- Email: [email protected]
Tidal-driven variation of suspended sediment in Hangzhou Bay based on GOCI data
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Contents lists available at ScienceDirect
Int J Appl Earth Obs Geoinformation journal homepage: www.elsevier.com/locate/jag
Tidal-driven variation of suspended sediment in Hangzhou Bay based on GOCI data
T
⁎
Yuekai Hua,b, Zhifeng Yua,c,h, , Bin Zhoua,h, Yuan Lid,e, Shoujing Yinf, Xianqiang Heg, Xiaoxue Penga, C.K. Shumc a
Institute of Remote Sensing and Earth Sciences, College of Science, Hangzhou Normal University, Hangzhou, 311121, China State Key Laboratory of Estuarine and Coastal Research, East China Normal University, Shanghai, 200062, China c Division of Geodetic Science, School of Earth Sciences, The Ohio State University, Columbus, OH, 43210, USA d School of Tourism and Urban & Rural Planning, Zhejiang Gongshang University, Hangzhou, 310018, China e Taihu Laboratory for Lake Ecosystem Research, State Key Laboratory of Lake Science and Environment, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing, 210008, China f Satellite Environment Centre, Ministry of Ecology and Environment, Beijing, 100094, China g State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography, State Oceanic Administration, Hangzhou, 310012, China h Zhejiang Provincial Key Laboratory of Urban Wetlands and Regional Change, Hangzhou, 311121, China b
A R T I C LE I N FO
A B S T R A C T
Keywords: Suspended sediment concentration Coefficient of variance Hangzhou Bay GOCI
The variation of suspended sediment concentration (SSC) in coastal waters plays a key role in the marine physical and chemical processes. Hangzhou Bay is a macrotidal estuary in eastern China. Under the influence of trapped tides, the SSC of Hangzhou Bay responds significantly commensurate with tidal frequencies, which leads to considerable changes in the ocean physical and chemical regimes, such as the underwater light field and turbidity front. However, polar-orbiting imaging satellites are unable to monitor the rapid variation of the SSC due to insufficient temporal resolution. In this study, we used the Geostationary Ocean Colour Imager (GOCI) data and in situ data of Hangzhou Bay to develop an appropriate remote sensing model to quantify the SSC of Hangzhou Bay and analyzed the impact of tides on the SSC. The research revealed the following findings: (1) The exponential model based on B8/B6 is best suited to remotely estimating the SSC in Hangzhou Bay, and the Mean Relative Error (MRE) is only 15.21%, (2) the variation of SSC in Hangzhou Bay is related to tidal forcing, especially during the ebb tide and the middle of the rising tide, (3) The effect of tidal driven SSC changes is less than the effect of diurnal tidal level variation of the middle tide day and the spring tide day tidal type, and (4) The maximum variation of SSC under tidal forcing is in the coastal ocean near the south bank of Hangzhou Bay and Nanhui Spit, while the origin of the variation of the Zhapu deep trough and the northern estuary is not obvious, which is speculated to be mainly related to the terrain and runoff of Hangzhou Bay. Furthermore, this study established a new model to map SSC in turbid seas and contributes to our understanding of SSC variation in the Hangzhou Bay, driven by strong tidal dynamics.
1. Introduction The estuary and coastal zones are at the junction of the mainland and coastal ocean. It is also the region where interactions and feedbacks of the Earth’s atmosphere, hydrosphere, lithosphere, biosphere, and biochemical processes are complex and with a wide range of temporal scales (Lyons et al., 1982). The biochemical processes in these zones are complex and variable. On the one hand, sediment transport in the estuary zone is governed by physical processes including land hydrology, ocean circulations, and gravitational interactions between the Earth,
⁎
the Moon and the Sun. Suspended sediment plays an important role in the geomorphological evolution and material circulation of coastal areas due to its characteristics of flocculation, adsorption and sedimentation. However, under rapid urbanization, the estuarine region has been adversely affected by significant human activities and by anthropogenic climate change (Wu et al., 2018). These factors resulted in the degradation of natural ecological environment with worsen water quality, exacerbated by semi-diurnal/diurnal suspended sediment concentration variations and chlorophyll concentration (Eom et al., 2016).
Corresponding author at: Institute of Remote Sensing and Earth Sciences, College of Science, Hangzhou Normal University, Hangzhou, 311121, China. E-mail address: [email protected] (Z. Yu).
https://doi.org/10.1016/j.jag.2019.101920 Received 17 October 2018; Received in revised form 28 June 2019; Accepted 4 July 2019 0303-2434/ © 2019 Elsevier B.V. All rights reserved.
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
source, and after atmospheric correction through the UV-AC model, the best empirical algorithm was used to calculate the hourly GOCI data to study the change in SSC in Hangzhou Bay under the influence of tides. This approach can realize the high-frequency dynamic monitoring of suspended sediment in Hangzhou Bay, and the results demonstrate a method and theoretical basis for studying the migration of SSC as a function of semi-diurnal and diurnal tides and the associated tidal currents.
High concentrations of suspended sediment are regarded as a typical characteristic of coastal waters, along with chlorophyll and yellow material. Suspended sediment affects the photosynthesis of phytoplankton and marine ecosystems by changing the light transmittance of surface water (Byun et al., 2007; Doxaran et al., 2009; Mao et al., 2012). In addition, the sedimentation of suspended sediment changes the characteristics of existing landforms and coastlines in estuarine coastal regions under the influence of changes in tidal intensity and riverine fluxes (Chatanantavet and L·amb, 2014;Yeshaneh et al., 2014). As a significant carrier of nutrients and pollutants, suspended sediment has a direct impact on marine chemistry under the influence of human activities and terrestrial inputs (Zhang, 1999). Therefore, studying the temporal and spatial variations and the distribution of suspended sediment concentration in estuary regions is of great importance in the investigations of water quality, underwater light field, marine ecology, estuary topography and carbon cycle (Schlünz and Schneider, 2000). Hangzhou Bay is one of the strongest tidal regions in China and in the world: the maximum tidal range can reach approximately 9 m, resulting in turbid seawater and producing SSC of up to 5000 mg/L, and the sediment is mainly originated from the Yangtze Estuary (He et al., 2013; Su and Wang, 1989; Xie et al., 2009). Monitoring the change in SCC in coastal water is a research topic significantly linked with severity of ocean water colour. During the past few decades, remote sensing technology has been widely used to map the variation and spatial distribution of SSC in coastal waters, including Landsat (Cai et al., 2015), Coastal Zone Color Scanner (CZCS) (Tang et al., 1998), Sea-viewing Wide Field-of-view Sensor (SeaWiFS) (Fettweis et al., 2007), and Moderate Resolution Imaging Spectroradiometer (MODIS) (Zhang et al., 2014). However, as a macrotidal estuary, Hangzhou Bay has undergone significant changes in SSC, especially under the influence of topography-trapped ocean tides (He et al., 2013), which has led to changes in the physical and chemical characteristics of the ocean, such as the underwater light field, turbidity front, etc. In such an extremely turbid estuary, the time resolution of traditional polar-orbiting satellites has been unable to meet the needs of ocean monitoring. In 2010, the South Korean Ocean Satellite Center (KOSC) launched the geostationary orbit satellite Geostationary Ocean Color Imager (GOCI), which can image 8 scenes a day (Lamquin et al., 2012). The high temporal resolution of the GOCI provides an opportunity to study hourly changes in suspended sediment in Hangzhou Bay. Moreover, the KOSC provided software system to process GOCI data, the GOCI Data Processing Software (GDPS), which integrates functions such as radiation calibration, atmospheric correction, and water colour factor inversion. Due to the spatial variability of ocean colour, the software's atmospheric correction algorithm has obvious errors in turbid water, and its single-band algorithm of SSC is not applicable to Hangzhou Bay. For this reason, Pan et al. (2017) proposed an improved spectral optimization algorithm for atmospheric correction. He et al. (2013) used the assumption of an ultraviolet (UV) “dark pixel” for atmospheric correction of the GOCI, which was useful in Hangzhou Bay (He et al., 2013; Liu et al., 2018; Yang and Mao, 2016). In the remote sensing inversion algorithm, GDPS provided the YOC algorithm to map the high-concentration suspended sediment in a subsequent version (Siswanto et al., 2011). He et al. (2013) established the model of B7 and B3 band ratios to map the SSC of Hangzhou Bay. The above research studies showed that the GOCI geostationary satellite measurements have great potential to advance in the investigation of ocean colour in Hangzhou Bay. However, the suitability of the proposed SSC model with GOCI data should be further validated, especially under tidal action. In the present study, we used in situ water quality data to analyse the spectral differences among collected samples of SSC. We propose a new empirical algorithm according to the analysis results. By comparison with existing algorithms, we chose the best algorithm for our study. Finally, cloudless GOCI data under a complete tidal cycle, for three days of the spring tide, middle tide and neap tide, was selected as the data
2. Materials and methods 2.1. Study area Hangzhou Bay, located in the middle of China's eastern coastline, is a typical funnel-shaped estuary. It extends from Changshan Gate–Xisan Gate in the west to Luwan Port–Waiyou Mount in the east. The tidal variation of the bay is classified as a semidiurnal tide, and the hydrodynamic environment in the bay is complex and significantly affected by ocean tides (Schoellhamer, 2002; Wang et al., 2009). Strong tidal action caused by the resuspension of bottom sediment combines with the Yangtze River diluted water every year to carry a large amount of sediment from the Nanhui Spit near Hangzhou Bay (Su and Wang, 1989). As a result, the SSC in Hangzhou Bay is higher than the SSC farther offshore, and it is also in a highly variable state. 2.2. In situ data Employing a combination of continuous flow measurements and fixed-point observations, our research team carried out water sampling and spectroscopic measurements simultaneously at 31 experimental locations (Fig. 1 right) during December 2nd to 13th, 2011, and May 16th and 22nd, 2014. To ensure the scientific value of the experimental data, the sample times of both samples covered the flood season and dry season as well as the periods of spring, middle and neap tides. During the sampling process, 500 mL of seawater was measured using a water collector and a measuring bottle. At the same time, the upward radiance from the ocean surface, the standard reflecting plate and the downward sky radiance were measured with an ASD handheld spectrometer. The reflectance of each location can be calculated based on these measured data. The water sample was filtered on the spot and stored in a refrigerator. In the laboratory, a crucible was placed in a muffle furnace and kept at 450 °C for two hours. After cooling, the sample was placed in a silica gel drier for two hours and weighed until the weight did not change. Subsequently, the thawed filter was placed in a crucible and the temperature of the muffle furnace was set at 500 °C for two hours, equilibrated with a silica gel dryer and weighed until the weight no longer changed. Finally, the SSC at each location was calculated by the weighing result and the volume of water. Based on the above method, 63 pairs of data (spectral and SSC) were obtained and used in this paper. 2.3. GOCI data GOCI data in this paper were downloaded from the Korea Ocean Satellite Centre (KOSC, http://kosc.kiost.ac.kr/). The observation centre of the GOCI is at 130 °E, 26 °N, and the spatial resolution is 500 m. GOCI data covers the East China Sea and the northern regions of Taiwan Province. The data includes eight bands (412 nm, 443 nm, 490 nm, 555 nm, 660 nm, 680 nm, 745 nm, and 865 nm). The band setting is similar to those of MODIS and SeaWiFS, which is suitable for ocean colour observation. In terms of data selection, studying the change in SSC under the tidal action in Hangzhou Bay required that the data could be covered by the complete tidal period and the cloud coverage in the study area should be less than 10%. We searched the GOCI data starting from 2011 to 2016 and determined three phases of research data that meet the requirements: 2015-07-25 (neap tide), 2
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 1. Study area and the distribution of in situ points.
