Time-Series Analysis of Surface-Water Quality in Xiong’an New Area, 2016–2019

The Xiong'an New Area was established by the Chinese government on 1 April 2017 as a new district in the Hebei Province, and soon after its establishment, investigations on the waterbodies of the district began to initiate governance. In this study, data from 105 Sentinel-2 cloudless images for the period April 2016–September 2019 were analysed to understand the changes in the concentrations of chlorophyll-a and water turbidity in surface waterbodies of the Xiong'an New Area. The results suggest that (1) the water quality shows an overall improving trend owing to surface-water pollution control measures by the Chinese government in the region; (2) the turbidity of the surface waterbody declined gradually particularly after the spring of 2017; (3) the concentration of chlorophyll-a shows a weak decreasing trend, with a low value in spring and a high value in summer with a four-year average of 31.3 μg/L; and (4) by comparing the results of the water quality of two ponds inside and outside the Lake Baiyangdian and the overall water quality of Xiong'an New Area, the changes in turbidity and chlorophyll-a concentration during the study period were consistent.


Introduction
On 1 April 2017, the Central Committee of the Communist Party of China and the State Council established a new state-level district called the Xiong'an New Area in the Hebei Province. The surface water of the Xiong'an New Area is dominated by the Lake Baiyangdian (Liu et al. 2006;Su et al. 2011;Xu et al. 2012;Zhao et al. 2012;Li et al. 2017a, b;Wang et al. 2018;Song et al. 2018), and its distribution is relatively small and broken. The area has many sewage pits and ponds, which impair the construction of the Xiong'an New Area. Therefore, after the district was established, investigations on and treatments of these small waterbodies began. As of November 2018, a total of 810 polluted pools have been inspected in the three counties of the Xiong'an New Area . The Chinese government has invested a lot of human and financial resources in governance, and there is an urgent need to evaluate the governance effects of the Chinese government.
Remote sensing has the advantages of large-scale applicability, high speed, and traceable history (Wei et al. 2020). The use of remote sensing for monitoring the surface-water quality of the Xiong'an New Area can quickly and accurately reflect the effect of treatment on the sewage treatment ponds. Because the surface waterbodies in this area are relatively small and fragmented, it is necessary to select a satellite image with a high spatial resolution for monitoring them (Brezonik et al. 2005;Vincent et al. 2004). In addition, for sufficient time-monitoring frequency, it is necessary to select images with a high temporal resolution (Lambin et al. 1994). Therefore, in this study, we used the images obtained from the multispectral imager of the Sentinel-2 twin satellites. Satellites A and B can cover 13 spectral bands with spatial resolutions of 10, 20, and 60 m and widths of up to 290 km. One satellite has a revisit period of 10 days, and both satellites have a complementary revisit period of 5 days each (Drusch et al. 2012;Chen et al. 2017;Kutser et al. 2016;Liu et al. 2017;Toming et al. 2016). In addition, Sentinel-2 datasets include information from three bands (central wavelengths of 705, 740, and 783 nm) in the red edge range ( Vanhellemont et al. 2019), which is more advantageous than other satellites in monitoring chlorophyll-a content in water.
To use the Sentinel-2 data for monitoring surface-water quality, the images must first be preprocessed, which involves atmospheric correction. Kuhn et al. (2019) performed comparisons among three atmospheric-correction methods, i.e. LaSRC, SeaDAS, and ACOLITE, and showed that in most cases, the three atmospheric-correction methods will produce the same order of magnitude of spectrum. However, for water applications, LaSRC is not as effective as SeaDAS and ACOLITE because the remotesensing reflectivity produced by LaSRC is 36% higher than those of ACOLITE and SeaDAS. Caballero et al. (2018) analysed various satellite data using ACOLITE, Sen2cor, and QUAC and showed that although ACOLITE can be applied to Sentinel-2 images, it tends to mask some small and broken waterbodies, and the atmospheric-correction results cannot be obtained. Sola et al. (2018) compared four atmospheric-correction methods (Sen2cor, MAJA, Icor, and 6S) based on Sentinel-2 images and measured the obtained spectra. The results obtained from these methods did not differ significantly and showed high accuracy with an R 2 value of [ 0.90. The root-mean-square error (RMSE) is less than 0.04 sr -1 . Warren et al. (2019) compared six kinds of atmospheric-correction algorithms (ACOLITE, C2RCC, Icor, l2gen, Polymer, and Sen2cor) available for Sentinel-2. The results show that Sen2cor has better accuracy. In summary, Sen2cor is a good atmospheric-correction method for Sentinel-2 data.
Many scholars, both domestically and internationally, have studied the remote-sensing retrieval of inland waterquality parameters (Chowdhury et al. 2014;Rostami et al. 2018). However, there are relatively few studies on the retrieval of water-quality parameters based on Sentinel-2 data. Based on the Sentinel-2 and Sentinel-3 satellites, Li et al. (2017a, b) solved the problem of unsuccessful retrieval of chlorophyll in some turbid waterbodies of Poyang Lake to a certain extent. Xu et al. (2019) proposed a spectral space-segmentation-guided ensemble method based on Sentinel-2 data and compared them with three separate empirical algorithms to better estimate the concentration of chlorophyll-a. Huovinen et al. (2019) studied the seasonal trends (long-term average) of the surfacewater temperature, turbidity, and chlorophyll-a of Lake Panguipulli in central and southern Chile through Sentinel-2 and Landsat images. The verification of the measured data showed that the turbidity and chlorophyll-a concentration of the lake appeared low in summer.
Moreover, some scholars studied the evolution of the waterbody in the Xiong'an New Area and the changes in wetlands in Lake Baiyangdian. Song et al. (2018) analysed the detailed evolution of the flooded area in the Xiong'an New Area based on the long-sequence Landsat images from 1984 to 2016 including inter-annual, seasonal, and abrupt changes. The study has shown that the Lake Baiyangdian waterbody degradation is closely related to many aspects of human activities including dam construction, groundwater extraction, agricultural irrigation, tourism, and aquaculture. Zhu et al. (2019) analysed the spatial and temporal characteristics of changes in the wetlands of Lake Baiyangdian over the past 43 years based on 10 Landsat satellite remote-sensing images from 1975 to 2018 supplemented by Gaofen-2 (GF-2) satellite remotesensing images of 2017-2018. There is still a lack of research on the water quality in the Xiong'an New Area.
Most retrieval studies have shown that the water-quality parameters are affected by the spatial resolution of the images, and most of them are aimed at large waterbodies such as lakes and reservoirs. Only few studies have been conducted on time-series changes of water quality of small and broken inland waterbodies. Therefore, in this study, the chlorophyll-a concentration and turbidity data of the surface waterbodies of the Xiong'an New Area from 2016 to 2019 were retrieved from higher-resolution Sentinel-2 data, and the time-series changes in water quality were statistically analysed.

