A multi-sensor framework in google earth engine for spatio-temporal trend analysis of water quality parameters in Anzali lagoon

doi:10.21203/rs.3.rs-3335434/v1

Download PDF

Research Article

A multi-sensor framework in google earth engine for spatio-temporal trend analysis of water quality parameters in Anzali lagoon

https://doi.org/10.21203/rs.3.rs-3335434/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

The Anzali wetland is one of the internationally known freshwater wetlands which provides habitats for many plants and migratory birds, and plays the role of a treatment plant for running water into the Caspian Sea. However, its lagoon is getting shallower due to the inflow of sediment and deposition of organic detritus, and its water quality is deteriorating due to the inflow of wastewater and solid waste from neighboring cities. Therefore, monitoring of water quality parameters (WQP) in Anzali Lagoon is deemed a priority for protection and prevention of its degradation. The current study investigates the environmental condition of the wetland between 2014 to 2021 using remote sensing technology. A multi-sensor framework was developed in google earth engine for spatio-temporal trend analysis of water quality parameters in Anzali lagoon. Water quality parameters incluing TSS, TDS, pH and nitrate were derived using Landsat satellite data based on the calibrated regression equations introduced for this wetland. Spatio-temporal variations of WQP was depicted to determine the critical points and evaluate the trend analysis. The results showed that the WQP fluctuate significantly over time. The spatial distributions indicated that the critical points of the lagoon are changing from the Western to the center and Eastern parts. The comparison of the WQP parameters with the environmental standards confirmed that the lagoon condition is in critical stage, especially in terms of TSS. Nitrate values showed that 50% of the lagoon area exceeds the environmental standards, which indicates the nutritional status. The investigated WQP has had a significant positive trend in most areas, so the lagoon is under high environmental risk and unfriendly human activities should be controlled.

Landsat

Mapping

Regression model

Trend analysis

Water quality

Wetlands are one of important ecosystems on earth that play a key role in filtering polluted water and provide habitats for many plants and animals. Wetlands perform a range of environmental, socio and economic benefits to the local communities. The coastal wetland functions as an important nutrient cycling capacity source to protect water quality [Schmidt and Skidmore, 2003]. Increasing human activities and their negative consequences cause pollution, degradation and loss of wetlands. The main threats that the wetlands struggle with are unsustainable and non-eco friendly human activities. Studies show that 87% of the wetlands have been lost in the last 300 years (BioLearn, 2021). The main reason for this loss is to open land for settlement, industry and agricultural activities. Water quality monitoring is a fundamental tool in the management of water resources. Environmental water quality monitoring aims to provide the data required for safeguarding the environment against adverse effects from anthropogenic point and nonpoint sources. Since, the ground survey of water quality parameter over a wide geographical area are very costly and time-consuming, use of remote sensing techniques has been expanded to obtain the required information using sensors without any physical contact based on the reflected and emitted energies Although, measuring the WQP from space is considered as a developing approach, many researchers have reported promising results after using satellite imagery [Yang, 2022]. The reports indicate the accuracy of measurements varies from case to case, and it depends on many factors (i.e.). But in general, high accuracy has been reported for measuring suspended solids [Pirali Zefrehei et al., 2021], turbidity [Weiqi et al., 2008], total nitrogen and phosphour [Sun, 2014], pH [Pereira et al., 2020], total dissolved solids [Aljoborey and Abdulhay, 2019], chloride [Amirsalari et al. 2013], organic matter [Li et al., 2018], chlorophyII [Palmer et al., 2015] and dissolved oxygen [El Din et al., 2017].

Weiqi et al., (2008) measured the values of turbidity, total nitrogen, ammonium, nitrate, total phosphorus and soluble phosphorus in Guanting Reservoir, Beijing, using Landsat 5 TM satellite images. They concluded that turbidity and nitrate can be measured with lower relative error, however, higher relative errors reported for total nitrogen, soluble phosphorus, ammonium nitrogen and total phosphorus, respectively.

Toming et al., (2016) implemented the remote sensing to evaluate the WQP in eight lakes in Estonia. They indicated that chlorophyll a and soluble organic matter can be estimated with higher accuracy. The reported correlation coefficient (R²) between the observed and estimated values for chlorophyll a and soluble organic matter were 0.83, 0.92, and 0.52, respectively. Hasab et al., (2020) used Landsat 8 satellite images to estimate the values of electro coductivity (EC), SO₄ and CaCO₃ within the Al-Hawizeh Marsh and evaluate the spatial distributions of these parameters during the two seasons in the year 2017. The values of R² between the observed and predicted values for the EC, SO₄ and CaCO₃ during the two seasons were 0.95, 0.96 and 0.92, respectively. Kapalanga et al., (2021) introduced regression equations for deriving the WQP from satellite images in the Olushandja Dam in Namibia. The correlation coefficients between the observed and predicted were reported as 0.76, 0.798, 0.91, 0.28 and 0.85 for turbidity, TN, TP, TSS and total algae count, respectively. They recommended satellite imagery for frequent and continuous monitoring of WQP in the Olushandja Dam. Sarwar et al., (2021) applied RS and GIS for monitoring water quality of Matta Tehsil District in Pakistan. They considered varieties of water parameters consist of pH, EC, DO, salinity, alkalinity, TDS, Cl^-1, SO₄^-1 and BOD, and concluded that remote sensing can be implemented as a reliable tools for mapping the water quality zones.