degree of the object. Therefore, in this paper, the SSC in Hangzhou Bay waters was quantitatively characterized by the CV, and the study evaluated the effect of tidal hour variation on a daily time scale on the SSC in different sea areas of Hangzhou Bay. The formula was as follows:
2015-07-29 (middle tide), and 2015-08-02 (spring tide), for a total of 24 scene images. 2.4. Atmospheric correction and validation Atmospheric correction is one of the key steps in ocean colour remote sensing, and its accuracy is critical for scientific interpretation. In terms of data processing, GOCI Level-1B data provided by KOSC was used as the data source in this paper. The pre-processing of remote sensing data such as clipping, projection transform, geometric correction, was done on the GDPS platform. The atmospheric correction of GOCI data was done using the UVeAC algorithm to obtain the normalized water-leaving radiance. The remote sensing reflectance (Rrs) could be calculated through the normalized water-leaving radiance. Applied UV band atmospheric correction of the turbid water was mainly because it can well estimate the aerosol scattering reflectance of the turbid water, and the UV band has a low value of the water-leaving radiance, which can be ignored(He et al., 2012). For the GOCI data, the 412 nm band can be used instead due to the lack of UV band. The details of UV-AC algorithm can be found in He et al. (2013). In this paper, we used the in situ reflectance field work measured data in the Hangzhou Bay, and the GOCI remotely sensed reflectance measurements from the UV-AC algorithm for satellite and in situ validation studies. The main criteria are as follows (Eom et al., 2016; Froidefond et al., 2002; Roberto et al., 2005; Sá et al., 2015):
CV=
n ¯ )2 /n ∑i = 1 (SSCi − SSC ¯ SSC
(1)
where n is the daily image number of GOCI, SSCi is the suspended se¯ is the diment concentration of a grid of pixels in image i, and SSC average value of suspended sediment concentration on that day on the grid. The root mean square error (RMSE), the absolute percentage difference (APD), and the Mean Relative Error (MRE) were used to measure the performance of the atmospheric correction method and SSC inversion model.
RMSE=
n=i
∑n=1
[(Xi )insitu − (Xi )calculated]2 i
n=i 1 (Xi )calculated − (Xi )insitu APD= 100* *∑ n=1 i (Xi )insitu
MRE=
n=i 1 (Xi )calculated − (Xi )insitu *∑ *100% n = 1 i (Xi )insitu
)insitu
and (Xi Where (Xi or SSC, respectively.
1 Calculate the number of valid pixels (NVP) in the measured pixel space window (3*3). If NVP > 4, use it; otherwise discard it; 2 Calculate the average (AVG) and standard deviation (SD) of valid pixels, remove data other than AVG ± 1.5*SD to avoid the impact of outliers on the results; 3 Recalculate the AVG and SD and calculate the coefficient of variation (CV). If CV ≤ 0.15, it means that the data group meets the spatial uniformity requirements and can be used for subsequent acting; 4 Considering the characteristics of high dynamics in Hangzhou Bay, the time window of the study was set to ± 1 h.
)calculated
(2)
(3)
(4)
represent the in situ and calculated Rrs
3. Model establishment and validation 3.1. Measured spectral data processing and analysis On December 12th, 2011 (spring tide), the measured point (D02) sampling results showed that the sediment content of water was significantly changed in one day (Fig. 2). The maximum daily variation of SSC was 1157.6 mg/L, and the maximum per unit time was 738.2 mg/ L/h. There were 63 samples of measured in situ data in the Hangzhou Bay area (Fig. 1, right), and the SSC varied from 150 mg/L to 2300 mg/ L. In the ocean colour remote sensing of case II waters, the amount of suspended matter affects the optical characteristics of the water spectrum (Binding and Bowers, 2005; Cuuran and Novo, 1988; Novo et al., 1989). According to the different concentration gradients of the
2.5. Data analysis and accuracy assessment When the mean and the variance of study object are not consistent, the coefficient of variance (CV) can be used to measure the dispersion 3
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
derived from the UV-AC algorithm were consistent with the in situ measured results. The APD of all GOCI bands was 23.12 and the RMSE was 0.0083. The APD of each band was 35.1 (412 nm), 32.7 (443 nm), 24.5 (490 nm), 14.0 (510 nm), 17.3 (555 nm), 15.6 (670 nm), 19.9 (765 nm) and 25.8 (865 nm). In this paper, we used the band ratio to calculate the SSC, the APD of B8/B6 was 12.74, RMSE was 0.1082. According to these results, we concluded that the UV-AV algorithm enabled an accurate estimate of the ocean surface reflectance, and that our results are novel.
3.3. Establishment and comparison of the inversion model Fig. 2. Daily variation of SSC (Point D02, 2011-12-12).
The 63 pairs of data measured in Hangzhou Bay were divided into two parts, two-thirds (42 pairs) of which were used for modelling and the other third (21 pairs) to evaluate the error of the model. The differences in B7 and B8 and B5 and B6 provided a new direction for building algorithms to research the SSC of Hangzhou Bay. In this study, we tested all the combinations to select the remote sensing factor and found that B8/B6 achieved the best result in the inversion algorithm; the R2 was 0.91 (Fig. 5). In the process of using GOCI data to map the ocean colour in Hangzhou Bay, the model established by He (2013) and the YOC model provided by GDPS are widely used in current studies (He et al., 2013; Hu et al., 2016; Liu et al., 2018; Yang and Mao, 2016). These models were used to map the total suspended matter. In the process of the inversion of SSC, some parameters might need to be re-set. Therefore, the present research used the modelling data to recalculate the model parameters and used the other data to evaluate the accuracy of the model (Fig. 6). Finally, the best model was selected to map the SSC in Hangzhou Bay. The three formulas are as follows:
measured data, the data were divided into 10 groups. After the average processing of the spectra of each group, 10 water bodies with different sand contents were obtained (Fig. 3a). Then, the study used the spectral response function of the GOCI to convert spectral data to the equivalent remote sensing reflectance of corresponding bands (Fig. 3b). The spectral data showed that the reflectance was positively correlated with the SSC. The reflectance at 350–600 nm increased with the increasing wavelength, and the relation between the reflectance and the SSC was not significant. There was a significant reflection peak at 685–720 nm, and the second reflection peak appeared at 790–830 nm. The reflectance of the second reflection peak was related to the SSC. When the suspended sediment concentration was below 700 mg/L, the second reflection peak was lower than the first reflection peak; with the increasing SSC, the reflection of the second peak gradually increased to exceed the first reflection peak. This phenomenon also appeared in the GOCI equivalent reflectance. The equivalent reflectance showed an upward trend in B1-B5, but the reflectance differences of different sediment samples were not obvious. However, in B5-B8, there was a significant relationship between the spectral band and the SSC; with the increasing SSC, the values of B7 and B8 (especially B8) showed an upward trend until they exceeded those of B5 and B6. Therefore, based on the above patterns, in this paper, the differences in the variation of B7 and B8 and B5 and B6 under changing SSC were considered as factors in the inversion algorithm.
SSC= 10
R (B 7) A1 + A2 * rs Rrs (B3) ,
SSC= 10
R (B 7) R (B6) (C1+ C2* rs − C3* rs ) Rrs (B 4) Rrs (B3) ,
(A1 = 1.1323, A2 = 1.0056)
(C1 = 1.5428, C2 = 1.8170, C3 (6)
= 0.4319) R (B8) (D2* rs ) Rrs (B6) ,
(5)
3.2. Results validation of atmospheric correction
SSC= D1*e
(D1 = 29.5693, D2 = 4.0210)
Due to the influence of cloud cover, some measured data lacked corresponding satellite data. According to the satellite-in situ matchup verification method, we filtered the data and finally obtained 13 groups of data. We calculated the RMSE and the APD to evaluate the accuracy of atmospheric correction results. Fig.4 shows that the Rrs values
According to the result of the comparison, model 3 based on B8/B6 achieved the best inversion effect of the SSC in Hangzhou Bay; its RMSE and MRE were lower than those of the other two models. Based on the spectral analysis results of the in situ measured data and the accuracy evaluation results of each model, model 3 was used in this study.
(7)
Fig. 3. Reflected spectrum of suspended sediment: (a) The measured spectrum of Hangzhou Bay and the range of the GOCI band; (b) measured spectral GOCI band equivalent remote sensing reflectance. 4
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 4. Comparison of Rrs between measured in situ and derived from UV-AC algorithm (a: Comparison for all bands; b: Comparison for B8/B6).
The SSC of Hangzhou Bay is influenced by numerous complicated factors, mainly including tidal effects, wind, and river runoff (Su and Wang, 1989). The influence of runoff could be ignored because the acquisition times of datasets were close. According to the 10-metre wind speed data obtained from European Centre for Medium-Rage Forecasts (ECMWF), the wind speed of Hangzhou Bay did not change significantly within the tidal cycle, remaining at approximately 5 m/s, and the daily variation in wind speed was not substantial, at 2 m/s. Thus, in this paper, we regard the environmental background on each day to have been relatively stable and not changed significantly throughout the tidal cycle. The tidal variation was considered as the main factor that influenced the SSC in Hangzhou Bay. Combined with the hourly tidal level data of the Zhapu tide gauge station (the data were obtained from a tidal table published by China’s State Oceanic Administration), the changes in suspended sediment in Hangzhou Bay under tidal forcing were studied.
Fig. 5. Regression fitting model (B8/B6).
4. Results and discussion The study took the flood season as an example, and GOCI images with minimal cloud cover were selected for a complete tidal cycle. We obtained 24 scenes, and the best model was used to estimate the SSC in Hangzhou Bay. In general, in the first half month of the lunar calendar, the neap tide occurred on the seventh and the eighth days of the lunar calendar, the spring tide occurred on the fifteenth and the sixteenth days of the lunar calendar, and the middle tide occurred around the tenth day of the lunar calendar. Due to its geographical location, the tidal effect in Hangzhou Bay will be two to three days later. Thus, the acquisition times of the datasets were selected to be July 25th, 2015 (neap tide), July 29 th, 2015 (middle tide), and August 2 nd, 2015 (spring tide).
4.1. Variation of the SSC in Hangzhou Bay under changes in tidal level The hourly change of the SSC in Hangzhou Bay under the neap tide, middle tide and spring tide was compared with the tidal data to investigate the effect of tidal changes caused by different tidal types. The results are shown in Figs. 7–9. (1) Under different tidal types, the change in tidal level had different influences on the SSC in Hangzhou Bay. The spring tidal influence
Fig. 6. Comparison of the accuracy of three empirical models (a: Model 1 provided by He et al.; b: Model 2 provided by GDPS; c: Model 3 provided by the authors). 5
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 7. (a–h) Hourly maps of SSC in Hangzhou Bay according to the GOCI on July 25th, 2015 (neap tide); (i) the hourly tidal level of the Zhapu tide gauge station on the same day.
July 25th, 2015, was the neap tide day (Fig. 7). The imaging time of the day was from the start of the ebb tide to the end of the flood tide. The change in tidal level on that day had little effect on the SSC. The high-value zone of SSC in the southern edge and the northeast of Hangzhou Bay appeared during the ebb tide. In the early time of the flood tide, the clean water in the outer sea was driven by the tide, and the main water channel was between the islands of Zhoushan. The area of relatively clean water (SSC < 80 mg/L) in the south of Hangzhou Bay was increased by 44.5% due to this phenomenon. July 29th, 2015, was the middle tide day (Fig. 8). On that day, the GOCI image recorded a relatively complete rise and fall period, among which the images shown in Fig. 8(a–b) and Fig. 8(g–h) had higher SSC
was the largest, and the neap tidal influence was smaller. The stronger the tidal action, the greater the change in water level and the greater the flow rate of water. Thus, the resuspension ability of the sediment in the water was enhanced, resulting in the high SSC and the increased turbidity area. (2) Although there were some differences in the influence of tidal intensity on SSC in Hangzhou Bay, there was a consistent spatial pattern in which during the middle period flood tide and ebb tide, the water flow rate was the largest and the disturbance of the bottom sediment was the most obvious so that the SSC of the surface water was increased obviously.