Research Area and Measured Data
The Xiong'an New Area is located in Baoding City, Hebei Province, China. The planning samples cover three small counties, namely Xiong County, Rongcheng, and Anxin, and some surrounding areas in the Hebei Province with an area of approximately 1563.31 km 2 . The dominant surface waterbody in the new area is Lake Baiyangdian with some ponds and other waterbodies scattered outside the area of the lake. On 1 May 2019, a quasi-synchronous transit experiment was performed on remote-sensing data obtained from the Sentinel-2 satellites of the Xiong'an New Area. A dataset of 29 samples was obtained, which was used to establish a retrieval algorithm of chlorophylla concentration and water turbidity. On 5 October 2018, a synchronous transit experiment was performed on the remote-sensing data obtained from the Sentinel-2 satellites of the Xiong'an New Area, and a total of nine samples were obtained. Among them, the waterbody from which two samples were obtained was considerably small to be clearly identified on the image. Therefore, these two samples were eliminated, and finally the data from seven samples were used to test the retrieval algorithm of chlorophyll-a concentration and water turbidity. The location of the study area and the distribution of the experimental samples are shown in Fig. 1.

Screening of Remote-Sensing Image Data
This study uses Sentinel-2 data of the spring (April to May), summer (June-August), and autumn (September-October) of 2016, 2017, 2018, and 2019. The first frost day in the Xiong'an New Area is usually 19 October, and the frost normally lasts until 12 April; this was the basis for selecting non-icing data in this paper. In addition, images with less cloudiness were selected for the analyses. As the Xiong'an New Area is at the junction of three Sentinel-2 images, the Sentinel Application Platform (SNAP) recommended by the European Space Agency (ESA) must be used to mosaic the cloudless images taken on the same day, and hence, only less amount of data could meet the requirements. Finally, a total of 102 images of 34 phases were screened out. The specific image-selection schedule is shown in Table 1. The 2016-2018 Sentinel-2 data downloaded from ESA in this article are L1C-level data, and the 2019 Sentinel-2 data are L2A-level data. The L1C-level data are geometrically corrected orthophotos, and the L2Alevel data are presented from L1C atmospherically corrected images.  Table 1 shows that when analysing temporal and spatial changes of water quality in each season, the pixel values of the same position of all the images in the season are averaged. The average image represents the water-quality status of that season; if there is only one phase in a season, it represents the water quality of that particular season by default.