Due to the importance role of Anzali lagoon in the ecosystem of the Caspian region, successful studies have been conducted on measuring the WQP using ground survey (Pirali Zefrehei et al., 2019, 2021). In a rare study, Navabian et al. [2019] studied the feasibility of monitoring WQP in Anzali lagoon using Landsat 5 and 7 satellite images. They introduced a sort of multi-variable regression equations and showed that pH and TSS can be measured with higher accuracy in comparison with nitrate, salinity, TDS and ammonium. In another study, Navabian et al. [2020] evaluated the possibility of estimation of pollution load entering to the Anzali lagoon using remotely sensing data. They measured the contamination loads of nitrate, TDS, TSS and orthophosphate at the entrance points of the rivers to the wetland in the period of April to July 2013, and concluded that Pirbazar and Bahmbar rivers discharged the most loads of pollution to the Anzali lagoon, respectively.

However, despite the threats caused by the entry of polluted waters and sediments into this lagoon, limited studies have been conducted on satellite monitoring of WQP. In addition, the spatio-temporal trend analysis of WQP has not been investigated in this lagoon. Therefore, the aim of this study was to map the WQP derived from Landsat satellite data over the longer periods and analysis the long term changes in the WQP of the Anzali lagoon. In addition, the areas with higher pollution were identified and the quality status of the lagoon were mapped from 2014 to 2022.

2.1. Study area

Anzali wetland is a costal wetland located in the southwest of the Caspian Sea, Guilan province, Northern Iran (Fig. 1).The area of Anzali wetland is 180 to 200 km² between 36°55′ and 37°32′ N, and 48°45′ and 49°42′ E.It registered as an international wetland in the 1975 Ramsar Convention. The wetland is treated by increasing the sediment and pollution loads, land use change and invasive species such as Azolla and Lotus flower. The main source of pollution originate from agricultural runoff, drained water from paddy lands, and industrial and municipal wastewater depleted into the rivers.

The wetland is geographically divided into four distinct sections consists of eastern, central, western and Siah keshim parts where their physical properties, chemical composition, morphology and their general geographic differences (Fig. 1). Rainfall in this region normally ranges between 400 and 2000 mm/year, and long-term annual air temperature varies from − 0.8 to 37.3°C (JICA, 2005). The daily climatologic parameters consisting of precipitation and temperature used in this study was collected from two local stations presented in Table 1.

Table 1

Specifications of ground-based data used in the monitoring of water quality parameters
Station	Specifications	Longitude (m)	Latitude (m)	Elevation (m)	Time scale	Period
Anzali	Synoptic station	4149178.049	363616.627	-23.6	Daily	2014–2022
Kiashahr	Synoptic station	4138619.546	402060.229	-22	Daily	2014–2022

2.2. Google Earth Engine platform

The GEE platform contains a substantial amount of satellite data acquired by other missions and it also hosts other ancillary data such as digital elevation models, vector-based datasets, land cover and meteorological data [ Gorelick and Clinton, 2018]. Since data hosted by GEE is analysis-ready, the need for pre-processing is circumvented as these data have been found to be consistent over time and well-suited for time series analysis and estimating long-term trends [Gorelick, 2017].

A framework was developed within the GEE environment (Fig. 2) to compute the time series of WQPs from 2014 to 2022 and analysis the trend and its intensity of changes. After calling the satellite datasets in GEE and applying the date filter function to the image collections, the desired time periods (2014–2022) were separated from the whole datasets. The spatial filter function was used to limit the borders to the Anzali lagoon. Then, using the water quality models (Navabian et al., 2019) and trend functions (Man Kendall, Sen's slope) in GEE, pixel-based spatio-temporal trend for the gridded datasets was carried out to analyze the trend and its intensity. The Mann-Kendall expression (alpha 0.05 and Z statistics = 1.96, generated from the standard normal table) was used to assess the Sen's slope [ Banerjee, 2020].

