*2.2. In Situ Data*

Four sampling cruises were carried out during 2021 (March, July, August, and November), where seawater was collected at 10 points homogeneously distributed in the Mar

Menor lagoon (Figure 5). The dates of the in situ campaigns during the study period corresponded to 23–24 March, 27 July, 26 August, and 16–17 November 2021. Matchups between in situ and satellite samples were generated when both data acquisitions occurred within 30 h of each other, and the satellite value was calculated as the mean of the 3 × 3 10 m pixel region around the sampling station.

**Figure 5.** Control points for data collection during the four in situ campaigns carried out in March, July, August, and November 2021.

In order to determine turbidity, samples were measured just after collection onboard using a portable turbidimeter (2100P, Hach). Prior to calibration, we ensured that the equipment did not suffer anomalies and that the necessary reagents were available. Calibration was performed quarterly, unless the equipment was malfunctioning. The turbidimeter was calibrated at 4 points: 0 NTU, 10 NTU, 200 NTU, and 800 NTU. The standards used in the calibration were certified commercial standards at room temperature in order to avoid misting interferences in the turbidity measurement. The turbidimeter was calibrated according to the equipment manual.

In the case of chl-a, water was collected from a 0.5 m depth in 1 L dark bottles to avoid enhanced photosynthetic activity and kept in a portable fridge until arrival at the laboratory. Three replicates per station were collected. Chl-a concentrations were measured in 700–1000 mL water samples for each replicate, which were filtered through Whatman GF/F 0.2 μm polycarbonate filters. The filters were immediately frozen at −20 ◦C until pigment extraction in 90% acetone at 4 ◦C overnight in the dark. Chl-a concentrations were determined with a 10-AU Turner Designs fluorometer calibrated with pure chl-a [50].

#### **3. Results and Discussion**

#### *3.1. Multisensor Approach and Preprocessing*

Until now, standard ocean color sensors at the moderate spatial resolution of 300–1000 m were generally used to map water quality, phytoplankton blooms, and eutrophication events in Mar Menor [25,26]. However, features in the lagoon show typical scales of tens of meters that cannot be detected with existing ocean color sensors (Figure 4). In recent years, several studies already demonstrated that, in order to comprehensively evaluate the ecological conditions of heterogeneous coastal areas and land-water inputs using remote sensing technologies, improved spatial resolution should be used with Landsat-8 and/or Sentinel-2 [28,30,31,51]. Undoubtedly, the Sentinel-2 twin mission, although originally not designed for coastal ocean monitoring, is a key tool for the detailed mapping of highly dynamic environments, such as coastal or inland water areas [52].

During the study period, in particular during summer, acute sunglint contamination also influenced the quality of the imagery over Mar Menor due to the specular reflection of sunlight off the water (see Table 1 for details). Figure 6 shows two images on 9 June 2021 and 4 July 2021 at the top-of-atmosphere (TOA) level, after ACOLITE processing at the bottom-of-atmosphere (BOA) level, and the Rrs of the blue band (492 nm). Accurate performance was accomplished by ACOLITE over low to moderate sunglint conditions, such as on 9 June 2021 (Figure 6a–c), but failed to retrieve Rrs during severe sunglint contamination on 4 July 2021 (Figure 6d–f), thus masking the data. The sunglint is clearly observed at TOA-Level-1C (Figure 6a,d), whereas the residuals are visible at BOA-Level2 (Figure 6e) with intense effects on the eastern area of the Sentinel-2 tile, exactly where Mar Menor is located. This effect is significant during spring and summer, given that minimum information could be retrieved with severe sunglint, restricting the amount of available data when the majority of the blooms occurred. Comparable results were generated in other coastal regions, such as in North Atlantic [53] or Caribbean waters [54]. Irregular residual issues still require further advancements in the sunglint and atmospheric correction approaches to facilitate the extensive combination of Landsat-8 and Sentinel-2 products during summer. In addition, the typical stripping patterns are clearly observed at both TOA and BOA levels.