6
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 8. Same as Fig. 7 but for July 29th, 2015 (middle tide).
August 2nd, 2015, was the spring tide day (Fig. 9). The imaging time of the GOCI covered a relatively complete tidal cycle. Fig. 9 shows that the SSC in Hangzhou Bay was relatively high and the turbid region was large. Compared with the middle tide and neap tide, the SSC of Hangzhou Bay did not significantly decrease beyond the change of the tidal level. The two images of 10:28 and 11:28 had obviously high-value areas of SSC, and the hourly tidal level data showed that the average tidal level change was 100 cm/h (middle tide: 79 cm/h; neap tide: 58 cm/h). The tidal level changes at 10:28 and 11:28 were respectively 232 cm/h and 157 cm/h, which were the two most dramatic changes of the tidal level of the day. Under the rapid change of water flow, the disturbance
than the other images. The imaging time of Fig. 8(a–b) was 8:28 and 9:28 in the morning. At this time, in the middle time period of the flood tide, the water flow and the tidal level changed more violently and the water flow velocity caused the sediment to suspend, causing the SSC to exhibit a high-value region in the top of Hangzhou Bay. Subsequently, the flood tide ended and the ebb tide began, a period lasting for nearly 4 h. During this period, the water level changed slowly, the suspended sediment gradually deposited, and the water was restored to a cleaner state. The imaging time of Fig. 8(g–h) was during the ebb tide, and the water level changed obviously during this period. The sediment source in this period was considered to be the bottom resuspended sediment caused by the increase of the flow velocity during the ebb tide. 7
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 9. Same as Fig. 7 but for August 2nd, 2015 (spring tide).
sediment concentration in the sea area decreased and the water flow mainly flowed from the south side, while the suspended sediment was gradually deposited to the south.
degree of suspended sediment was strengthened and the SSC of surface water was higher. The slack tide period was short during the period of spring tide, but the water level change was still obvious. Even at 8:30 at the end of the ebb tide, the changing range of the tidal level was comparable to that of the middle period of the mid-tide ebb tide; thus, the SSC in water of the spring tide was high throughout day. In addition, it can be clearly seen in the images of the day that the SSC in several major watercourses in the Zhoushan archipelago was relatively low during the flood tide period. Due to the flushing of clean water from the outer sea, the muddy waters in the Hangzhou Bay area have several relatively clean areas in the east, which are located in the channels of large watercourses. During the period of ebb tide, the
4.2. Spatial distribution of SSC in Hangzhou Bay under different tidal types According to the GOCI imaging time, the study took the average daily data of suspended sediment in Hangzhou Bay during the eight time periods of spring, middle and neap tides and obtained the average daily SSC of spring, middle and neap tides in Hangzhou Bay (Fig.10). In western Hangzhou Bay, the SSC was high in all three tidal types, but the spatial difference was not obvious. Eastern Hangzhou Bay was 8
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 10. Average daily SSC in Hangzhou Bay (a: neap tide; b: middle tide; c: spring tide).
the inflow of water from the open sea made the northern waters relatively clean, while the southern waters had relatively high sediment content. During the spring tide period, the large flow of water caused the Yangtze Estuary to have a strong resuspension effect near the Nanhui Spit; thus, the water content was similar between the north and south sides. Because of the water flow "in from the north and out to the south" phenomenon, in the estuarine area, the turbidity belt in the south was wider than that in the north, and the turbid waters in the southern region extended to the island of Zhoushan, while the edge of the northern turbid waters did not reach the Nanhui Spit.
mostly sea area, and the water throughout this area was relatively clean. The SSC in Hangzhou Bay was positively correlated with tidal strength. In general, the distribution of suspended sediment concentration in Hangzhou Bay waters had the following characteristics: (1) At the top of the bay, the suspended sediment concentration was higher in the north and south and lower in the middle. On the whole, the SSC of the top of the bay was higher than that of other large areas in the neap tide and middle tide, while the sediment concentration in the spring tide was similar to that of other waters. The top of Hangzhou Bay is 100 km wide, but the west of Hangzhou Bay is only 20 km; within a distance of 100 km, the width of Hangzhou Bay is reduced by 80 km. Due to the sharp contraction of the Hangzhou Bay estuary, the strength of water flow near the top of the bay is greater so that the capacity of the water carrying sediment is enhanced and the bottom sediment is resuspended. Therefore, in the three different tidal types, the sediment concentration in the estuary was at a higher level. (2) In the middle of Hangzhou Bay, especially from Haiyan to Jinshan, due to the terrain difference between the north and the south, there was a significant difference in SSC. This phenomenon was evident in the image of spring tide and was mainly manifested in the higher SSC in the southern Hangzhou Bay, while the northern region was relatively clean. Andon, which is on the south bank of Hangzhou Bay, was the silting coast area. The terrain was relatively low, the water depth was shallow, and the area of sediment resuspension effect was stronger, causing the SSC on the coast of Andon area to reach the highest level. In the central area of Hangzhou Bay, the water of the north shore area was relatively clean. This phenomenon was mainly related to the topography of the north side of Hangzhou Bay. There is a deep trough on the north shore from Haiyan to Jinshan. The depth of the area is deep, and the water flow weakens the resuspension of the sediment. Thus, the rapid settlement of sediment causes the water body in the region to be relatively clean. (3) In the estuarine area, the water content in the north side of the water body was lower than that in the south side during the spring tide and the neap tide periods. In the spring tide period, the water content of the north and south sides was similar, which was mainly related to the trend direction. The study showed that the water flows from the north to Hangzhou Bay during the flood tide and outflows from the south during the ebb tide, and due to the geographical barrier on the south side, the velocity of this area was slow and sediment deposition was obvious. During the neap tide period and middle tide period, the tides were not very strong and
4.3. Analysis of daily variation of SSC in Hangzhou Bay under different tidal types The SSC in Hangzhou Bay was significantly different under different tidal types. The average concentration of suspended sediment in the water in the neap tide was 98.33 mg/L, while that in the middle tide was 146.03 mg/L and that in the spring tide was 239.69 mg/L. In contrast, the variation in SSC over a tidal day in Hangzhou Bay was far greater than that on monthly scale. The variation intensity and frequency distribution of CV of the suspended sediment were obtained by calculating the impact of the three days of the spring tide, middle tide and neap tide (Fig. 11). It can be seen from Fig. 11 that under the action of three different tidal types—spring tide, middle tide and neap tide—in the waters of Hangzhou Bay, the CV of SSC in water was far greater than that of the outer sea in one day. This phenomenon was more obvious during the middle tide period and spring tide period. In the case of Hangzhou Bay waters, the CV was the lowest in the neap tide period and the CV of most regions was less than 0.6. The CV of most sea areas during the middle tide period and spring tide period was concentrated in the range of 0.2 to 0.8. The area with high CV was concentrated in eastern Hangzhou Bay. The region with high CV (0.4-0.8) during the spring tide was the largest, with an area 17.3% greater than that during the middle tide. During the middle tide, approximately 18.05% of region was in an extremely high CV (> 1.0), and the region was mainly distributed on the south bank. The reason was that during the ebb tide of the middle tide, the SSC in the Hangzhou Bay area was low and the region was relatively clean; thus, the differences in SSC values among the eight times were large. During the spring tide, most regions of Hangzhou Bay exhibited a high SSC value and the water quality was turbid throughout the day; therefore, the CV during the spring tide was slightly lower than that during the other tidal types. The effect of tidal type variation was relatively small compared with 9
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 11. The variation and distribution of SSC. (a: the CV map of SSC on July 25th, 2015; b: the CV map of SSC on July 29th, 2015; c: the CV map of SSC on August 2nd, 2015; d: the CV map of the three tidal types).
Although cloud coverage was avoided in data selection, the first three images of July 29th, 2015 (middle tide), still included clouds, causing the CV of the middle tide to show a false height. Therefore, from the perspective of CV, the variation of SSC was more obvious in one day than for the entire tidal cycle.
Table 1 The CV of the different tidal types.
Neap tide day Middle tide day Spring tide day Tidal type change
0-0.2
0.2-0.4
0.4-0.6
0.6-0.8
0.8-1.0
> 1.0
21.78% 6.05% 3.22% 12.18%
45.21% 23.36% 23.04% 25.67%
17.25% 24.94% 30.21% 32.46%
7.61% 16.37% 28.40% 21.96%
3.75% 11.23% 11.72% 7.52%
4.40% 18.05% 3.41% 0.28%
4.4. Analysis of SSC change state in typical regions under different tidal patterns
the change in SSC caused by the change in tide time on a single day (Table 1). If a CV of 0.6 was set as the standard, on the neap tide day, approximately 8.15% of Hangzhou Bay reached this standard; the value for the middle tide was 29.29%, and the value for the spring tide was 15.13%. However, the value for the entire tidal cycle was only 7.77%.
Due to the complex topography of Hangzhou Bay, under the combined action of hydrodynamic conditions and tidal currents, the change in SSC in water was obvious, but the variation patterns and range of different regions differed. Five areas (5 km*5 km) (Fig. 12) were sampled in the typical waters of Hangzhou Bay. Region A was located on 10
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
due to the different hydrodynamic factors in the area outside Hangzhou Bay. The influence of different tidal types on the SSC in Hangzhou Bay was in the order spring tide > middle tide > neap tide. 5. Conclusions This study used the GOCI spectral response function to transform the reflectance of the in situ spectral data into the GOCI bands in Hangzhou Bay. According to the variation difference of SSC under different bands used to build models, the optimal model was selected to map the hourly SSC in Hangzhou Bay. The main conclusions of the study were as follows: 1 The SSC in Hangzhou Bay was high and the reflectance in the spectrum of the near-infrared band was positively correlated with the SSC, but it was not obvious in the visible range. Based on this result, the ratio of B8/B6 was suitable to map SSC of Hangzhou Bay with the least error. 2 According to the SSC maps made by hourly GOCI images, the change in tide level in one day had an obvious effect on the SSC. During the middle period of the flood tide and ebb tide, the change in SSC was the most dramatic. 3 Using the CV to quantitatively characterize the variation of SSC revealed that the overall level of the CV of SSC was higher in Hangzhou Bay during the spring tide period. The region of high variation was the largest in the middle tide day and was mainly concentrated in the middle and southern parts of Hangzhou Bay. Compared with the tidal type variation, the daily variation of the SSC was more obvious during changes in the tide. 4 Analysis of the changes in SSC in one day in the typical regions of Hangzhou Bay showed that the change in SSC in Hangzhou Bay was the strongest on the south bank, followed by Nanhui Spit. The changes in SSC near the Zhapu deep trough and the middle-northern part of the estuary were similar, both lower than the overall level, but still higher than the outer sea.