A rich time series of remote sensing data including Landsat TM/ETM+/OLI (30 m spatial resolution) and GPM Precipitation dataset (10 km spatial resolution) was called in GEE to derive the WQP as follow:

2.3. GPM dataset

The Global Precipitation Mission (GPM) is a quasi-global (60°S-60°N) rainfall dataset that offers data from 2014 to almost the present at high resolution (0.1). GPM is a reliable source of rainfall data for examining spatial precipitation patterns and seasonal and interannual variability. GPM produces and distributes a wide variety of precipitation data products. Data is provided at multiple "levels" of processing, from raw satellite measurements to best-estimate global precipitation maps using combinations of all the constellation observations and other meteorological data. All data from the mission is made freely available to the public on NASA websites. Precipitation data is made available in a variety of formats, spatial and temporal resolutions, and processing levels. It offers enough spatio-temporal coverage and is especially appropriate for regions without many rainfall gauges ( Huffman et al., 2013). Due to inaccessibility or lack of ground data in the the study area, GPM data were used to fill the data gap and to produce precipitation quantitative maps for the period of 2014 to 2022.

2.4. Landsat Images

Water quantitative maps are regularly determined from optical satellite imagery such as MODIS and LANDSAT [ Plug et al., 2008; Gao et al., 2012; Tian et al., 2017]. Many researchers have also reported promising results after using LANDSAT satellite images for measuring the WQP [Yang, 2022]. LANDSAT has a high resolution (30 m), which is its main benefit, but it also has a low repetition frequency. Therefore, to generate the water quality maps over the Anzali lagoon, both the reflective and thermal bands of LANDSAT images from ETM + boarded on the LANDSAT 7 and OLI sensor boarded on LANDSAT 8 & 9 were implemented. LANDSAT images from ETM + and OLI sensors were acquired for the period of 2014–2022 from GEE. Specification of all satellite data used in this research is listed in Table 2. Due to cloudiness, 76 images could be used to monitor the quality of Anzali wetland in this period.

Table 2

Specifications of LANDSAT satellite datasets available in Google Earth Engine
parameters	Temporal Resolution	Products Information	Spatial Resolution	Time Period
GPM	Daily	Integrated Multi-satellitE Retrievals (IMERG) (Version final)	10 km	2014–2022
Landsat 7 ETM	16 days	Level 1	30m	2003–2022
Landsat 8 OLI	16 days	Level 1	30m	2013–2022
Landsat 9 OLI–2	16 days	Level 1	30m	2021–2022

2.6. Prediction of water quality parameters

To predice WQPs, two forms of calibrated multi-variable regression models developted by Navabian et al. (2019) was implemented in GEE as follow:

$$Y=a \times R_{1}^{b} \times R_{2}^{c} \times R_{3}^{d} \times R_{4}^{e} \times R_{5}^{f} \times R_{7}^{g} \times SS{T^h} \times {(rain1+rain2+rain3+T)^i}$$

$$TSS=1{0}^{a\times {R}_{1}^{b}\times {R}_{2}^{c}\times {R}_{3}^{d}\times {R}_{4}^{e}\times {R}_{5}^{f}\times {R}_{7}^{g}\times (0.001SST{)}^{h}\times (rain1+rain2+rain3+T{)}^{i}}$$

Where Y refer to the WQP, TSS refer to total suspended solids, and R₁, R₂, R₃, R₄, R₅ and R₇ refer to the reflectance bands of Landsat 7 (bands 1 to 5 and 7) and Landsat 8 & 9 (bands 2 to 7), respectively. SST is sea surface temperature derived from thermal bands in both Landsat 7 & 8. Also rain₁, rain₂ and rain₃ refer to the precipitation amount one, two and three days before sampling (mm) and T is air temperature on sampling day (C⁰). Parameters a, b, c, d, e, f, g, h and i are empirical coefficient that were defined in calibration stage. For more details regarding to the regression models and their empirical coefficients, we refer to Navabian et al. (2019).

2.7. Mann-Kendall trend and Sen’s slope test

In this study, the nonparametric statistical test methods, Mann-Kendall (MK) [Liu et al. 2012] and Sen's slope estimator [Sen 1968], was applied in GEE to compute the long-term magnitude of change on water quality parameters and their statistical significance. The MK statistical test was used to quantify the significance of trends in the time series. The MK test null hypothesis (H₀) says that there is no monotonic trend at the specified level of significance. This may be used to detect the monotonic trend in a time series. In this test, the alternative hypothesis (H_a) suggests that the data exhibit a monotonic trend over time that is described by Eq. 3:

$${\text{S}}_{(k, l)}={\sum }_{i=1}^{n-1}{\sum }_{j=i+1}^{n}sgn\left({X}_{j(k, l)}-{X}_{i(k, l)}\right)$$