Obtaining Measured Data
A field-SpecR3 portable ground spectrometer of the American ASD was used, and the water-spectrum measurement method proposed by NASA Marine Optical Specifications (Mueller et al. 2003) and Tang et al. (2004) was used to collect the remote-sensing reflectance of each sampling point. Furthermore, to better analyse the band characteristics of Sentinel-2 data and evaluate the effect of atmospheric correction, the equivalent reflectance measured on 5 October 2018 must be calculated to obtain the equivalent spectral reflectance corresponding to the Sentinel-2 band. The spectral equivalent can be calculated as where F 0 is the solar irradiance outside the atmosphere; q(k) is the reflectance at each wavelength on the ground, which is the reflectance spectrum measured on our ground; and f(k) is the spectral response function of each band of the remote sensor. The measured spectral reflectance and Sentinel-2 equivalent spectral reflectance are shown in Fig. 2. Turbidity (NTU) was measured using the Orion AQ3010 turbidity meter (Thermo Scientific, USA). The hot ethanol method was used to measure the concentration of chlorophyll-a in water. A GF/F glass fibre filter membrane with a pore size of 0.7 lm and diameter of 47 mm was used to filter the water samples collected in the field experiment. Then, the used folded filter membrane was put it in a test tube for freezing for 48 h. Later, it was taken out of the refrigerator and approximately 4 mL of hot ethanol was added to the test tube with the filter membrane and heated to 85°C in a water bath for 4-6 h and the extract was filtered with another 0.7-lm membrane with a pore diameter of 25 mm and 90% ethanol was added to the filtrate to make up a volume of 10 ml. Then, the filtered extract was poured into a cuvette and put it into a Shimadzu UV2550 spectrophotometer to measure the absorbances at 665 nm at E665 and 750 nm at E750. After that 1 mol/L of hydrochloric acid was added to the cuvette for acidification, covered, and shaken well, and finally, the absorbances after 1 min at A665 and A750 at 665 nm and 750 nm, respectively, were measured. The concentration of chlorophyll-a was calculated according to the absorbances.

Accuracy Evaluation Index
The atmospheric-correction and water-quality parameter retrieval use the mean relative error (MRE) and RMSE for accuracy evaluation: The standard deviation (SD) of the dispersion of statistical values of the water-quality retrieval results can be calculated as follows: In Eqs. (2-4), n represents the number of points, R i represents the value of water-quality parameter after retrieval or atmospheric-corrected reflectance value, and R j represents the actual measurement. The water-quality parameter or measured equivalent spectral reflectance value, K, represents the number of samples, x a represents the a-th sample, and l represents the average value of the samples.

Evaluation of Atmospheric Correction and Effect
To quickly perform atmospheric correction on the Sentinel-2 data, the ESA released a plug-in, Sen2cor, which specialises in the production of L2A-level data. Sen2cor can perform atmospheric correction on Sentinel 2A/2B to obtain L2A-level data and was used in this study. To ensure the accuracy of the atmospheric-correction results, the spectral reflectance of each water sample after equivalence is fitted to the remote-sensing reflectance of Sentinel-2 satellite image after atmospheric correction. The reflectance of each band after fitting is shown in Fig. 3. The measured spectral reflectance of seven waterbodies collected in the Xiong'an New Area on 5 October 2018 and the atmospheric correction of the remote-sensing reflectance of the image were subjected to corresponding regression analysis. To analyse the accuracy of Sen2cor atmospheric correction, the measured spectral reflectance of the waterbody and the results of atmospheric correction were used to extract the remote-sensing reflectance in the Band2-Band8A range for regression analysis. The results are shown in Fig. 4. Figure 4 shows that the results obtained by Sen2cor have a high correlation with the measured values, and the data are more concentrated overall. When the reflectivity is 0.004-0.008 sr -1 , the data obtained are relatively discrete. This may be because the error may be larger in the band with the largest rate of change in the spectral curve. The determination coefficient R 2 of the remote-sensing reflectance results obtained by the Sen2cor atmospheric-correction method is 0.85, the RMSE is 0.0011 sr -1 , and the average relative error is 29.51%.

