Dataset of Total Suspended Solids in the Jiaozhou Bay Retrieved from Landsat Images (1984–2020)

Huang, J.¹  Zhang, X.¹  Zhao, Y. F.^2*

1. College of Geodesy and Geomatics, Shandong University of Science and Technology, Qingdao 266590, China;

2. Jiaozhou Bay Marine Ecosystem Research Station, Chinese Academy of Sciences, Qingdao 266071, China

Abstract: The total suspended solids (TSS) concentration (mg/L) is an important water environment parameter in coastal waters. The dataset of total suspended solids in the Jiaozhou Bay (1984–2020) was retrieved from the Landsat TM/ETM /OLI images between 1984 and 2020 with a quantitative retrieval model. In total, 23 pairs of remote sensing reflectance and TSS concentration data were used for model development with the Leave-One-Out Cross-Validation method. Judged by R² and RMSE, the model with the best performance was selected for TSS retrieval. The dataset includes: (1) the geographic boundary data of Jiaozhou Bay in 1984 and 2020; (2) 318 images of TSS concentration with spatial resolution of 30 m. The dataset was archived in .shp and .tif data formats and consisted of 341 data files with a data size of 236 MB (compressed into one single file of 232 MB). This dataset provided data support for watercolor remote sensing research, water quality evaluation, and water environmental protection.

Keywords: Jiaozhou Bay; Landsat; total suspended solids; quantitative retrieval; long-term monitoring

DOI: https://doi.org/10.3974/geodp.2022.03.14

CSTR: https://cstr.escience.org.cn/CSTR:20146.14.2022.03.14

Dataset Availability Statement:

The dataset supporting this paper was published and is accessible through the Digital Journal of Global Change Data Repository at: https://doi.org/10.3974/geodb.2022.02.09.V1 or https://cstr.escience.org.cn/CSTR:20146.11.2022.02.09.V1.

1 Introduction

Jiaozhou Bay (JZB) is a semi-closed bay with an approximate trumpet shape, and its exit faces east. It is located in the middle of the Yellow Sea in China, close to the southern coast of Shandong Peninsula (Figure 1). JZB covers an area of 370 km², with narrow entrances, deep waterways, low turbidity, weak tides, and low waves. As a natural port in Qingdao, it promotes the development of Qingdao??s transportation industry^[1]. The mouth of the bay is bounded by the southern end of the Tuandao and the northern end of Xuejiadao and is about 2.5 km wide. The width of the central part of JZB is about 27.8 km from east to west, and the maximum length from north to south is about 33.3 km. The coastline is about 163 km long, and the tidal flat area is about 125 km². However, like most bays bordering urbanized areas, JZB coastal areas are also undergoing major changes to resolve the conflict between land demand and supply. A previous study showed that the area of JZB had decreased by 36.6% compared to that in 1935, and the length of the coastline was shortened by 30.4%^[2].

As one of the important parameters in coastal water environment, total suspended solids (TSS) is fundamental to many physical and chemical processes in the waters. On the one hand, the TSS concentration as a common water quality parameter plays an important role in indicating the pollution degree and estimating the soil erosion. On the other hand, TSS reduces the light energy entering the water body, restricting the growth of aquatic plants and the primary productivity of plankton^[3].

In recent years, with the development of industry and growth in the standard of living, water resources such as oceans, rivers, and lakes have been contaminated. Monitoring of the TSS concentration has become an important part of water environmental improvement. The traditional method of TSS monitoring heavily relies on laboratory measurement of field sampling^[4], which is time-consuming and labor-intensive. Especially for some hard-to-reach places, the number of field sampling sites are limited, and periodic and comprehensive measurement is difficult to achieve. Remote sensing has the advantages of saving time, economy, and repeat observation. It can realize real-time, macroscopic, and dynamic water environment monitoring for large bodies of water^[5].

Although the variation of TSS in JZB and surrounding waters has always been a concern to scientists and managers, long-term TSS monitoring results in JZB are still lacking. Therefore, in this paper, a quantitative retrieval model of TSS concentration was established based on field-measured data and Landsat data. Then, the JZB TSS dataset from 1984 to 2020 was obtained, which provided data support for watercolor remote sensing and water environment monitoring.

2 Metadata of the Dataset

The metadata of Landsat image-based retrieval dataset of total suspended solids in the Jiaozhou Bay (1984-2020)^[6] is summarized in Table 1. It includes the dataset full name, short name, authors, year of the dataset, temporal resolution, spatial resolution, data format, data size, data files, data publisher, and data sharing policy, etc.

Table 1  Metadata summary of the Landsat image-based retrieval dataset of total suspended solids in the Jiaozhou Bay (1984–2020)

3 Methods   

3.1 Jiaozhou Bay Vector Data

In this study, the geographic boundary data of Jiaozhou Bay in 1984 and 2020 were obtained using ArcGIS. Due to the influence of land reclamation in recent years, the area of JZB has decreased, and the boundary is changing, so the 2020 boundary was used. The boundary of JZB in 2020 has a certain distance to the coast in the north, because the northern part of JZB is mostly tidal flats with shallow water depth, and the high reflectance of the sediment seriously affects the retrieval results of TSS. As a result, the areas seriously affected by sediment were excluded, and this boundary was finally determined as the ??retrieval boundary?? to minimize the impact of the tidal flat. In addition, the JZB Bridge has a significant difference in reflectance from seawater. In order to reduce the influence of the JZB Bridge on TSS retrieval, the JZB Bridge was masked.