Where n is the number of data points, ${X}_{i(k, l)}$ and ${X}_{j(k, l)}$are the data values in the time series i and j ($j>i$), respectively, k and l indicate point location in the data matrix and $sgn\left({X}_{j(k, l)}-{X}_{i(k, l)}\right)$ is the sign function, as illustrated in Eq. (4).

$$sgn\left({X}_{j(k, l)}-{X}_{i(k, l)}\right)=\left\{\begin{array}{c}+1 if {X}_{i(k, l)}-{X}_{j(k, l)} >0\\ 0 if {X}_{i(k, l)}-{X}_{j(k, l)}=0\\ -1 if {X}_{i(k, l)}-{X}_{j(k, l)}=0\end{array}\right.$$

In cases where the sample size $\varvec{n}>30$, the standard normal test statistic ZS_(l,k) is computed using Eq. (5):

$$Z{s}_{(k, l)}=\left\{\begin{array}{c}\frac{{S}_{(k, l)}-1}{\sqrt{Var\left({S}_{(k, l)}\right)}} if {S}_{(k, l)} >0\\ 0 if {S}_{(k, l)}=0\\ \frac{{S}_{(k, l)}+1}{\sqrt{Var\left({S}_{(k, l)}\right)}} if {S}_{(k, l)}<0\end{array}\right.$$

Positive values of ZS_(l,k) indicate increasing trends while negative ZS_(l,k) values show decreasing trends. A two-tailed test was conducted using the hypothesis test technique at alpha = 50% level of significance, with the null hypothesis being no monotonic trend in time series at the 95% level of significance (${H}_{0}: \tau =0$) and the alternative hypothesis being a substantial monotonic trend in time series in the 95% significance level (${H}_{a}: \tau =0$).

Sen’s nonparametric method was used to estimate the magnitude of trends in the time series (Eq. 6):

$${T}_{i(\text{l},\text{k}) }=\frac{{X}_{j(\text{l},\text{k}) }-{X}_{i(\text{l},\text{k}) }}{j-m}$$

In this equation, ${X}_{j}$ and ${X}_{k}$ represent data values at time j and m, respectively.

$${Q}_{i(\text{l},\text{k}) }=\left\{\begin{array}{c}{T}_{(N+1)/2} N is odd\\ \\ \frac{1}{2}\left({T}_{N/2}+{T}_{(N+2)/2}\right) N is even.\end{array}\right.$$

A positive ${Q}_{i(\text{l},\text{k}) }$ value represents an increasing trend and a negative ${Q}_{i(\text{l},\text{k}) }$ value represents a decreasing trend over time.

3.1. Spatio-Temporal changes in the extent of Anzali lagoon (2014–2022)

The three-day cumulative rainfall time series for the period 2014–2022 is shown in Fig. 3. As shown in this figure, the highest rainfall occurrs in September to October. The summer precipitations is observed in all years, although in 2019 and 2020 the amount of summer precipitation has significantly reduced. The lowest and highest air temperature were recorded in the months of January and August, respectively. However, the bounds of maximum and minimum air temperature are almost within the same range and no specific trend is observed.

Figure 4a shows the changes in the area of the lagoon in the Anzali Wetland along with the three-day cumulative rainfall on the dates of receiving the Landsat images. The reason for choosing three-day rains is the effect it had on calibrated WQP equations in the previous study [Navabian et al., 2019]. As can be seen in this figure, the three-day cumulative rainfall has a relatively good match with the total number of water pixels, which represents the extent of the lagoon. Considering that the water pixels in the lagoon were extracted using the NDWI index, it is possible that due to the masking of the cloudy area, the extent of the water surface on cloudy days may have been underestimated. The inconsistency between rainfall and the water area in May can be had different reasons for example the occurrence of more than 3- days of rainfall in some years like 2018 (a significant increase compared to the 3 days of rain). Also, compared to the rainfall amount in September 2016, the same rainfall in May 2019 could not extend the water area of the lagoon due to a decrease in the river flow leading to the wetland (beginning of rice planting on May 5). The trend of changes in the water pixels ( Fig. 4b) indicates that the water area in Anzali Wetland is decreasing dramatically, which is mainly due to the accumulation of sediments on the bed of lagoon and the reduction of its water storage capacity ( JICA, 2020).

Figure 5 shows the annual maximum extent of the Anzali lagoon between 2014 and 2022. Based on the figure, the area of Anzali lagoon has started decreasng since 2014 specially in the western and centeral parts. The dramatic redution of this waterbady is usually refreted to the entry of polluted waters and sediments into the wetland and invasion of exotic species such as water fern (A. filiculoides), water hyacinth (E. crassipes), and water lettuce (Pistia stratiotes) [Sadeghi et al., 2021].