Water-Extraction Method and Result Evaluation
The waterbody-extraction algorithm used in this study uses the newly proposed relative water index (RWI) (Wu et al. 2020), which is formulated as follows: RWI ¼ ðR rs ð560Þ þ R rs ð705ÞÞ À ðR rs ð842Þ þ R rs ð865ÞþR rs ð2190ÞÞ R rs ð560Þ þ R rs ð705Þ þ R rs ð842Þ þ R rs ð865Þ þ R rs ð2190Þ where R rs (560), R rs (705), R rs (842), R rs (865), and R rs (2190) are the remote-sensing reflectance of Sentinel-2 bands 3, 5, 8, 8a, and 12, respectively. The waterbody boundary extracted by the RWI algorithm shows a better extraction effect than the other methods, and it can eliminate the influence of mountains and building shadows, cloud shadows, and mixed pixels to a certain extent. The water-boundary effect extracted by RWI is shown in Fig. 5a. The RWI was used to extract waterbodies from 35 time-phase mosaic images selected from 2016 to 2019, and each of the waterbodies differed with respect to area. This study was aimed at determining the temporal changes of surface-water quality in the Xiong'an New Area and did not consider the changes in the area of the waterbodies. Therefore, the results of water extraction were intersected. This method can remove the influence of changes in the area of a waterbody and in the case where a part of the image, which could be a building shadow, is mistaken for a waterbody, such an error can be eliminated by obtaining the intersection of the water surface pixels. Finally, the waterbody with an unchanged area within Xiong'an New Area was obtained from 2016 to 2019. The waterbodyextraction results are shown in Fig. 5b.

Water-Quality Parameter Retrieval Algorithm and Accuracy Evaluation
Turbidity Retrieval Algorithm In the near-infrared and visible light bands, the water turbidity increases along with the spectral reflectance. The turbidity level and spectral reflectance of the water have a linear correlation in the red band (Lin et al. 2004). In this study, blue, green, and red bands were selected for performing turbidity retrieval, and the measured data from May 2019 were used to construct a turbidity retrieval algorithm, which can be expressed as follows:  where R rs (490), R rs (560), and R rs (665) correspond to the blue, green, and red bands of the Sentinel data, respectively.

Chlorophyll-A Retrieval Algorithm
The absorption coefficient spectrum of chlorophyll-a in an inland waterbody mainly has two absorption peaks that are usually located near 440 and 675 nm. The peak at 440 nm is greatly affected by the absorption of non-pigmented suspended solids and yellow substances, while that at 675 nm is less affected by other water elements; therefore, the characteristic spectrum near 675 nm was selected in this study corresponding to the fifth band of the red edge band of Sentinel-2 data (Gilerson et al. 2010;Matthews 2011;Odermatt et al. 2012). The chlorophyll-a concentration retrieval algorithm was established using the measured data of 29 samples of Lake Baiyangdian in May 2019 as follows: where C chla represents the retrieved chlorophyll-a concentration [lg/L], and R rs (665) and R rs (705) represent the red bands of the Sentinel-2 data and the fifth band, respectively.

Accuracy of Results of the Water-Quality Parameter Retrieval Algorithm
To ensure the reliabilities of remote-sensing monitoring and retrieval algorithms, the measured turbidity and chlorophyll-a concentration of the seven water samples in the Xiong'an New Area obtained on 5th October 2018 were regressed. In addition, the turbidity and chlorophylla concentration values obtained through the retrieval showed a significant correlation between the two sets of data. When the correlation coefficient between the measured and retrieved turbidity values of the waterbody reached 0.89, the MRE and RMSE were 23.31% and 3.18 NTU, respectively. When the correlation coefficient between the measured and retrieved chlorophyll-a concentrations in the waterbody reached 0.96, the MRE and RMSE were 31.23% and 9.09 lg/L, respectively. The results are shown in Fig. 6.