Fig. 12. Distribution of the typical regions of Hangzhou Bay.
the north bank of Hangzhou Bay with deep waters and weak resuspension ability. Region B was located on the south bank of Hangzhou Bay with a shallow depth and strong resuspension ability. Regions C and D were mainly located in the key parts of the tidal channel. Region E was located outside of Hangzhou Bay, where the SSC was low and was not affected by Hangzhou Bay. In this paper, the average SSC of 100 data points was substituted for the SSC in the region and the variation of the SSC in the typical regions was obtained. The graphs in Fig. 13 show that the change of tide during the neap tide (July 25th, 2018) had little effect on the changes in SSC in the five typical regions. Region C was located at Nanhui Spit, which was the main channel for freshwater from the Yangtze River estuary to enter Hangzhou Bay, and the SSC was slightly higher than that of other sea areas. Overall, the SSC in all of the regions exhibited a small increase at the end of the ebb tide and began to decrease and maintain a low level after the flood tide began. During the middle tide period (July 29th, 2015), the SSC was high in region B and region C. In general, the SSC was in a low-value state during the flood tide period, but there was an obvious trend of increasing in the middle period of the ebb tide in the Hangzhou Bay area but not in the outer sea area. During the spring tide period (August 2nd, 2015), the SSC values in region C, region B and region A were higher than those in other regions. The peak period was the most intense period of the flood tide, in which the fluctuation of the south bank was the most obvious, with a CV value of 0.58. From the perspective of the CV, in addition to the outer sea area of Zhoushan, the changes in SSC in the other Hangzhou Bay regions were the largest during the middle tide period. The imaging time of GOCI covered a tide time in the periods of neap tide, middle tide and spring tide, and the change in tide level had the most obvious effect on the SSC. In addition, during the spring and neap tide periods, the CV of SSC in a single day was generally higher than that of the whole tidal cycle. This finding indicated that the influence of tidal fluctuation in one day was the key factor for the variation of SCC in Hangzhou Bay. The influence of the middle tide on the change in SSC was the largest, followed by the spring tide day and neap tide day. The SSC in the region outside Hangzhou Bay was less affected by different tidal types and was basically in a low-level clean state, mainly
The study used GOCI images of Hangzhou Bay in the flood period as the data source and deduced the inversion formula of SSC in Hangzhou Bay waters based on spectral analysis. With the advantage of GOCI's high spatial and temporal resolution, the distribution and variation trend of SSC in Hangzhou Bay were obtained. The results can provide a reference for the inversion of SSC in turbid estuaries and provide a theoretical reference for marine water quality monitoring based on remote sensing technology. In addition to Hangzhou Bay, this research model can be applied when studying other macrotidal estuaries of the world. Strengthening the monitoring capability of marine environment changes on a short time scale can further refine the migration and settlement patterns of suspended sediment, which is also useful in assessing the risks of sediment deposition to marine ecology, channel dredging, and port security. Author contributions Yuekai Hu, Zhifeng Yu and Bin Zhou conceived the experimental design of the study. Zhifeng Yu and Bin Zhou participated in the fieldwork to gather in situ data in the Hangzhou Bay. All authors contributed to writing the paper. Acknowledgements This research was jointly supported by the National Key Research and Development Program of China (#2016YFC1400906), the High Resolution Earth Observation Systems of National Science and Technology Major Projects (grant 41-Y20A31-9003-15/17), the National Natural Science Foundation of China (41501374), and the Zhejiang Provincial Natural Science Foundation of China 11
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 13. Variation of SSC in typical regions of Hangzhou Bay (0725-0802).
(LQ16D010001). We acknowledge Korea Ocean Satellite Centre (KOSC) for providing the GOCI data, and China’s State Oceanic Administration for the tidal data, used in this study. We are also grateful for the suggestions from the reviewers, which helped us to improve our work.
Doxaran, D., Froidefond, J.M., Castaing, P., Babin, M., 2009. Dynamics of the turbidity maximum zone in a macrotidal estuary (the Gironde, France): observations from field and MODIS satellite data. Estuar. Coast. Shelf Sci. 81 (3), 321–332. https://doi.org/ 10.1016/j.ecss.2008.11.013. Eom, J., Choi, J.-K., Won, J.-S., Ryu, J.-H., Doxaran, D., Ruddick, K., Lee, S., 2016. Spatiotemporal Variation in Suspended Sediment Concentrations and Related Factors of Coastal Waters Based on Multispatial Satellite Data in Gyeonggi Bay, Korea. J. Coast. Res. 333, 653–667. https://doi.org/10.2112/jcoastres-d-16-00012.1. Fettweis, M., Nechad, B., Van den Eynde, D., 2007. An estimate of the suspended particulate matter (SPM) transport in the southern North Sea using SeaWiFS images, in situ measurements and numerical model results. Cont. Shelf Res. 27 (10–11), 1568–1583. https://doi.org/10.1016/j.csr.2007.01.017. Froidefond, J.M., Lavender, S., Laborde, P., Herbland, A., Lafon, V., 2002. Sea WiFS data interpretation in a coastal area in the bay of Biscay. Int. J. Remote Sens. 23 (5), 881–904. https://doi.org/10.1080/01431160110040332. He, X., Bai, Y., Pan, D., Huang, N., Dong, X., Chen, J., et al., 2013. Using geostationary satellite ocean color data to map the diurnal dynamics of suspended particulate matter in coastal waters. Remote Sens. Environ. 133, 225–239. https://doi.org/10. 1016/j.rse.2013.01.023. He, X., Bai, Y., Pan, D., Tang, J., Wang, D., 2012. Atmospheric correction of satellite ocean color imagery using the ultraviolet wavelength for highly turbid waters. Opt. Express 20 (18), 20754. https://doi.org/10.1364/oe.20.020754. Hu, Z., Pan, D., He, X., Bai, Y., 2016. Diurnal variability of turbidity fronts observed by geostationary satellite ocean color remote sensing. Remote Sens. (Basel) 8 (2), 1–15. https://doi.org/10.3390/rs8020147. Lamquin, N., Mazeran, C., Doxaran, D., Ryu, J.H., Park, Y.J., 2012. Assessment of GOCI
References Binding, C.E., Bowers, D.G., 2005. Estimating suspended sediment concentrations from ocean colour measurements in moderately turbid waters ; the impact of variable particle scattering properties. Remote Sens. Environ. 94, 373–383. https://doi.org/ 10.1016/j.rse.2004.11.002. Byun, D.S., Wang, X.H., Zavatarelli, M., Cho, Y.K., 2007. Effects of resuspended sediments and vertical mixing on phytoplankton spring bloom dynamics in a tidal estuarine embayment. J. Mar. Syst. 67 (1–2), 102–118. https://doi.org/10.1016/j.jmarsys. 2006.10.003. Cai, L., Tang, D., Li, C., 2015. An investigation of spatial variation of suspended sediment concentration induced by a bay bridge based on Landsat TM and OLI data. Adv. Space Res. 56 (2), 293–303. https://doi.org/10.1016/j.asr.2015.04.015. Chatanantavet, P., Lamb, M.P., 2014. Sediment transport and topographic evolution of a coupled river and river plume system: an experimental and numerical study. J. Geophy. Res. 119, 1263–1282. https://doi.org/10.1002/2013JF002810.Received. Cuuran, Pj., Novo, E.M.M., 1988. The relationship between suspended sediment concentration and remotely sensed spectral radiance - a review. J. Coast. Res. 4 (3), 351–368.
12
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al. radiometric products using MERIS, MODIS and field measurements. Ocean. Sci. J. 47 (3), 287–311. https://doi.org/10.1007/s12601-012-0029-z. Liu, J., Liu, J., He, X., Pan, D., Bai, Y., Zhu, F., et al., 2018. Diurnal dynamics and seasonal variations of total suspended particulate matter in highly turbid hangzhou bay waters based on the geostationary ocean color imager. Ieee J. Sel. Top. Appl. Earth Obs. Remote. Sens. 11 (7), 2170–2180. https://doi.org/10.1109/JSTARS.2018.2830335. Lyons, W.B., Neilson, B.J., Cronin, L.E., 1982. Estuaries and nutrients. Estuaries 5 (4), 326. https://doi.org/10.2307/1351759. Mao, Z., Chen, J., Pan, D., Tao, B., Zhu, Q., 2012. A regional remote sensing algorithm for total suspended matter in the East China Sea. Remote Sens. Environ. 124, 819–831. https://doi.org/10.1016/j.rse.2012.06.014. Novo, E.M., Hansom, J.D., Curran, P.J., 1989. The effect of sediment type on the relationship between reflectance and suspended sediment concentration. Int. J. Remote Sens. 10 (7), 1283–1289. https://doi.org/10.1080/01431168908903967. Pan, Y., Shen, F., Verhoef, W., 2017. An improved spectral optimization algorithm for atmospheric correction over turbid coastal waters: a case study from the Changjiang (Yangtze) estuary and the adjacent coast. Remote Sens. Environ. 191, 197–214. https://doi.org/10.1016/j.rse.2017.01.013. Roberto, Barbini, Francesco, Colao, Roberta, Fantoni, Luca, Fiorani, G Okladnikov, I., Antonio, Palucci, 2005. Comparison of SeaWiFS, MODIS‐Terra and MODIS‐Aqua in the Southern Ocean. Int. J. Remote Sens. 26 (22), 2471–2478. Sá, C., D’Alimonte, D., Brito, A.C., Kajiyama, T., Mendes, C.R., Vitorino, J., et al., 2015. Validation of standard and alternative satellite ocean-color chlorophyll products off Western Iberia. Remote Sens. Environ. 168, 403–419. https://doi.org/10.1016/j.rse. 2015.07.018. Schlünz, B., Schneider, R.R., 2000. Transport of terrestrial organic carbon to the oceans by rivers: Re-estimating flux- and burial rates. Int. J. Earth Sci. 88 (4), 599–606. https://doi.org/10.1007/s005310050290. Schoellhamer, D.H., 2002. Variability of suspended-sediment concentration at tidal to annual time scales in San Francisco Bay, USA. Cont. Shelf Res. 22 (11–13), 1857–1866. https://doi.org/10.1016/S0278-4343(02)00042-0. Siswanto, E., Tang, J., Yamaguchi, H., 2011. Empirical ocean-color algorithms to retrieve chlorophyll- a, total suspended matter, and colored dissolved organic matter
absorption coefficient in the Yellow and East China Seas. J. Oceanogr. 67, 627–650. https://doi.org/10.1007/s10872-011-0062-z. Su, J., Wang, K., 1989. Changjiang River plume and suspended sediment transport in Hangzhou. Cont. Shelf Res. 9 (1), 93–111. https://doi.org/10.1016/0278-4343(89) 90085-X. Tang, D.L., Ni, I.H., Müller-Karger, F.E., Liu, Z.J., 1998. Analysis of annual and spatial patterns of CZCS-derived pigment concentration on the continental shelf of China. Cont. Shelf Res. 18 (12), 1493–1515. https://doi.org/10.1016/S0278-4343(98) 00039-9. Wang, F., Zhou, B., Xu, J., Song, L., Wang, X., 2009. Application of neural network and MODIS 250 m imagery for estimating suspended sediments concentration in Hangzhou Bay, China. Environ. Geol. 56 (6), 1093–1101. https://doi.org/10.1007/ s00254-008-1209-0. Wu, W., Yang, Z., Tian, B., Huang, Y., Zhou, Y., Zhang, T., 2018. Estuarine, Coastal and Shelf Science Impacts of coastal reclamation on wetlands : loss, resilience, and sustainable management. Estuar. Coast. Shelf Sci. 210 (July (2017)), 153–161. https:// doi.org/10.1016/j.ecss.2018.06.013. Xie, D., Wang, Z., Gao, S., De Vriend, H.J., 2009. Modeling the tidal channel morphodynamics in a macro-tidal embayment, Hangzhou Bay. China. Continental Shelf Research 29 (15), 1757–1767. https://doi.org/10.1016/j.csr.2009.03.009. Yang, X., Mao, Z., 2016. Using GOCI retrieval data to initialize and validate a sediment transport model for monitoring diurnal. Water 8 (3), 108–125. https://doi.org/10. 3390/w8030108. Yeshaneh, E., Eder, A., Blöschl, G., 2014. Temporal variation of suspended sediment transport in the Koga catchment, North Western Ethiopia and environmental implications. Hydrol. Process. 28 (24), 5972–5984. https://doi.org/10.1002/hyp. 10090. Zhang, J., 1999. Heavy metal compositions of suspended sediments in the Changjiang (Yangtze River) estuary: significance of riverine transport to the ocean. Cont. Shelf Res. 19 (12), 1521–1543. https://doi.org/10.1016/S0278-4343(99)00029-1. Zhang, X., Hu, F., Zhang, J., Tang, D., 2014. Northward drift of suspended sediments in the Yangtze estuary in spring. Int. J. Remote Sens. 35 (11–12), 4114–4126. https:// doi.org/10.1080/01431161.2014.916455.