3.3. Spatio-Temporal distribution annual pH in the Anzali Lagoon (2014–2021)

Annual distribution and trend of pH show a detailed picture of the acidity of water in the Anzali lagoon. Although it is shown in Fig. 5 that some parts of the wetland are dry, an estimate of the wetland quality was tried to be determined during the flooding of the dried parts and water quality maps were produced for the entire wetland. Figure 6 depicts that the pH varies from 6 to 8 in the lagoon. As the figure shows, from 2014 to 2015, the lagoon has lower pH, while in the period from 2015 to 2018 a higher value of pH is observed. Spatial maps of average pH in different years showed that more alkalinity occurred in 2015, 2016, 2017, and 2019 but the lagoon experienced acidity conditions in 2020. Comparing the pH of the lagoon in different seasons showed that in months 5, 6, and 8 to 12, the lagoon experiences more alkalinity conditions but acidic conditions occur often in month 3. It seems the acidic and alkaline conditions correspond to the rainy months in the watershed and the beginning of the rice cultivation season and its harvest, respectively. From 2014 to 2015, alkaline conditions were observed in the western part of the lagoon, while during 2016, alkaline conditions prevailed in most points. From 2017 onwards, more alkalinity happened on the marginal of the central part, but with the approaching to the year 2021, it tends to the eastern part of the lagoon. The significant trend of pH in the central and the edges of the eastern parts of the lagoon shows the affecting of pH changes in the rivers leading to these parts. Influenced by the pH values of wastewater entering rivers, the lagoon goes through an alkaline process, which can affect the chemical reactions and activities of the microorganisms.

Figure 7 shows the spatial distribution of trends analysis in the annual pH based on LANDSAT estimates from 2014 to 2021. Figure 7a indicated the result of Man Kendall trend analysis. This figure indicates that the central part of the lagoon goes toward alkalinity conditions and the eastern and western parts go to acidity conditions. The trend slope in the Fig. 7b shows the central part is changing with a positive slope of 0.00043 while the western and eastern parts are changing with a negative slope of 0.0003. However, the changes in most parts of the lagoon are nonsignificant, except for a small region in the western part, which significantly progresses towards acidification. The map of P-values in the Fig. 7c confirms the insignificant rise in trend (p > 0.05) of pH values from 2014 to 2021.

The Pareto diagram (Fig. 7d) showed that 43 images out of the 76 satellite images implemented for trend analysis contained average values of less than 6.5. Therefore, based on environmental classification "Order 161/2006", the lagoon has been in the unsatisfactory conditions in more than 55% of the times.

Based on Fig. 8, the time series of area-average of pH during the studied years varies between 6.15 and 7.5. The highest and lowest values occurred in 2017 and 2018, respectively. The temporal changes show a remarkable increase from Jul 2014 (~ 6.23) to Oct 2015 (~ 7.07), from Jun 2016 (~ 6.34) to Nov 2017 (~ 7.46), and from Mar 2018 (~ 6.15) to Oct 2019 (~ 7.22). Parts of the lagoon with a pH less than 6.5 are in the unfavorable range of environmental issues. Considering the significant Sen's slope in pH value in the west of the lagoon, a serious concern in this sector is observed.

3.4. Spatio-Temporal distribution annual TDS in the Anzali Lagoon (2014–2021)

Figure 9 shows the spatial distribution of the annual TDS derived from LANDSAT images in the period of 2014 to 2021. Annual mean TDS values varies from 100 to 800 mg/l in the lagoon. Spatial maps showed that the highest TDS often occurres in May, June, and September while its lowest is observed in July. It seems fertilization in the rice cultivation affects the amount of TDS more than salinity in the months of May and June. Also, leaching salts of soil during the heavy seasonal rains can be the reason of high salinity in September. The spatial trend of the maximum TDS in the years of 2014–2021 demonstrates that in the early years, the western part has experienced high TDS, and after that high TDS moved toward the margin of the center part. Since 2019, the salinity has intensified in the center and eastern parts. The spatial pattern of the Man Kendall trend exhibits that TDS is rising in the central parts and falling in the western and eastern of the Anzali lagoon (Fig. 10a). The Sen’s slope map in the figure (10b) reveals pisitive and negetive rising slopes in TDS which varies from 0.00043 (upper limit) to -0.0012 (lower limit) with the significant tendencies at 95% confidence level. The statistically significant increasing tendency is detectable across the central part (+ 0.00043). The map of p-value in the Fig. 10c exhibits that changes in TDS is significant almost in the whole lagoon. The majority areas in the western and eastern parts has a significant negative trend but in the marginal of these parts near the river effluent, the trend is observed nonsignificant positive. According to the Pareto diagram (Fig. 10d), the frequency of occurrence of TDS with values above 750 mg/liter was 17 occurrences from 76 images, which includes 22% of the investigated dates.