3.2 In-Situ Data

In this study, 23 sets of field-measured data between 2011 and 2015 were collected from the JZB Marine Ecosystem Research Station (http://jzw.qdio.cas.cn/). The TSS concentration was measured gravimetrically. The water samples were first filtered through a dry preweighed 0.45 µm filter. Then, the filters were dried at 45??C for 24 hours and then reweighed. The drying process was repeated until the difference in the TSS values of consecutive weighing was less than 0.01 mg/L. Information on the location of the sample sites is shown in Figure 1.

3.3 Acquisition and Preprocessing of Landsat Data

In this study, Landsat reflectance data were collected from the Google Earth Engine (GEE) platform, including Theme Mapper (TM), Enhanced Theme Mapper Plus (ETM+), and Operating Land Imager (OLI). Images with cloud cover >10% were excluded for quality control. The atmospheric correction method for Landsat TM/ETM+ and OLI reflectance data were LEDAPS (Landsat Ecosystem Disturbance Adaptive Processing System) and LaSRC (Landsat Land Surface Reflectance Code), respectively.

The Quality Assessment (QA) band was used to detect clouds and cloud shadows on Landsat images and generate cloud mask files for cloud removal. Although TM/ETM+ and OLI have similar spectral bands, there were still differences between the image reflectance obtained by the three sensors used in this study. Therefore, a linear relationship between the reflectance of OLI and TM/ETM+ was developed using the empirical line method^[8], and the reflectance of Landsat TM/ETM+ images was adjusted to the same level as the reflectance of OLI images. On May 31, 2003, the Landsat ETM+ airborne scan line corrector (SLC) failed, resulting in the loss of data bands in subsequently acquired images. Therefore, the focal mean function was applied to fix the mistakes.

3.4 Development of TSS Quantitative Retrieval Model

Through band sensitivity analysis, it was found that the red band had the strongest correlation with the TSS concentration. Therefore, the combinations of the red band and other bands were used as indicators to analyze the correlation between the TSS concentration and remote sensing data. Table 2 shows that the band combination with the strongest correlation was (G+R)/(G/R) (G is the remote sensing reflectance of the green band, R is the remote sensing reflectance of the red band). Therefore, in this study, the band combination mentioned above

was used for the establishment of a quantitative inversion model of TSS concentration.

The model was calibrated and validated using a robust leave-one-out cross-validation approach^[9]. In total, 22 sets of remote sensing reflectance and TSS concentration data were used for modeling, and the remaining one set of data was used for validation. The above process was carried out in sequence, a total of 23 models were obtained, and the coefficient of determination (R²) and root-mean-square error (RMSE) were calculated. The model with the highest accuracy was finally selected as the retrieval model of TSS concentration. The model proposed in this study achieved the best performance, and the retrieval model was set as follows:

                                                                                                               (1)

where X was (B2 + B3)/(B2/B3) for Landsat 5 and Landsat 7 and (B3 + B4)/(B3/B4) for , while Y was the TSS concentration. B2, B3, and B4 were the reflectance associated with the Landsat channels.

4 Data Results and Validation

4.1 Data Composition

After the preprocessing, the retrieval model was applied to the Landsat images, and finally, 318 images of TSS concentration of JZB from 1984 to 2020 were obtained. The data format, spatial resolution, and unit were. tif, 30 m, and mg/L, respectively. Among them, there were 144 TM images with a time range of 1984-2011, 120 ETM+ images with a time range of 1999–2020, and 54 OLI images with a time range of 2013–2020. The image was named as follows: XXXX_120035_YYYYMMDD.tif, where XXXX represented different Landsat sensors (LT05, LE07, and LC08), 120035 represented the row and column number of the image where the JZB is located, and YYYYMMDD represented the date. The TSS concentration derived from different sensors can be mixed. In addition, this dataset also included the vector data of JZB boundaries in 1984 and 2020.

4.2 Data Results

The spatial distribution of the TSS in JZB on August 19, 2013, was taken as an example and is shown in Figure 2. The missing pixels in the image were due to cloud removal and generally did not affect usability. The TSS concentration in the JZB mainly varied in the range of 0–150 mg/L, showing a low turbidity and gradually decreasing from northwest to southeast. The TSS concentration was generally above 30 mg/L in the estuary and coastal areas but was mostly below 30 mg/L in the central sea area.

Figure 3 shows the variation of the annual average TSS in JZB inversed from the Landsat data from 1984 to 2020. The annual average TSS concentration reached a maximum value (26.94 mg/L) in 1993 and a minimum value (10.69 mg/L) in 2016. However, there was an overall decreasing trend, which was consistent with the research by Gao et al^[10].