13
Contents lists available at ScienceDirect
Int J Appl Earth Obs Geoinformation journal homepage: www.elsevier.com/locate/jag
Tidal-driven variation of suspended sediment in Hangzhou Bay based on GOCI data
T
⁎
Yuekai Hua,b, Zhifeng Yua,c,h, , Bin Zhoua,h, Yuan Lid,e, Shoujing Yinf, Xianqiang Heg, Xiaoxue Penga, C.K. Shumc a
Institute of Remote Sensing and Earth Sciences, College of Science, Hangzhou Normal University, Hangzhou, 311121, China State Key Laboratory of Estuarine and Coastal Research, East China Normal University, Shanghai, 200062, China c Division of Geodetic Science, School of Earth Sciences, The Ohio State University, Columbus, OH, 43210, USA d School of Tourism and Urban & Rural Planning, Zhejiang Gongshang University, Hangzhou, 310018, China e Taihu Laboratory for Lake Ecosystem Research, State Key Laboratory of Lake Science and Environment, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing, 210008, China f Satellite Environment Centre, Ministry of Ecology and Environment, Beijing, 100094, China g State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography, State Oceanic Administration, Hangzhou, 310012, China h Zhejiang Provincial Key Laboratory of Urban Wetlands and Regional Change, Hangzhou, 311121, China b
A R T I C LE I N FO
A B S T R A C T
Keywords: Suspended sediment concentration Coefficient of variance Hangzhou Bay GOCI
The variation of suspended sediment concentration (SSC) in coastal waters plays a key role in the marine physical and chemical processes. Hangzhou Bay is a macrotidal estuary in eastern China. Under the influence of trapped tides, the SSC of Hangzhou Bay responds significantly commensurate with tidal frequencies, which leads to considerable changes in the ocean physical and chemical regimes, such as the underwater light field and turbidity front. However, polar-orbiting imaging satellites are unable to monitor the rapid variation of the SSC due to insufficient temporal resolution. In this study, we used the Geostationary Ocean Colour Imager (GOCI) data and in situ data of Hangzhou Bay to develop an appropriate remote sensing model to quantify the SSC of Hangzhou Bay and analyzed the impact of tides on the SSC. The research revealed the following findings: (1) The exponential model based on B8/B6 is best suited to remotely estimating the SSC in Hangzhou Bay, and the Mean Relative Error (MRE) is only 15.21%, (2) the variation of SSC in Hangzhou Bay is related to tidal forcing, especially during the ebb tide and the middle of the rising tide, (3) The effect of tidal driven SSC changes is less than the effect of diurnal tidal level variation of the middle tide day and the spring tide day tidal type, and (4) The maximum variation of SSC under tidal forcing is in the coastal ocean near the south bank of Hangzhou Bay and Nanhui Spit, while the origin of the variation of the Zhapu deep trough and the northern estuary is not obvious, which is speculated to be mainly related to the terrain and runoff of Hangzhou Bay. Furthermore, this study established a new model to map SSC in turbid seas and contributes to our understanding of SSC variation in the Hangzhou Bay, driven by strong tidal dynamics.
1. Introduction The estuary and coastal zones are at the junction of the mainland and coastal ocean. It is also the region where interactions and feedbacks of the Earth’s atmosphere, hydrosphere, lithosphere, biosphere, and biochemical processes are complex and with a wide range of temporal scales (Lyons et al., 1982). The biochemical processes in these zones are complex and variable. On the one hand, sediment transport in the estuary zone is governed by physical processes including land hydrology, ocean circulations, and gravitational interactions between the Earth,
⁎
the Moon and the Sun. Suspended sediment plays an important role in the geomorphological evolution and material circulation of coastal areas due to its characteristics of flocculation, adsorption and sedimentation. However, under rapid urbanization, the estuarine region has been adversely affected by significant human activities and by anthropogenic climate change (Wu et al., 2018). These factors resulted in the degradation of natural ecological environment with worsen water quality, exacerbated by semi-diurnal/diurnal suspended sediment concentration variations and chlorophyll concentration (Eom et al., 2016).
Corresponding author at: Institute of Remote Sensing and Earth Sciences, College of Science, Hangzhou Normal University, Hangzhou, 311121, China. E-mail address: [email protected] (Z. Yu).
https://doi.org/10.1016/j.jag.2019.101920 Received 17 October 2018; Received in revised form 28 June 2019; Accepted 4 July 2019 0303-2434/ © 2019 Elsevier B.V. All rights reserved.
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
source, and after atmospheric correction through the UV-AC model, the best empirical algorithm was used to calculate the hourly GOCI data to study the change in SSC in Hangzhou Bay under the influence of tides. This approach can realize the high-frequency dynamic monitoring of suspended sediment in Hangzhou Bay, and the results demonstrate a method and theoretical basis for studying the migration of SSC as a function of semi-diurnal and diurnal tides and the associated tidal currents.
High concentrations of suspended sediment are regarded as a typical characteristic of coastal waters, along with chlorophyll and yellow material. Suspended sediment affects the photosynthesis of phytoplankton and marine ecosystems by changing the light transmittance of surface water (Byun et al., 2007; Doxaran et al., 2009; Mao et al., 2012). In addition, the sedimentation of suspended sediment changes the characteristics of existing landforms and coastlines in estuarine coastal regions under the influence of changes in tidal intensity and riverine fluxes (Chatanantavet and L·amb, 2014;Yeshaneh et al., 2014). As a significant carrier of nutrients and pollutants, suspended sediment has a direct impact on marine chemistry under the influence of human activities and terrestrial inputs (Zhang, 1999). Therefore, studying the temporal and spatial variations and the distribution of suspended sediment concentration in estuary regions is of great importance in the investigations of water quality, underwater light field, marine ecology, estuary topography and carbon cycle (Schlünz and Schneider, 2000). Hangzhou Bay is one of the strongest tidal regions in China and in the world: the maximum tidal range can reach approximately 9 m, resulting in turbid seawater and producing SSC of up to 5000 mg/L, and the sediment is mainly originated from the Yangtze Estuary (He et al., 2013; Su and Wang, 1989; Xie et al., 2009). Monitoring the change in SCC in coastal water is a research topic significantly linked with severity of ocean water colour. During the past few decades, remote sensing technology has been widely used to map the variation and spatial distribution of SSC in coastal waters, including Landsat (Cai et al., 2015), Coastal Zone Color Scanner (CZCS) (Tang et al., 1998), Sea-viewing Wide Field-of-view Sensor (SeaWiFS) (Fettweis et al., 2007), and Moderate Resolution Imaging Spectroradiometer (MODIS) (Zhang et al., 2014). However, as a macrotidal estuary, Hangzhou Bay has undergone significant changes in SSC, especially under the influence of topography-trapped ocean tides (He et al., 2013), which has led to changes in the physical and chemical characteristics of the ocean, such as the underwater light field, turbidity front, etc. In such an extremely turbid estuary, the time resolution of traditional polar-orbiting satellites has been unable to meet the needs of ocean monitoring. In 2010, the South Korean Ocean Satellite Center (KOSC) launched the geostationary orbit satellite Geostationary Ocean Color Imager (GOCI), which can image 8 scenes a day (Lamquin et al., 2012). The high temporal resolution of the GOCI provides an opportunity to study hourly changes in suspended sediment in Hangzhou Bay. Moreover, the KOSC provided software system to process GOCI data, the GOCI Data Processing Software (GDPS), which integrates functions such as radiation calibration, atmospheric correction, and water colour factor inversion. Due to the spatial variability of ocean colour, the software's atmospheric correction algorithm has obvious errors in turbid water, and its single-band algorithm of SSC is not applicable to Hangzhou Bay. For this reason, Pan et al. (2017) proposed an improved spectral optimization algorithm for atmospheric correction. He et al. (2013) used the assumption of an ultraviolet (UV) “dark pixel” for atmospheric correction of the GOCI, which was useful in Hangzhou Bay (He et al., 2013; Liu et al., 2018; Yang and Mao, 2016). In the remote sensing inversion algorithm, GDPS provided the YOC algorithm to map the high-concentration suspended sediment in a subsequent version (Siswanto et al., 2011). He et al. (2013) established the model of B7 and B3 band ratios to map the SSC of Hangzhou Bay. The above research studies showed that the GOCI geostationary satellite measurements have great potential to advance in the investigation of ocean colour in Hangzhou Bay. However, the suitability of the proposed SSC model with GOCI data should be further validated, especially under tidal action. In the present study, we used in situ water quality data to analyse the spectral differences among collected samples of SSC. We propose a new empirical algorithm according to the analysis results. By comparison with existing algorithms, we chose the best algorithm for our study. Finally, cloudless GOCI data under a complete tidal cycle, for three days of the spring tide, middle tide and neap tide, was selected as the data
2. Materials and methods 2.1. Study area Hangzhou Bay, located in the middle of China's eastern coastline, is a typical funnel-shaped estuary. It extends from Changshan Gate–Xisan Gate in the west to Luwan Port–Waiyou Mount in the east. The tidal variation of the bay is classified as a semidiurnal tide, and the hydrodynamic environment in the bay is complex and significantly affected by ocean tides (Schoellhamer, 2002; Wang et al., 2009). Strong tidal action caused by the resuspension of bottom sediment combines with the Yangtze River diluted water every year to carry a large amount of sediment from the Nanhui Spit near Hangzhou Bay (Su and Wang, 1989). As a result, the SSC in Hangzhou Bay is higher than the SSC farther offshore, and it is also in a highly variable state. 2.2. In situ data Employing a combination of continuous flow measurements and fixed-point observations, our research team carried out water sampling and spectroscopic measurements simultaneously at 31 experimental locations (Fig. 1 right) during December 2nd to 13th, 2011, and May 16th and 22nd, 2014. To ensure the scientific value of the experimental data, the sample times of both samples covered the flood season and dry season as well as the periods of spring, middle and neap tides. During the sampling process, 500 mL of seawater was measured using a water collector and a measuring bottle. At the same time, the upward radiance from the ocean surface, the standard reflecting plate and the downward sky radiance were measured with an ASD handheld spectrometer. The reflectance of each location can be calculated based on these measured data. The water sample was filtered on the spot and stored in a refrigerator. In the laboratory, a crucible was placed in a muffle furnace and kept at 450 °C for two hours. After cooling, the sample was placed in a silica gel drier for two hours and weighed until the weight did not change. Subsequently, the thawed filter was placed in a crucible and the temperature of the muffle furnace was set at 500 °C for two hours, equilibrated with a silica gel dryer and weighed until the weight no longer changed. Finally, the SSC at each location was calculated by the weighing result and the volume of water. Based on the above method, 63 pairs of data (spectral and SSC) were obtained and used in this paper. 2.3. GOCI data GOCI data in this paper were downloaded from the Korea Ocean Satellite Centre (KOSC, http://kosc.kiost.ac.kr/). The observation centre of the GOCI is at 130 °E, 26 °N, and the spatial resolution is 500 m. GOCI data covers the East China Sea and the northern regions of Taiwan Province. The data includes eight bands (412 nm, 443 nm, 490 nm, 555 nm, 660 nm, 680 nm, 745 nm, and 865 nm). The band setting is similar to those of MODIS and SeaWiFS, which is suitable for ocean colour observation. In terms of data selection, studying the change in SSC under the tidal action in Hangzhou Bay required that the data could be covered by the complete tidal period and the cloud coverage in the study area should be less than 10%. We searched the GOCI data starting from 2011 to 2016 and determined three phases of research data that meet the requirements: 2015-07-25 (neap tide), 2
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 1. Study area and the distribution of in situ points.
degree of the object. Therefore, in this paper, the SSC in Hangzhou Bay waters was quantitatively characterized by the CV, and the study evaluated the effect of tidal hour variation on a daily time scale on the SSC in different sea areas of Hangzhou Bay. The formula was as follows:
2015-07-29 (middle tide), and 2015-08-02 (spring tide), for a total of 24 scene images. 2.4. Atmospheric correction and validation Atmospheric correction is one of the key steps in ocean colour remote sensing, and its accuracy is critical for scientific interpretation. In terms of data processing, GOCI Level-1B data provided by KOSC was used as the data source in this paper. The pre-processing of remote sensing data such as clipping, projection transform, geometric correction, was done on the GDPS platform. The atmospheric correction of GOCI data was done using the UVeAC algorithm to obtain the normalized water-leaving radiance. The remote sensing reflectance (Rrs) could be calculated through the normalized water-leaving radiance. Applied UV band atmospheric correction of the turbid water was mainly because it can well estimate the aerosol scattering reflectance of the turbid water, and the UV band has a low value of the water-leaving radiance, which can be ignored(He et al., 2012). For the GOCI data, the 412 nm band can be used instead due to the lack of UV band. The details of UV-AC algorithm can be found in He et al. (2013). In this paper, we used the in situ reflectance field work measured data in the Hangzhou Bay, and the GOCI remotely sensed reflectance measurements from the UV-AC algorithm for satellite and in situ validation studies. The main criteria are as follows (Eom et al., 2016; Froidefond et al., 2002; Roberto et al., 2005; Sá et al., 2015):
CV=
n ¯ )2 /n ∑i = 1 (SSCi − SSC ¯ SSC
(1)
where n is the daily image number of GOCI, SSCi is the suspended se¯ is the diment concentration of a grid of pixels in image i, and SSC average value of suspended sediment concentration on that day on the grid. The root mean square error (RMSE), the absolute percentage difference (APD), and the Mean Relative Error (MRE) were used to measure the performance of the atmospheric correction method and SSC inversion model.