Investigating the pattern of temperature and rainfall changes with TDS shows that the rain changes is more coordinated with TDS changes than temperature (Fig. 11). The matching of the precipitation pattern and TDS shows that increasing rain leads to entering more salts into the lagoon, but this load is increased at the time of spreading fertilizers in the paddy lands in May and April. The second increase occurs in the rainy months (September) and at the time of rice harvest. Although, an increase in TDS with the occurrence of rainfall events is observed in the agricultural season, temperature is more effective than rainfall on TDS in the other months. The maximum value of TDS was 1345 mg/liter on March 2018, and the best condition of the lagoon was on June 2015, when the maximum TDS reached 229 mg/liter. The standard "Order 161/2006", does not approve values above 750 mg/liter of TDS for surface water bodies such as lagoon.

3.5. Spatio-Temporal distribution annual TSS in the Anzali lagoon (2014–2021)

Figure 12 depicts the spatial distribution of TSS, which varies from 1 to 200 mg/l in the lagoon. From this figure, it is clear that the TSS parameter fluctuated during the years 2014 to 2012. Spatial distribution maps of 76 satellite images showed that the lagoon has encountered high levels of TSS in March, May, and November in all years, and in Januray and October in some years. The distribution pattern revealed that in months March and May, the points of the entrance of the rivers in the western, central, and eastern parts, had higher values of TSS. But in months October and November, high amounts of TSS were observed sporadically on the lagoon's surface. It seems the difference in the source of suspended solids entering is the reason for the difference in a focused area of TSS between the beginning (agricultural effluent) and the end (soil erosion caused by flood rains) months of the year. The minimum amounts of TSS occurred in months August and September, which can be due to soil erosion reduction due to the presence of plant residues from rice harvesting in the paddy fields of the lagoon catchment.

Man Kendal test in the Fig. 13a shows that the middle of the central part, the lower edge of the western, and the edge of the eastern parts of the lagoon have an upward trend of TSS. The slope of increasing TSS values in these areas was 0.00043 (Sen’s Slope Estimator), and this trend was significant in some places at a 95% confidence level (Fig. 13c).

The time series of area-average of TSS over Anzali lagoon (Fig. 14) showed high fluctuation in TSS due to the precipitation in the upstream of watershed and the agricultural activities in the area (plowing of soil and rice planting). For example, the maximum TSS was estimated as 87 mg/liter in May 2015 which coincides with the soil preparation for rice cultivation. The soil of the Anzali watershed includes high levels of clay and silt, so precipitation can be the main reason for increasing soil erosion and TSS. The time series of TSS trend and precipitation values in 76 different dates (Fig. 13) indicated that three-days cumulative rainfall have a more effective role in tracing the TSS values than the previous day's rainfall values.

The Pareto diagram (Fig. 13d) showed that in three out of 76 images, the TSS values exceeded 40 mg/liter, which puts the lagoon in the worst condition of the surface water according to the China National Standards for Surface Water Quality (CNSSWQ). This lead to disrupt the production process of dissolved oxygen in the lagoon. Based on class II of CNSSWQ (15 mg/liter), which shows good ecological status for surface water, the TSS values have exceeded the permissible limit in 5 occurrences.

3. 6. Spatio-Temporal distribution annual nitrate in the Anzali lagoon (2014–2021)

Figure 15 depicts the spatial distribution of nitrate that varies from 1 to 250 mg/l in the lagoon. Spatial distribution maps in different years showed that the nitrate levels usually increase in months of June to August (about two months after transplanting rice to the paddy lands). Due to the large number of paddy fields in the Anzali lagoon watershed, higher values of nitrates during the cultivation season was predicted. The distribution pattern of nitrate in the mentioned three months was different, as in the 6th month, the marginal, and in months 7 and 8, the marginal and middle parts of the lagoon had high nitrate levels.

Figure 14 shows the spatial distribution of nitrate values based on MODIS (MOD16A2) product estimates from 2014 to 2021. The MK test exhibits that the nitrate trend is rising in the eastern and western parts and falling in the central part (Fig. 14(A)). In Fig. 14(B), Sen’s slope calculation varies from 0.00043 (upper limit) to -0.0012 (lower limit) for upward and downward trends for the significant tendencies at a 95% confidence level, respectively. Figure 14 (C) exhibits that nitrate rising in the west and east is significant as (p-value = 0) is less than alpha = 0.05. The MOD16 temporal obtained by the linear regression model shows the lowest and highest nitrate spatial mean values are 21.5 and 89.2 mg/l in May 2019 and Jul 2014, respectively (Fig. 14 (D)).

