Y. Wang, “The world’s largest macroalgal bloom in the Yellow Sea, China: Formation and implications,” Estuar. Coast. Shelf Sci. 129, 2–10 (2013). [CrossRef]
5. L. Qi, C. Hu, Q. Xing, and S. Shang, “Long-term trend of Ulva prolifera blooms in the western Yellow Sea,” Harmful Algae 58, 35–44 (2016). [CrossRef]
6. L. Qi, C. Hu, K. Mikelsons, M. Wang, V. Lance, S. Sun, B. Barnes, J. Zhao, and D. Zande, “In search of floating algae and other organisms in global oceans and lakes,” Remote Sens. Environ. 239, 111659 (2020). [CrossRef]
7. Y. Zhang, P. He, H. Li, G. Li, J. Liu, F. Jiao, J. Zhang, Y. Huo, X. Shi, R. Su, N. Ye, D. Liu, R. Yu, Z. Wang, M. Zhou, and N. Jiao, “Ulva prolifera green-tide outbreaks and their environmental impact in the Yellow Sea, China,” Natl. Sci. Rev. 6(4), 825–838 (2019). [CrossRef]
8. C. Hu, “A novel ocean color index to detect floating algae in the global oceans,” Remote Sens. Environ. 113(10), 2118–2129 (2009). [CrossRef]
9. H. Yu, C. Wang, J. Li, and Y. Sui, “Automatic Extraction of Green Tide From GF-3 SAR Images Based on Feature Selection and Deep Learning,” IEEE J. Sel. Top Appl. Earth. Obs. Remote Sens 14, 10598–10613 (2021). [CrossRef]
10. Q. Xing and C. Hu, “Mapping macroalgal blooms in the Yellow Sea and East China Sea using HJ-1 and Landsat data: Application of a virtual baseline reflectance height technique,” Remote Sens. Environ. 178, 113–126 (2016). [CrossRef]
11. T. Lu, Y. Lu, L. Hu, J. Jiao, M. Zhang, and Y. Liu, “Uncertainty in the optical remote estimation of the biomass of Ulva prolifera macroalgae using MODIS imagery in the Yellow Sea,” Opt. Express 27(13), 18620–18627 (2019). [CrossRef]
12. L. Hu, C. Hu, and M. He, “Remote estimation of biomass of Ulva prolifera macroalgae in the Yellow Sea,” Remote Sens. Environ. 192, 217–227 (2017). [CrossRef]
13. L. Hu, K. Zeng, C. Hu, and M. He, “On the remote estimation of Ulva prolifera areal coverage and biomass,” Remote Sens. Environ. 223, 194–207 (2019). [CrossRef]
14. Y. Lu, W. Zhan, and C. Hu, “Detecting and quantifying oil slick thickness by thermal remote sensing: a ground-based experiment,” Remote Sens. Environ. 181, 207–217 (2016). [CrossRef]
15. J. Jiao, Y. Lu, C. Hu, J. Shi, S. Sun, and Y. Liu, “Quantifying ocean surface oil thickness using thermal remote sensing,” Remote Sens. Environ. 261, 112513 (2021). [CrossRef]
16. J. Liu, J. Liu, J. Ding, and Y. Lu, “A refined imagery algorithm to extract green tide in the Yellow Sea from HY-1C satellite CZI measurements,” Haiyang Xuebao. 44(5), 1–11 (2022). [CrossRef]
17. J. Keesing, D. Liu, P. Fearns, and R. Garcia, “Inter-and intra-annual patterns of Ulva prolifera green tides in the Yellow Sea during 2007–2009, their origin and relationship to the expansion of coastal seaweed aquaculture in China,” Mar. Pollut. Bull. 62(6), 1169–1182 (2011). [CrossRef]
18. J. Li, X. Chen, L. Tian, and L. Feng, “Tracking radiometric responsivity of optical sensors without on-board calibration systems-case of the Chinese HJ-1A/1B CCD sensors,” Opt. Express 23(2), 1829–1847 (2015). [CrossRef]
19. Q. Zhou, J. Li, L. Tian, et al., “Coupled approach for radiometric calibration and parameter retrieval to improve SPM estimations in turbid inland/coastal waters,” Opt. Express 28(4), 5567 (2020). [CrossRef]
20. J. Li, X. Chen, L. Tian, and L. Feng, “Improved Capabilities of Chinese High-Resolution Remote Sensing Satellite GF-1 in Suspended Particulate Matter (SPM) Monitoring of Inland Waters: Spectral and Spatial Considerations,” ISPRS J. Photogramm. Remote Sens. 106, 145–156 (2015). [CrossRef]