4.3 Data Validation

Leave-one-out cross-validation was used to minimize the influence of random factors caused by the small amount of field-measured data in the TSS model development. Figure 4 showed the correlation between the measured and predicted TSS values. The model had high explanatory and predictive power. The R², RMSE, and trendline slope were 0.77, 1.82, and 1.07, respectively. The scatter points were evenly distributed on both sides of the 1:1 line, and all predicted TSS values were within the range of the measured data.

5 Discussion and Conclusion

The data format, spatial resolution, and unit were .tif, 30 m, and mg/L, respectively. The data source of the dataset was reliable, and it was mainly produced based on platforms such as GEE. Users can open it with the help of image processing software such as ArcGIS or ENVI. If other formats are required, format conversion can be performed in the software above.

It is worth noting that there may be certain uncertainties in this dataset. Firstly, there was a difference in the acquisition time between field-measured data and satellite data due to the limited number of field-measured data, inevitably introducing uncertainty to the retrieval model. Secondly, due to the lack of near-shore observation sites, the highest concentration in the field-measured data was 13 mg/L, but a small part of the TSS concentration in the retrieval results was higher than 30 mg/L. Retrieval model developed based on a narrower concentration dynamic range also introduced certain uncertainties. In addition, less remote sensing data were available in summer than in other seasons because of the cloudy summer weather in the JZB. Finally, the northern coast of the JZB was greatly affected by the high reflectance of sediments from shallow waters; therefore, the high TSS concentration in this area revealed by the retrieval model might be inconsistent with reality.

Long-term dynamic TSS datasets can effectively provide data support for management departments and accelerate water quality improvement. This dataset provided the TSS concentration products in JZB from 1984 to 2020, with a large amount of data and a wide time range. It provided important data support for the study of ocean color remote sensing such as the pattern of the temporal and spatial distribution of TSS in JZB and its influencing factors and also provided a basis for water environment monitoring and environmental protection policy formulation.

Author Contributions

Huang, J. made the overall design for the development of the dataset, participated in the model design and the writing and revision of the paper. Zhao, Y. F. collected the field-measured data and gave general guidance for the research. Zhang, X. processed basic data, participated in model development, performed data validation, and wrote data papers.

Acknowledgements

Thanks to the Jiaozhou Bay Marine Ecosystem Research Station for the field-measured data of the total suspended solid concentration in Jiaozhou Bay.

Conflicts of Interest

The authors declare no conflicts of interest.

References

[1]      Yang, G., Jiang, T., Zhao, Y., et al. Study on variation in chlorophyll-a concentration and its influencing factors of Jiaozhou Bay based on long term remote sensing images [J]. Haiyang Xuebao, 2019, 41(1): 183–190.

[2]      Yuan, Y., Jal??n-Rojas, I., Wang, X. H., et al. Design, construction and application of a regional ocean database: a case study in Jiaozhou Bay, China [J]. Limnology and Oceanography, 2019, 17(3): 210‒222.

[3]      Zhang, Y., Wu, Z., Liu, M. Thermal structure and response to long-term climatic changes in Lake Qiandaohu, a deep subtropical reservoir in China [J]. Limnology and Oceanography, 2014, 59(4): 1193‒1202.

[4]      Zhang, M. H. Distribution and seasonal variation of suspended matter in sea water of Jiaozhou Bay [J]. Studia Marina Sinica, 2000(00): 49–54.

[5]      Zhang, X., Song, Y., Chen, J. J., et al. Landsat image-based retrieval and analysis of spatiotemporal variation of total suspended solid concentration in Jiaozhou Bay, China [J]. Remote Sensing, 2021, 13(23): 4796‒4796.

[6]      Huang, J., Zhang, X., Zhao, Y., F. Landsat image-based retrieval dataset of total suspended solids in the Jiaozhou Bay (1984–2020) [J/DB/OL]. Digital Journal of Global Change Data Repository, 2022. https://doi.org/10.3974/geodb.2022.02.09.V1. https://cstr.escience.org.cn/CSTR:20146.11.2022.

02.09.V1.

[7]      GCdataPR Editorial Office. GCdataPR data sharing policy [OL]. https://doi.org/10.3974/dp.policy.2014.05 (Updated 2017).

[8]      Han, X. X., Chen, X. L., Feng, L., et al. Four decades of winter wetland changes in Poyang Lake based on Landsat observations between 1973 and 2013 [J]. Remote Sensing of Environment, 2015, 156: 426‒437.

[9]      Huang, J., Wu, M., Cui, T. W., et al. Quantifying DOC and its controlling factors in major Arctic rivers during ice-free conditions using Sentinel-2 data [J]. Remote Sensing, 2019, 11(24): 2904.

[10]   Gao, G., Wang, X., Bao, X., et al. The impacts of land reclamation on suspended-sediment dynamics in Jiaozhou Bay, Qingdao, China [J]. Estuarine Coastal and Shelf Science, 2018, 206: 61‒75.