High nitrate levels were often observed at the marginal of the lagoon and close to river effluents. The highest level of nitrate occurred in 2014 and then in 2018, and this fluctuation shows that the nitrate values are affected by the rainfall pattern. Figures (14 (E)) shows that it is possible to predict the time of occurrence of the high nitrate values and the critical conditions of the lagoon based on the rainfall pattern. The maximum mean spatial nitrate value was obtained 89 mg/liter, so 100% of the images had a spatial average above the standard limit (3 mg/liter based on Class II of Order 161/2006). Also, based on the desirable and maximum permissible limit of the European Community (25 and 50 mg/liter), about 92 and 40% of the images were over the standard limit. Therefore, the lagoon has a high potential for eutrophication, and it is essential to reduce nitrogen loadings.

This study has well established the efficiency of remote sensing for trending the Anzali lagoon condition in the parameters of TSS, TDS, pH, nitrate, and the lagoons area. The information obtained from Landsat images from 2014 to 2021 showed that the lagoon has environmental problems. The results of this study are consistent with objective observations. The growth of algae, the death of aquatic life, the reduction of the lagoons area, and the difficulties in navigating through the lagoon have increased in recent years. The comparison of the investigated parameter values with environmental standards also showed that the whole of the lagoon has a critical condition in TSS, and in terms of nitrates and TDS, at least 50 and 22% of the lagoon are in critical condition, respectively. The trend of changes in quality parameters in some parts obtained a significant increase, which means an increase in the crisis in the lagoon. The pattern of spatial changes showed that the problems in the first years of the investigation were more on the western side, but in recent years, it has spread to the center and east of the lagoon. The evaluation of the coordination of precipitation pattern, temperature, and agricultural activity in the watershed with the studied parameters showed that the quality condition of the lagoon is affected by these factors an example, the nitrate concentration strongly affects by the rainfall pattern. Soil erosion caused by floods and plowing of paddy fields of the watershed is problematic due to the soil texture of the region. Since rainfall and agriculture affect the condition of the lagoon, it suggests presenting the measures to reduce the pollution of the agricultural sector (nitrogen fertilizer control based on soil needs and irrigation and drainage management) and reduce erosion and retain sediments. Lagoon monitoring by the proposed method in this study can be an efficient tool to investigate lagoon changes, find the critical points, and evaluate the effects of management decisions.

Acknowledgments

This article was supported by Iran Nation Science Foundation with 89002948 code and the researchers are grateful for the support provided by the fund. It is hoped that the findings of this study will be used by water authorities.