RMSE=
n=i
∑n=1
[(Xi )insitu − (Xi )calculated]2 i
n=i 1 (Xi )calculated − (Xi )insitu APD= 100* *∑ n=1 i (Xi )insitu
MRE=
n=i 1 (Xi )calculated − (Xi )insitu *∑ *100% n = 1 i (Xi )insitu
)insitu
and (Xi Where (Xi or SSC, respectively.
1 Calculate the number of valid pixels (NVP) in the measured pixel space window (3*3). If NVP > 4, use it; otherwise discard it; 2 Calculate the average (AVG) and standard deviation (SD) of valid pixels, remove data other than AVG ± 1.5*SD to avoid the impact of outliers on the results; 3 Recalculate the AVG and SD and calculate the coefficient of variation (CV). If CV ≤ 0.15, it means that the data group meets the spatial uniformity requirements and can be used for subsequent acting; 4 Considering the characteristics of high dynamics in Hangzhou Bay, the time window of the study was set to ± 1 h.
)calculated
(2)
(3)
(4)
represent the in situ and calculated Rrs
3. Model establishment and validation 3.1. Measured spectral data processing and analysis On December 12th, 2011 (spring tide), the measured point (D02) sampling results showed that the sediment content of water was significantly changed in one day (Fig. 2). The maximum daily variation of SSC was 1157.6 mg/L, and the maximum per unit time was 738.2 mg/ L/h. There were 63 samples of measured in situ data in the Hangzhou Bay area (Fig. 1, right), and the SSC varied from 150 mg/L to 2300 mg/ L. In the ocean colour remote sensing of case II waters, the amount of suspended matter affects the optical characteristics of the water spectrum (Binding and Bowers, 2005; Cuuran and Novo, 1988; Novo et al., 1989). According to the different concentration gradients of the
2.5. Data analysis and accuracy assessment When the mean and the variance of study object are not consistent, the coefficient of variance (CV) can be used to measure the dispersion 3
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
derived from the UV-AC algorithm were consistent with the in situ measured results. The APD of all GOCI bands was 23.12 and the RMSE was 0.0083. The APD of each band was 35.1 (412 nm), 32.7 (443 nm), 24.5 (490 nm), 14.0 (510 nm), 17.3 (555 nm), 15.6 (670 nm), 19.9 (765 nm) and 25.8 (865 nm). In this paper, we used the band ratio to calculate the SSC, the APD of B8/B6 was 12.74, RMSE was 0.1082. According to these results, we concluded that the UV-AV algorithm enabled an accurate estimate of the ocean surface reflectance, and that our results are novel.
3.3. Establishment and comparison of the inversion model Fig. 2. Daily variation of SSC (Point D02, 2011-12-12).
The 63 pairs of data measured in Hangzhou Bay were divided into two parts, two-thirds (42 pairs) of which were used for modelling and the other third (21 pairs) to evaluate the error of the model. The differences in B7 and B8 and B5 and B6 provided a new direction for building algorithms to research the SSC of Hangzhou Bay. In this study, we tested all the combinations to select the remote sensing factor and found that B8/B6 achieved the best result in the inversion algorithm; the R2 was 0.91 (Fig. 5). In the process of using GOCI data to map the ocean colour in Hangzhou Bay, the model established by He (2013) and the YOC model provided by GDPS are widely used in current studies (He et al., 2013; Hu et al., 2016; Liu et al., 2018; Yang and Mao, 2016). These models were used to map the total suspended matter. In the process of the inversion of SSC, some parameters might need to be re-set. Therefore, the present research used the modelling data to recalculate the model parameters and used the other data to evaluate the accuracy of the model (Fig. 6). Finally, the best model was selected to map the SSC in Hangzhou Bay. The three formulas are as follows:
measured data, the data were divided into 10 groups. After the average processing of the spectra of each group, 10 water bodies with different sand contents were obtained (Fig. 3a). Then, the study used the spectral response function of the GOCI to convert spectral data to the equivalent remote sensing reflectance of corresponding bands (Fig. 3b). The spectral data showed that the reflectance was positively correlated with the SSC. The reflectance at 350–600 nm increased with the increasing wavelength, and the relation between the reflectance and the SSC was not significant. There was a significant reflection peak at 685–720 nm, and the second reflection peak appeared at 790–830 nm. The reflectance of the second reflection peak was related to the SSC. When the suspended sediment concentration was below 700 mg/L, the second reflection peak was lower than the first reflection peak; with the increasing SSC, the reflection of the second peak gradually increased to exceed the first reflection peak. This phenomenon also appeared in the GOCI equivalent reflectance. The equivalent reflectance showed an upward trend in B1-B5, but the reflectance differences of different sediment samples were not obvious. However, in B5-B8, there was a significant relationship between the spectral band and the SSC; with the increasing SSC, the values of B7 and B8 (especially B8) showed an upward trend until they exceeded those of B5 and B6. Therefore, based on the above patterns, in this paper, the differences in the variation of B7 and B8 and B5 and B6 under changing SSC were considered as factors in the inversion algorithm.
SSC= 10
R (B 7) A1 + A2 * rs Rrs (B3) ,
SSC= 10
R (B 7) R (B6) (C1+ C2* rs − C3* rs ) Rrs (B 4) Rrs (B3) ,
(A1 = 1.1323, A2 = 1.0056)
(C1 = 1.5428, C2 = 1.8170, C3 (6)
= 0.4319) R (B8) (D2* rs ) Rrs (B6) ,
(5)
3.2. Results validation of atmospheric correction
SSC= D1*e
(D1 = 29.5693, D2 = 4.0210)
Due to the influence of cloud cover, some measured data lacked corresponding satellite data. According to the satellite-in situ matchup verification method, we filtered the data and finally obtained 13 groups of data. We calculated the RMSE and the APD to evaluate the accuracy of atmospheric correction results. Fig.4 shows that the Rrs values
According to the result of the comparison, model 3 based on B8/B6 achieved the best inversion effect of the SSC in Hangzhou Bay; its RMSE and MRE were lower than those of the other two models. Based on the spectral analysis results of the in situ measured data and the accuracy evaluation results of each model, model 3 was used in this study.
(7)
Fig. 3. Reflected spectrum of suspended sediment: (a) The measured spectrum of Hangzhou Bay and the range of the GOCI band; (b) measured spectral GOCI band equivalent remote sensing reflectance. 4
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 4. Comparison of Rrs between measured in situ and derived from UV-AC algorithm (a: Comparison for all bands; b: Comparison for B8/B6).
The SSC of Hangzhou Bay is influenced by numerous complicated factors, mainly including tidal effects, wind, and river runoff (Su and Wang, 1989). The influence of runoff could be ignored because the acquisition times of datasets were close. According to the 10-metre wind speed data obtained from European Centre for Medium-Rage Forecasts (ECMWF), the wind speed of Hangzhou Bay did not change significantly within the tidal cycle, remaining at approximately 5 m/s, and the daily variation in wind speed was not substantial, at 2 m/s. Thus, in this paper, we regard the environmental background on each day to have been relatively stable and not changed significantly throughout the tidal cycle. The tidal variation was considered as the main factor that influenced the SSC in Hangzhou Bay. Combined with the hourly tidal level data of the Zhapu tide gauge station (the data were obtained from a tidal table published by China’s State Oceanic Administration), the changes in suspended sediment in Hangzhou Bay under tidal forcing were studied.
Fig. 5. Regression fitting model (B8/B6).
4. Results and discussion The study took the flood season as an example, and GOCI images with minimal cloud cover were selected for a complete tidal cycle. We obtained 24 scenes, and the best model was used to estimate the SSC in Hangzhou Bay. In general, in the first half month of the lunar calendar, the neap tide occurred on the seventh and the eighth days of the lunar calendar, the spring tide occurred on the fifteenth and the sixteenth days of the lunar calendar, and the middle tide occurred around the tenth day of the lunar calendar. Due to its geographical location, the tidal effect in Hangzhou Bay will be two to three days later. Thus, the acquisition times of the datasets were selected to be July 25th, 2015 (neap tide), July 29 th, 2015 (middle tide), and August 2 nd, 2015 (spring tide).
4.1. Variation of the SSC in Hangzhou Bay under changes in tidal level The hourly change of the SSC in Hangzhou Bay under the neap tide, middle tide and spring tide was compared with the tidal data to investigate the effect of tidal changes caused by different tidal types. The results are shown in Figs. 7–9. (1) Under different tidal types, the change in tidal level had different influences on the SSC in Hangzhou Bay. The spring tidal influence
Fig. 6. Comparison of the accuracy of three empirical models (a: Model 1 provided by He et al.; b: Model 2 provided by GDPS; c: Model 3 provided by the authors). 5
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 7. (a–h) Hourly maps of SSC in Hangzhou Bay according to the GOCI on July 25th, 2015 (neap tide); (i) the hourly tidal level of the Zhapu tide gauge station on the same day.
July 25th, 2015, was the neap tide day (Fig. 7). The imaging time of the day was from the start of the ebb tide to the end of the flood tide. The change in tidal level on that day had little effect on the SSC. The high-value zone of SSC in the southern edge and the northeast of Hangzhou Bay appeared during the ebb tide. In the early time of the flood tide, the clean water in the outer sea was driven by the tide, and the main water channel was between the islands of Zhoushan. The area of relatively clean water (SSC < 80 mg/L) in the south of Hangzhou Bay was increased by 44.5% due to this phenomenon. July 29th, 2015, was the middle tide day (Fig. 8). On that day, the GOCI image recorded a relatively complete rise and fall period, among which the images shown in Fig. 8(a–b) and Fig. 8(g–h) had higher SSC
was the largest, and the neap tidal influence was smaller. The stronger the tidal action, the greater the change in water level and the greater the flow rate of water. Thus, the resuspension ability of the sediment in the water was enhanced, resulting in the high SSC and the increased turbidity area. (2) Although there were some differences in the influence of tidal intensity on SSC in Hangzhou Bay, there was a consistent spatial pattern in which during the middle period flood tide and ebb tide, the water flow rate was the largest and the disturbance of the bottom sediment was the most obvious so that the SSC of the surface water was increased obviously.
6
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 8. Same as Fig. 7 but for July 29th, 2015 (middle tide).