Results
Owing to the large area of the Xiong'an New Area and the small surface waterbodies, it is difficult to show the surface water of the entire district as most of the water is distributed in the Baiyangdian area of Anxin County. The area with a larger number and density of waterbodies in the inner Lake Baiyangdian is used as the display range.
According to the results of the time-series analysis of turbidity retrieval of surface water in Xiong'an New Area from 2016 to 2019 (Fig. 7) and statistics (Fig. 8), the turbidity of surface water was higher in 2016 and spring of 2017. After that, a significant drop was observed. In addition, from April 2016 to September 2019, the turbidity gradually declined with slight fluctuations during autumn 2016 and spring 2017. The seasonal-average highest and lowest values of turbidity were 40 NTU and 8.3 NTU in spring of 2016 and 2019, respectively. The four-year average value of turbidity was 22 NTU, and the SD was 9.5 NTU. No obvious change rule for turbidity in space existed, and the distributions of high and low turbidity were relatively uniform.
The time-series results of chlorophyll-a concentration retrieval in the surface water of the Xiong'an New Area from 2016 to 2019 (Fig. 8) and the statistical chart (Fig. 9) show that the chlorophyll-a concentration is low in spring and high in summer, followed by autumn. The chlorophylla concentration in the waterbody generally showed a weak downward trend with a low value in spring and a high value in summer. The four-year average was 31.3 lg/L, and the SD was 2.5 lg/L. No obvious change pattern for chlorophyll-a concentration in space was observed, and the distributions of high and low values of chlorophyll-a concentration were relatively uniform.

Analysis of Impact of Policies on Surface-Water Quality in the Xiong'an New Area
The mean values of the chlorophyll-a concentration and turbidity of the whole image were calculated for each date selected in this study and plotted as a change curve as shown in Fig. 10. The figure shows that since the establishment of the Xiong'an New Area, the environmental governance work has been fruitful.
Earlier, many industries such as paper, leather, shoemaking, printing and dyeing, rubber, down and feather, and workshops existed in the upper reaches of Lake Baiyangdian with high levels of water consumption, energy consumption, pollution and sewage discharge along with rural sewage discharge and garbage accumulation along the river. With the economic development of the villages along the river, the unreasonable use of pesticides and chemical fertilisers required by the planting industry and the accumulation of bait and animal faeces from the breeding industry increased the agricultural non-point source of pollution to the nearby rivers. Increase in population led to an increase in cultivation area in the region causing largescale reclamation of cultivated land leading to water loss, shallow river siltation, soil fertility decline, and poor surface-water quality. Since 1958, many large-and mediumsized reservoirs were built in the upper reaches of Lake Baiyangdian. Due to the impoundment of reservoirs, many rivers have lost their natural water supply sources and are in a state of disconnection. The imbalance between economic and social development and water resources in the basin has led to the pollution of Lake Baiyangdian. After 2002, The Lake Baiyangdian Natural Wetland Reserve has become a tourist attraction again. The development of tourism has driven the surrounding industries. However, due to the lack of relevant laws and regulations and a standard system of ecological environment protection, the pollution caused by tourism is increased gradually.
Since the establishment of the Xiong'an New Area, the government has taken a series of measures including shutting down or rectifying large-scale high-pollution and high-emission enterprises, strengthening the treatment of domestic sewage and garbage, and improving farming methods in and around the lake region, thereby promoting green and pollution-free agriculture, reducing the use of pesticides and fertilisers and pollutant emissions, and saving water resources. Designated breeding areas, optimised breeding mode, and the removal of cages, net fences, and net enclosures, earthwork and enclosure cleanup, scientific planning and management, and control on the scale and capacity of breeding have been implemented. Furthermore, sediment dredging, dismantling of water-intensive projects, and increasing the amount of water inflow into the Lake Baiyangdian have been implemented. Additionally, water-saving measures such as rational use of reclaimed water to prevent water diversion outside the river and interception of groundwater along the river have been implemented. These comprehensive measures from May to June 2017 have caused a significant decrease in turbidity in the surface waterbodies (Fig. 10), indicating that the Chinese government has achieved remarkable results in water treatment and conservation in the region. Figure 8 shows that the chlorophyll-a concentration in the Xiong'an New District during 2016-2019 is low in spring and high in summer and autumn. To understand this pattern, the authors collected historical weather data of the region through a weather-based website (http://tianqi.2345. com/wea_history/ 72,105.htm) and found that the temperature in the district was lower during spring and higher in summer and autumn. The warm-water environment can effectively promote the growth of phytoplankton and produce chlorophyll-a concentrations that are lower in spring and higher in summer and autumn. Therefore, chlorophylla concentration is proportional to the temperature in the district.