Schmidt, K.; Skidmore, A. Spectral discrimination of vegetation types in a coastal wetland. Remote Sens. Environ. 2003, 85, 92–108.
Biolearn .2021 .Threats to lagoons and lagoon conservation. Eco- conscious minds to stop pollution in the valuable lagoons of black sea basin. Biolearn-BSB142. Republic of Turkey district government of Enez.
Yang, H., Kong, J., Hu, H., Du, Y., Gao, M. and F. Chen. 2022. A Review of Remote Sensing for Water Quality Retrieval: Progress and Challenges. Remote Sens. 14, 1770: 1-21.
El Din, E.S.; Zhang, Y.; Suliman, A. Mapping concentrations of surface water quality parameters using a novel remote sensing and artificial intelligence framework. Int. J. Remote Sens. 2017, 38, 1023–1042.
Palmer, S.C.; Hunter, P.D.; Lankester, T.; Hubbard, S.; Spyrakos, E.; Tyler, A.N.; Presing, M.; Horvath, H.; Lamb, A.; Balzter, H.; et al. Validation of Envisat MERIS algorithms for chlorophyll retrieval in a large, turbid and optical-ly-complex shallow lake. Remote Sens. Environ. 2015, 157, 158–169
Li, J.; Yu, Q.; Tian, Y.Q.; Becker, B.L.; Siqueira, P.; Torbick, N. Spatio-temporal variations of CDOM in shallow inland waters from a semi-analytical inversion of Landsat-8. Remote Sens. Environ. 2018, 218, 189–200.
Sun, D.; Qiu, Z.; Li, Y.; Shi, K.; Gong, S. Detection of Total Phosphorus Concentrations of Turbid InlandWaters Using a Remote Sensing Method. Water Air Soil Pollut. 2014, 225, 1–17.
Osvaldo J. R. Pereira 1,*, Eder R. Merino 1 , Célia R. Montes 2, Laurent Barbiero 3, Ary T. Rezende-Filho 4 , Yves Lucas 5 and Adolpho J. Melfi . 2020. Estimating Water pH Using Cloud-Based Landsat Images for a New Classification of the Nhecolândia Lakes (Brazilian Pantanal). Remote Sens. 2020, 12, 1090: 1-21.
Abdulmunem Dherar Abdullah Aljoborey1, Hind Suhail Abdulhay. 2019. Estimating total dissolved solids and total suspended solids in Mosul dam lake in situ and using remote sensing technique. Periodicals of Engineering and Natural Sciences ISSN 2303-4521 Vol. 7, No. 4, pp.1755-1767.
Yuanyuan Zhang, Shuanggen Jin, Ning Wang, Jiarui Zhao, Hongwei Guo and Petri Pellikka.2022. Total Phosphorus and Nitrogen Dynamics and Influencing Factors in Dongting Lake Using Landsat Data. Remote Sens.2022, 14(22), 5648: 1-28.
Faranak Amirsalari, Jonathan Li, Xian Guan, & William G. Booty. 2013. Investigation of correlation between remotely sensed impervious surfaces and chloride concentrations. International Journal of Remote Sensing Volume 34(5): 1507-1525.
MARYAM NAVABIAN1*, MAJID VAZIFEHDOST2, MEHDI ESMAEILI VARAKI2. 2019. Feasibility Study of Anzali Lagoon Quality Monitoring Using Remote Sensing. Iran J Soil Water Research, 50(7): 1813-1824.
Maryam Navabian, Majid Vazifehdost, Mehdi Esmaeili Varaki. 2020. Estimation of pollution load to Anzali Lagoon using remote sensing technique. Caspian Journal of Environmental Science, 18(3): 251-264
Gorelick, N. and N. Clinton. Multitemporal Supervised Classification Using Google Earth Engine. in AGU Fall Meeting 2018. 2018. AGU.
Banerjee, A., et al., An analysis of long-term rainfall trends and variability in the uttarakhand himalaya using google earth engine. Remote Sensing, 2020. 12(4): p. 709.
Gorelick, N., et al., Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote sensing of Environment, 2017. 202: p. 18-27.
Tian, H., et al., Dynamic monitoring of the largest freshwater lake in China using a new water index derived from high spatiotemporal resolution Sentinel-1A data. Remote Sensing, 2017. 9(6): p. 521.
Plug, L.J., C. Walls, and B. Scott, Tundra lake changes from 1978 to 2001 on the Tuktoyaktuk Peninsula, western Canadian Arctic. Geophysical Research Letters, 2008. 35(3).
Gao, H., C. Birkett, and D.P. Lettenmaier, Global monitoring of large reservoir storage from satellite remote sensing. Water Resources Research, 2012. 48(9).
Sen, P. K., 1968: Estimates of the regression coefficient based on Kendall’s tau. J. Amer. Stat. Assoc., 63, 1379–1389
Liu, D., S. Guo, X. Chen, and Q. Shao, 2012: Analysis of trends of annual and seasonal precipitation from 1956 to 2000 in Guangdong Province, China. Hydrol. Sci. J., 57, 358–369
Huffman, G.J.; Bolvin, D.T.; Braithwaite, D.; Hsu, K.; Joyce, R.; Xie, P.; Yoo, S.H. Algorithm Theoretical Basis Document, Version 4.1: NASA Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG); NASA/GSFC: Greenbelt, MD, USA, 2013.
Sadeghi Pasvisheh R, Eurie Forio MA, Ho LT, Goethals PLM. Evidence-Based Management of the Anzali Lagoon System (Northern Iran) Based on Innovative Monitoring and Modeling Methods. Sustainability. 2021; 13(10):5503. https://doi.org/10.3390/su13105503
JICA. The Study on Integrated Management of the Anzali Wetland in the Islamic Republic of Iran-Final Report. 2005, p. 182. Available online: https://openjicareport.jica.go.jp/pdf/11784097.pdf (accessed on 11 May 2021).

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

A multi-sensor framework in google earth engine for spatio-temporal trend analysis of water quality parameters in Anzali lagoon

Status:

Version 1

Abstract

Figures

1. Introduction

2. Material and Methods

2.1. Study area

2.2. Google Earth Engine platform

2.3. GPM dataset

2.4. Landsat Images

2.6. Prediction of water quality parameters

2.7. Mann-Kendall trend and Sen’s slope test

3. Results and Discussion

3.1. Spatio-Temporal changes in the extent of Anzali lagoon (2014–2022)

3.3. Spatio-Temporal distribution annual pH in the Anzali Lagoon (2014–2021)

3.4. Spatio-Temporal distribution annual TDS in the Anzali Lagoon (2014–2021)

3.5. Spatio-Temporal distribution annual TSS in the Anzali lagoon (2014–2021)

3. 6. Spatio-Temporal distribution annual nitrate in the Anzali lagoon (2014–2021)

4. Conclusion

Declarations

Acknowledgments

References

Additional Declarations

Status:

Version 1