August 2nd, 2015, was the spring tide day (Fig. 9). The imaging time of the GOCI covered a relatively complete tidal cycle. Fig. 9 shows that the SSC in Hangzhou Bay was relatively high and the turbid region was large. Compared with the middle tide and neap tide, the SSC of Hangzhou Bay did not significantly decrease beyond the change of the tidal level. The two images of 10:28 and 11:28 had obviously high-value areas of SSC, and the hourly tidal level data showed that the average tidal level change was 100 cm/h (middle tide: 79 cm/h; neap tide: 58 cm/h). The tidal level changes at 10:28 and 11:28 were respectively 232 cm/h and 157 cm/h, which were the two most dramatic changes of the tidal level of the day. Under the rapid change of water flow, the disturbance
than the other images. The imaging time of Fig. 8(a–b) was 8:28 and 9:28 in the morning. At this time, in the middle time period of the flood tide, the water flow and the tidal level changed more violently and the water flow velocity caused the sediment to suspend, causing the SSC to exhibit a high-value region in the top of Hangzhou Bay. Subsequently, the flood tide ended and the ebb tide began, a period lasting for nearly 4 h. During this period, the water level changed slowly, the suspended sediment gradually deposited, and the water was restored to a cleaner state. The imaging time of Fig. 8(g–h) was during the ebb tide, and the water level changed obviously during this period. The sediment source in this period was considered to be the bottom resuspended sediment caused by the increase of the flow velocity during the ebb tide. 7
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 9. Same as Fig. 7 but for August 2nd, 2015 (spring tide).
sediment concentration in the sea area decreased and the water flow mainly flowed from the south side, while the suspended sediment was gradually deposited to the south.
degree of suspended sediment was strengthened and the SSC of surface water was higher. The slack tide period was short during the period of spring tide, but the water level change was still obvious. Even at 8:30 at the end of the ebb tide, the changing range of the tidal level was comparable to that of the middle period of the mid-tide ebb tide; thus, the SSC in water of the spring tide was high throughout day. In addition, it can be clearly seen in the images of the day that the SSC in several major watercourses in the Zhoushan archipelago was relatively low during the flood tide period. Due to the flushing of clean water from the outer sea, the muddy waters in the Hangzhou Bay area have several relatively clean areas in the east, which are located in the channels of large watercourses. During the period of ebb tide, the
4.2. Spatial distribution of SSC in Hangzhou Bay under different tidal types According to the GOCI imaging time, the study took the average daily data of suspended sediment in Hangzhou Bay during the eight time periods of spring, middle and neap tides and obtained the average daily SSC of spring, middle and neap tides in Hangzhou Bay (Fig.10). In western Hangzhou Bay, the SSC was high in all three tidal types, but the spatial difference was not obvious. Eastern Hangzhou Bay was 8
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 10. Average daily SSC in Hangzhou Bay (a: neap tide; b: middle tide; c: spring tide).
the inflow of water from the open sea made the northern waters relatively clean, while the southern waters had relatively high sediment content. During the spring tide period, the large flow of water caused the Yangtze Estuary to have a strong resuspension effect near the Nanhui Spit; thus, the water content was similar between the north and south sides. Because of the water flow "in from the north and out to the south" phenomenon, in the estuarine area, the turbidity belt in the south was wider than that in the north, and the turbid waters in the southern region extended to the island of Zhoushan, while the edge of the northern turbid waters did not reach the Nanhui Spit.
mostly sea area, and the water throughout this area was relatively clean. The SSC in Hangzhou Bay was positively correlated with tidal strength. In general, the distribution of suspended sediment concentration in Hangzhou Bay waters had the following characteristics: (1) At the top of the bay, the suspended sediment concentration was higher in the north and south and lower in the middle. On the whole, the SSC of the top of the bay was higher than that of other large areas in the neap tide and middle tide, while the sediment concentration in the spring tide was similar to that of other waters. The top of Hangzhou Bay is 100 km wide, but the west of Hangzhou Bay is only 20 km; within a distance of 100 km, the width of Hangzhou Bay is reduced by 80 km. Due to the sharp contraction of the Hangzhou Bay estuary, the strength of water flow near the top of the bay is greater so that the capacity of the water carrying sediment is enhanced and the bottom sediment is resuspended. Therefore, in the three different tidal types, the sediment concentration in the estuary was at a higher level. (2) In the middle of Hangzhou Bay, especially from Haiyan to Jinshan, due to the terrain difference between the north and the south, there was a significant difference in SSC. This phenomenon was evident in the image of spring tide and was mainly manifested in the higher SSC in the southern Hangzhou Bay, while the northern region was relatively clean. Andon, which is on the south bank of Hangzhou Bay, was the silting coast area. The terrain was relatively low, the water depth was shallow, and the area of sediment resuspension effect was stronger, causing the SSC on the coast of Andon area to reach the highest level. In the central area of Hangzhou Bay, the water of the north shore area was relatively clean. This phenomenon was mainly related to the topography of the north side of Hangzhou Bay. There is a deep trough on the north shore from Haiyan to Jinshan. The depth of the area is deep, and the water flow weakens the resuspension of the sediment. Thus, the rapid settlement of sediment causes the water body in the region to be relatively clean. (3) In the estuarine area, the water content in the north side of the water body was lower than that in the south side during the spring tide and the neap tide periods. In the spring tide period, the water content of the north and south sides was similar, which was mainly related to the trend direction. The study showed that the water flows from the north to Hangzhou Bay during the flood tide and outflows from the south during the ebb tide, and due to the geographical barrier on the south side, the velocity of this area was slow and sediment deposition was obvious. During the neap tide period and middle tide period, the tides were not very strong and
4.3. Analysis of daily variation of SSC in Hangzhou Bay under different tidal types The SSC in Hangzhou Bay was significantly different under different tidal types. The average concentration of suspended sediment in the water in the neap tide was 98.33 mg/L, while that in the middle tide was 146.03 mg/L and that in the spring tide was 239.69 mg/L. In contrast, the variation in SSC over a tidal day in Hangzhou Bay was far greater than that on monthly scale. The variation intensity and frequency distribution of CV of the suspended sediment were obtained by calculating the impact of the three days of the spring tide, middle tide and neap tide (Fig. 11). It can be seen from Fig. 11 that under the action of three different tidal types—spring tide, middle tide and neap tide—in the waters of Hangzhou Bay, the CV of SSC in water was far greater than that of the outer sea in one day. This phenomenon was more obvious during the middle tide period and spring tide period. In the case of Hangzhou Bay waters, the CV was the lowest in the neap tide period and the CV of most regions was less than 0.6. The CV of most sea areas during the middle tide period and spring tide period was concentrated in the range of 0.2 to 0.8. The area with high CV was concentrated in eastern Hangzhou Bay. The region with high CV (0.4-0.8) during the spring tide was the largest, with an area 17.3% greater than that during the middle tide. During the middle tide, approximately 18.05% of region was in an extremely high CV (> 1.0), and the region was mainly distributed on the south bank. The reason was that during the ebb tide of the middle tide, the SSC in the Hangzhou Bay area was low and the region was relatively clean; thus, the differences in SSC values among the eight times were large. During the spring tide, most regions of Hangzhou Bay exhibited a high SSC value and the water quality was turbid throughout the day; therefore, the CV during the spring tide was slightly lower than that during the other tidal types. The effect of tidal type variation was relatively small compared with 9
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 11. The variation and distribution of SSC. (a: the CV map of SSC on July 25th, 2015; b: the CV map of SSC on July 29th, 2015; c: the CV map of SSC on August 2nd, 2015; d: the CV map of the three tidal types).
Although cloud coverage was avoided in data selection, the first three images of July 29th, 2015 (middle tide), still included clouds, causing the CV of the middle tide to show a false height. Therefore, from the perspective of CV, the variation of SSC was more obvious in one day than for the entire tidal cycle.
Table 1 The CV of the different tidal types.
Neap tide day Middle tide day Spring tide day Tidal type change
0-0.2
0.2-0.4
0.4-0.6
0.6-0.8
0.8-1.0
> 1.0
21.78% 6.05% 3.22% 12.18%
45.21% 23.36% 23.04% 25.67%
17.25% 24.94% 30.21% 32.46%
7.61% 16.37% 28.40% 21.96%
3.75% 11.23% 11.72% 7.52%
4.40% 18.05% 3.41% 0.28%
4.4. Analysis of SSC change state in typical regions under different tidal patterns
the change in SSC caused by the change in tide time on a single day (Table 1). If a CV of 0.6 was set as the standard, on the neap tide day, approximately 8.15% of Hangzhou Bay reached this standard; the value for the middle tide was 29.29%, and the value for the spring tide was 15.13%. However, the value for the entire tidal cycle was only 7.77%.
Due to the complex topography of Hangzhou Bay, under the combined action of hydrodynamic conditions and tidal currents, the change in SSC in water was obvious, but the variation patterns and range of different regions differed. Five areas (5 km*5 km) (Fig. 12) were sampled in the typical waters of Hangzhou Bay. Region A was located on 10
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
due to the different hydrodynamic factors in the area outside Hangzhou Bay. The influence of different tidal types on the SSC in Hangzhou Bay was in the order spring tide > middle tide > neap tide. 5. Conclusions This study used the GOCI spectral response function to transform the reflectance of the in situ spectral data into the GOCI bands in Hangzhou Bay. According to the variation difference of SSC under different bands used to build models, the optimal model was selected to map the hourly SSC in Hangzhou Bay. The main conclusions of the study were as follows: 1 The SSC in Hangzhou Bay was high and the reflectance in the spectrum of the near-infrared band was positively correlated with the SSC, but it was not obvious in the visible range. Based on this result, the ratio of B8/B6 was suitable to map SSC of Hangzhou Bay with the least error. 2 According to the SSC maps made by hourly GOCI images, the change in tide level in one day had an obvious effect on the SSC. During the middle period of the flood tide and ebb tide, the change in SSC was the most dramatic. 3 Using the CV to quantitatively characterize the variation of SSC revealed that the overall level of the CV of SSC was higher in Hangzhou Bay during the spring tide period. The region of high variation was the largest in the middle tide day and was mainly concentrated in the middle and southern parts of Hangzhou Bay. Compared with the tidal type variation, the daily variation of the SSC was more obvious during changes in the tide. 4 Analysis of the changes in SSC in one day in the typical regions of Hangzhou Bay showed that the change in SSC in Hangzhou Bay was the strongest on the south bank, followed by Nanhui Spit. The changes in SSC near the Zhapu deep trough and the middle-northern part of the estuary were similar, both lower than the overall level, but still higher than the outer sea.