Characteristics of Chlorophyll-a Concentration Changes
Through repeated field observations and using images in May 2019, the growth of diverse types of water plants was Fig. 6 Scatterplots of the comparison between the retrieval and measured values of turbidity and chlorophyll-a concentration. a Regression analysis of the retrieved and measured turbidities and b regression analysis of the retrieved and measured chlorophyll-a concentration observed on the surface of the waterbody in Lake Baiyangdian Area as shown in Fig. 11. Figure 11a and c shows two kinds of floating-leaf water plants that are removed when the waterbody is extracted. The water grass in Fig. 11e grows underwater. When the water is extracted, it may be mistakenly divided into waterbodies. The growth of underwater plants will influence the concentration of chlorophyll-a, and the growth of water grass is affected by seasonal changes that are more significant in summer and autumn. Therefore, the change in chlorophyll-a concentration in the surface water of the region is affected by air temperature and submerged water plants.

Characteristics of Water-Quality Changes in Partial Waterbodies
The sizes of the selected waterbodies are shown in Fig. 12.
In addition, Fig. 13 shows the changes in the chlorophylla concentration and turbidity in the two areas. As shown in Fig. 13, the local waterbodies in the region show a declining trend in turbidity from 2016 to 2019. The chlorophyll-a concentration also fluctuates, i.e. it is low in spring and high in summer and autumn in the waterbodies outside the Lake Baiyangdian area. During 2016-2019, the turbidity shows an overall downward trend; however, this trend is not obvious. The change in chlorophyll-a concentration is relatively gentle with an overall downward trend showing a low concentration because although the waterbody in the lake is divided into small sections, the water channels are connected and thus can be regarded as a single unit. Therefore, the trends in the quality of the waterbody (Lake Baiyangdian) and the overall quality of all the surface waterbodies in the Xiong'an New Area are the same. The waterbody outside the Lake Baiyangdian area is an independent pond. Water quality fluctuations are highly susceptible to environmental influences. Therefore, although turbidity fluctuates, no obvious downward trend is observed. Moreover, because the algae content in the pond is relatively stable, it is not strongly affected by weather. Therefore, the content of chlorophyll-a during 2016-2019 is relatively stable, and the change is not obvious.

Comparison with Other Researches
The main advantages of this article are as follows. (1) The article analyses the surface water quality of Xiong'an New Area in time series based on remote-sensing data and not only analyses the reasons for the change of Lake Baiyangdian water body area (Liu et al. 2006;Li et al. 2017a, b;Wang et al. 2018;Song et al. 2018). (2) The retrieval algorithm for chlorophyll-a and turbidity suitable for Xiong'an New Area was constructed. The article compares the chlorophyll-a and turbidity retrieval algorithms proposed by other scholars in recent years and verifies the retrieval accuracy of each method based on the R rs of Sentinel-2 image, as shown in Table 2 and 3. It can be seen from the table that the chlorophyll-a and turbidity retrieval algorithm proposed in this paper has higher accuracy. According to the retrieval results of chlorophylla in this article, we calculated the nutrient index TSI (Carlson. 1977). The results of TSI show that most surface water bodies in Xiong'an New Area have serious eutrophication. The chlorophyll-a and turbidity retrieval algorithms proposed in the article have good applicability to this type of water bodies and can provide a certain reference basis for subsequent researchers.

Conclusion
The surface waterbodies in the Xiong'an New Area were chosen for research as this region acts as the fulcrum for easing the non-capital administrative functions of Beijing, the capital of China, and plays a vital role in the coordinated development of the Beijing-Tianjin-Hebei economic belt, which is currently the national focus. Multispectral Sentinel-2 data, field experiments, and spectral data of the surface waterbodies of the Xiong'an New Area, and various water-quality parameters were used to analyse the   The changes in waterquality parameters outside the Lake Baiyangdian region are not obvious. (4) Based on the turbidity and chlorophylla concentrations from 2016 to 2019, the water quality in this area shows an overall improvement, which is in line with the mitigating measures of the Chinese government for the region. This paper provides a feasible theoretical basis for the inversion of surface water quality parameters in areas similar to Xiong'an New Area and will serve as a reference for other researchers to carry out similar monitoring studies. We provide a reasonable scheme to analyse water quality of relatively small surface water areas and facilitate effective technical support to and environmental management of small watersheds and discontinuous water areas through scientific decision-making. Furthermore, our approach can be used to improve the management of urban water environments and conduct quality-based classification of small surface water bodies in China and other parts of the world.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this  Accuracy matrices for Chla estimates using satellite R rs are listed in the table. G = R rs (560), R = R rs (665), RE = R rs (705), N = R rs (842)  (490), G = R rs (560), R = R rs (665), RE = R rs (705), N = R rs (842) article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/.