Fig. 12. Distribution of the typical regions of Hangzhou Bay.
the north bank of Hangzhou Bay with deep waters and weak resuspension ability. Region B was located on the south bank of Hangzhou Bay with a shallow depth and strong resuspension ability. Regions C and D were mainly located in the key parts of the tidal channel. Region E was located outside of Hangzhou Bay, where the SSC was low and was not affected by Hangzhou Bay. In this paper, the average SSC of 100 data points was substituted for the SSC in the region and the variation of the SSC in the typical regions was obtained. The graphs in Fig. 13 show that the change of tide during the neap tide (July 25th, 2018) had little effect on the changes in SSC in the five typical regions. Region C was located at Nanhui Spit, which was the main channel for freshwater from the Yangtze River estuary to enter Hangzhou Bay, and the SSC was slightly higher than that of other sea areas. Overall, the SSC in all of the regions exhibited a small increase at the end of the ebb tide and began to decrease and maintain a low level after the flood tide began. During the middle tide period (July 29th, 2015), the SSC was high in region B and region C. In general, the SSC was in a low-value state during the flood tide period, but there was an obvious trend of increasing in the middle period of the ebb tide in the Hangzhou Bay area but not in the outer sea area. During the spring tide period (August 2nd, 2015), the SSC values in region C, region B and region A were higher than those in other regions. The peak period was the most intense period of the flood tide, in which the fluctuation of the south bank was the most obvious, with a CV value of 0.58. From the perspective of the CV, in addition to the outer sea area of Zhoushan, the changes in SSC in the other Hangzhou Bay regions were the largest during the middle tide period. The imaging time of GOCI covered a tide time in the periods of neap tide, middle tide and spring tide, and the change in tide level had the most obvious effect on the SSC. In addition, during the spring and neap tide periods, the CV of SSC in a single day was generally higher than that of the whole tidal cycle. This finding indicated that the influence of tidal fluctuation in one day was the key factor for the variation of SCC in Hangzhou Bay. The influence of the middle tide on the change in SSC was the largest, followed by the spring tide day and neap tide day. The SSC in the region outside Hangzhou Bay was less affected by different tidal types and was basically in a low-level clean state, mainly
The study used GOCI images of Hangzhou Bay in the flood period as the data source and deduced the inversion formula of SSC in Hangzhou Bay waters based on spectral analysis. With the advantage of GOCI's high spatial and temporal resolution, the distribution and variation trend of SSC in Hangzhou Bay were obtained. The results can provide a reference for the inversion of SSC in turbid estuaries and provide a theoretical reference for marine water quality monitoring based on remote sensing technology. In addition to Hangzhou Bay, this research model can be applied when studying other macrotidal estuaries of the world. Strengthening the monitoring capability of marine environment changes on a short time scale can further refine the migration and settlement patterns of suspended sediment, which is also useful in assessing the risks of sediment deposition to marine ecology, channel dredging, and port security. Author contributions Yuekai Hu, Zhifeng Yu and Bin Zhou conceived the experimental design of the study. Zhifeng Yu and Bin Zhou participated in the fieldwork to gather in situ data in the Hangzhou Bay. All authors contributed to writing the paper. Acknowledgements This research was jointly supported by the National Key Research and Development Program of China (#2016YFC1400906), the High Resolution Earth Observation Systems of National Science and Technology Major Projects (grant 41-Y20A31-9003-15/17), the National Natural Science Foundation of China (41501374), and the Zhejiang Provincial Natural Science Foundation of China 11
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al.
Fig. 13. Variation of SSC in typical regions of Hangzhou Bay (0725-0802).
(LQ16D010001). We acknowledge Korea Ocean Satellite Centre (KOSC) for providing the GOCI data, and China’s State Oceanic Administration for the tidal data, used in this study. We are also grateful for the suggestions from the reviewers, which helped us to improve our work.
Doxaran, D., Froidefond, J.M., Castaing, P., Babin, M., 2009. Dynamics of the turbidity maximum zone in a macrotidal estuary (the Gironde, France): observations from field and MODIS satellite data. Estuar. Coast. Shelf Sci. 81 (3), 321–332. https://doi.org/ 10.1016/j.ecss.2008.11.013. Eom, J., Choi, J.-K., Won, J.-S., Ryu, J.-H., Doxaran, D., Ruddick, K., Lee, S., 2016. Spatiotemporal Variation in Suspended Sediment Concentrations and Related Factors of Coastal Waters Based on Multispatial Satellite Data in Gyeonggi Bay, Korea. J. Coast. Res. 333, 653–667. https://doi.org/10.2112/jcoastres-d-16-00012.1. Fettweis, M., Nechad, B., Van den Eynde, D., 2007. An estimate of the suspended particulate matter (SPM) transport in the southern North Sea using SeaWiFS images, in situ measurements and numerical model results. Cont. Shelf Res. 27 (10–11), 1568–1583. https://doi.org/10.1016/j.csr.2007.01.017. Froidefond, J.M., Lavender, S., Laborde, P., Herbland, A., Lafon, V., 2002. Sea WiFS data interpretation in a coastal area in the bay of Biscay. Int. J. Remote Sens. 23 (5), 881–904. https://doi.org/10.1080/01431160110040332. He, X., Bai, Y., Pan, D., Huang, N., Dong, X., Chen, J., et al., 2013. Using geostationary satellite ocean color data to map the diurnal dynamics of suspended particulate matter in coastal waters. Remote Sens. Environ. 133, 225–239. https://doi.org/10. 1016/j.rse.2013.01.023. He, X., Bai, Y., Pan, D., Tang, J., Wang, D., 2012. Atmospheric correction of satellite ocean color imagery using the ultraviolet wavelength for highly turbid waters. Opt. Express 20 (18), 20754. https://doi.org/10.1364/oe.20.020754. Hu, Z., Pan, D., He, X., Bai, Y., 2016. Diurnal variability of turbidity fronts observed by geostationary satellite ocean color remote sensing. Remote Sens. (Basel) 8 (2), 1–15. https://doi.org/10.3390/rs8020147. Lamquin, N., Mazeran, C., Doxaran, D., Ryu, J.H., Park, Y.J., 2012. Assessment of GOCI
References Binding, C.E., Bowers, D.G., 2005. Estimating suspended sediment concentrations from ocean colour measurements in moderately turbid waters ; the impact of variable particle scattering properties. Remote Sens. Environ. 94, 373–383. https://doi.org/ 10.1016/j.rse.2004.11.002. Byun, D.S., Wang, X.H., Zavatarelli, M., Cho, Y.K., 2007. Effects of resuspended sediments and vertical mixing on phytoplankton spring bloom dynamics in a tidal estuarine embayment. J. Mar. Syst. 67 (1–2), 102–118. https://doi.org/10.1016/j.jmarsys. 2006.10.003. Cai, L., Tang, D., Li, C., 2015. An investigation of spatial variation of suspended sediment concentration induced by a bay bridge based on Landsat TM and OLI data. Adv. Space Res. 56 (2), 293–303. https://doi.org/10.1016/j.asr.2015.04.015. Chatanantavet, P., Lamb, M.P., 2014. Sediment transport and topographic evolution of a coupled river and river plume system: an experimental and numerical study. J. Geophy. Res. 119, 1263–1282. https://doi.org/10.1002/2013JF002810.Received. Cuuran, Pj., Novo, E.M.M., 1988. The relationship between suspended sediment concentration and remotely sensed spectral radiance - a review. J. Coast. Res. 4 (3), 351–368.
12
Int J Appl Earth Obs Geoinformation 82 (2019) 101920
Y. Hu, et al. radiometric products using MERIS, MODIS and field measurements. Ocean. Sci. J. 47 (3), 287–311. https://doi.org/10.1007/s12601-012-0029-z. Liu, J., Liu, J., He, X., Pan, D., Bai, Y., Zhu, F., et al., 2018. Diurnal dynamics and seasonal variations of total suspended particulate matter in highly turbid hangzhou bay waters based on the geostationary ocean color imager. Ieee J. Sel. Top. Appl. Earth Obs. Remote. Sens. 11 (7), 2170–2180. https://doi.org/10.1109/JSTARS.2018.2830335. Lyons, W.B., Neilson, B.J., Cronin, L.E., 1982. Estuaries and nutrients. Estuaries 5 (4), 326. https://doi.org/10.2307/1351759. Mao, Z., Chen, J., Pan, D., Tao, B., Zhu, Q., 2012. A regional remote sensing algorithm for total suspended matter in the East China Sea. Remote Sens. Environ. 124, 819–831. https://doi.org/10.1016/j.rse.2012.06.014. Novo, E.M., Hansom, J.D., Curran, P.J., 1989. The effect of sediment type on the relationship between reflectance and suspended sediment concentration. Int. J. Remote Sens. 10 (7), 1283–1289. https://doi.org/10.1080/01431168908903967. Pan, Y., Shen, F., Verhoef, W., 2017. An improved spectral optimization algorithm for atmospheric correction over turbid coastal waters: a case study from the Changjiang (Yangtze) estuary and the adjacent coast. Remote Sens. Environ. 191, 197–214. https://doi.org/10.1016/j.rse.2017.01.013. Roberto, Barbini, Francesco, Colao, Roberta, Fantoni, Luca, Fiorani, G Okladnikov, I., Antonio, Palucci, 2005. Comparison of SeaWiFS, MODIS‐Terra and MODIS‐Aqua in the Southern Ocean. Int. J. Remote Sens. 26 (22), 2471–2478. Sá, C., D’Alimonte, D., Brito, A.C., Kajiyama, T., Mendes, C.R., Vitorino, J., et al., 2015. Validation of standard and alternative satellite ocean-color chlorophyll products off Western Iberia. Remote Sens. Environ. 168, 403–419. https://doi.org/10.1016/j.rse. 2015.07.018. Schlünz, B., Schneider, R.R., 2000. Transport of terrestrial organic carbon to the oceans by rivers: Re-estimating flux- and burial rates. Int. J. Earth Sci. 88 (4), 599–606. https://doi.org/10.1007/s005310050290. Schoellhamer, D.H., 2002. Variability of suspended-sediment concentration at tidal to annual time scales in San Francisco Bay, USA. Cont. Shelf Res. 22 (11–13), 1857–1866. https://doi.org/10.1016/S0278-4343(02)00042-0. Siswanto, E., Tang, J., Yamaguchi, H., 2011. Empirical ocean-color algorithms to retrieve chlorophyll- a, total suspended matter, and colored dissolved organic matter
absorption coefficient in the Yellow and East China Seas. J. Oceanogr. 67, 627–650. https://doi.org/10.1007/s10872-011-0062-z. Su, J., Wang, K., 1989. Changjiang River plume and suspended sediment transport in Hangzhou. Cont. Shelf Res. 9 (1), 93–111. https://doi.org/10.1016/0278-4343(89) 90085-X. Tang, D.L., Ni, I.H., Müller-Karger, F.E., Liu, Z.J., 1998. Analysis of annual and spatial patterns of CZCS-derived pigment concentration on the continental shelf of China. Cont. Shelf Res. 18 (12), 1493–1515. https://doi.org/10.1016/S0278-4343(98) 00039-9. Wang, F., Zhou, B., Xu, J., Song, L., Wang, X., 2009. Application of neural network and MODIS 250 m imagery for estimating suspended sediments concentration in Hangzhou Bay, China. Environ. Geol. 56 (6), 1093–1101. https://doi.org/10.1007/ s00254-008-1209-0. Wu, W., Yang, Z., Tian, B., Huang, Y., Zhou, Y., Zhang, T., 2018. Estuarine, Coastal and Shelf Science Impacts of coastal reclamation on wetlands : loss, resilience, and sustainable management. Estuar. Coast. Shelf Sci. 210 (July (2017)), 153–161. https:// doi.org/10.1016/j.ecss.2018.06.013. Xie, D., Wang, Z., Gao, S., De Vriend, H.J., 2009. Modeling the tidal channel morphodynamics in a macro-tidal embayment, Hangzhou Bay. China. Continental Shelf Research 29 (15), 1757–1767. https://doi.org/10.1016/j.csr.2009.03.009. Yang, X., Mao, Z., 2016. Using GOCI retrieval data to initialize and validate a sediment transport model for monitoring diurnal. Water 8 (3), 108–125. https://doi.org/10. 3390/w8030108. Yeshaneh, E., Eder, A., Blöschl, G., 2014. Temporal variation of suspended sediment transport in the Koga catchment, North Western Ethiopia and environmental implications. Hydrol. Process. 28 (24), 5972–5984. https://doi.org/10.1002/hyp. 10090. Zhang, J., 1999. Heavy metal compositions of suspended sediments in the Changjiang (Yangtze River) estuary: significance of riverine transport to the ocean. Cont. Shelf Res. 19 (12), 1521–1543. https://doi.org/10.1016/S0278-4343(99)00029-1. Zhang, X., Hu, F., Zhang, J., Tang, D., 2014. Northward drift of suspended sediments in the Yangtze estuary in spring. Int. J. Remote Sens. 35 (11–12), 4114–4126. https:// doi.org/10.1080/01431161.2014.916455.
13