On 3 August 2021, Landsat-8 and Sentinel-2 acquired a scene at 10:45 am and 11:00 am GMT, respectively. Figure 7 shows the spectral signal of both satellites with only a 15 min time difference over three control points distributed across different areas of the lagoon. This exhibits the consistent performance of ACOLITE for both satellite missions, retrieving the spectrum with similar Rrs values for each point and sensor (P1, turbidity of 18.11 and 17.28 FNU; P2, turbidity of 5.5 and 6.1 FNU; and P3, turbidity of 4.08 and 3.93 FNU for Sentinel-2 and Landsat-8, respectively), thus further corroborating the remarkable value of the combined products. Comparison of Sentinel-2 and Landsat-8 Rrs over the visible and NIR bands yielded a bias of −0.00035 sr−1, MAE of 0.00072 sr−1, and MedAE of 0.00049 sr−1. Consistent sunglint and atmospheric correction models are needed to empower the application community to explore these products in order to thoroughly address the ecological conditions of Mar Menor using remote sensing technologies. Recent research already demonstrated the potential of ACOLITE to provide robust information for aquatic and marine applications [30,43,44,54]. A study applied Sentinel-2 data to monitor water quality in Mar Menor using the Sen2Cor atmospheric correction processor (designated for land application with restricted performance in water application) and to generate products at 60 m spatial resolution [35]. However, Pahlevan et al. (2019, 2021) suggested that enhanced information for inland and coastal water quality mapping is required as a critical and urgent task to evaluate spatial and spectral differences under several atmospheric and aquatic conditions [28,55]. This data record is crucial to ensuring a detailed monitoring of the Mar Menor coastal lagoon with both satellite platforms working in tandem.

**Figure 6.** RGB (Red–Green–Blue) composite image on 9 June 2021 from the Sentinel-2 satellite (10 m spatial resolution) at (**a**) top-of-atmosphere (TOA) level, (**b**) bottom-of-atmosphere (BOA) level after ACOLITE, and (**c**) remote sensing reflectance (Rrs, sr−1) of the blue band (492 nm); (**d**–**f**) the same on 4 July 2021. Severe sunglint contamination can be clearly observed in the eastern section of the Sentinel-2 tile affecting Mar Menor.

**Figure 7.** (**a**) RGB (Red–Green–Blue) composite image on 3 August 2021 of the Sentinel-2 satellite at bottom-of-atmosphere (BOA) level, (**b**) spectral signal of the Sentinel-2 and Landsat-8 satellites over different control points (P1, turbidity of 18.11 and 17.28 FNU; P2, turbidity of 5.5 and 6.1 FNU; and P3, turbidity of 4.08 and 3.93 FNU for Sentinel-2 and Landsat-8, respectively). Yellow circles in (**a**) indicate the location of the control pixels.

#### *3.2. Validation of the Water Quality Algorithms*

Figure 8 shows the validation matchups for the water quality parameters obtained with Sentinel-2 and Landsat-8 during the four in situ campaigns carried out in 2021. We applied the standard OC3 algorithm to calculate chl-a concentration and a regularly used semianalytical algorithm for the determination of turbidity [44,45]. The performance of both algorithms is illustrated in Figure 8a,b, respectively. The chl-a matchups cover the range of 0.5–5 mg/m<sup>3</sup> with a bias of 0.37 mg/m3, MAE of 0.43 mg/m3, and MedAE of 0.41 mg/m3 (R2 = 0.903, *n* = 37), whereas the turbidity ranges from 0.5–6 FNU with a bias of 2.09 FNU, MAE of 2.09 FNU, and MedAE of 2.04 FNU (R<sup>2</sup> = 0.54, *n* = 35). The validation assessment indicated robust statistical analysis, with accurate chl-a retrieval and minimum bias. Predictions of chl-a from both Sentinel-2 and Landsat-8 yielded precise results after the ACOLITE atmospheric and sunglint correction. The performance of the ACOLITE and the turbidity model is accurate, but a general satellite overestimation was encountered with biased outcomes, as seen in Figure 8b. Pahlevan et al. (2022) also found overestimation of turbidity retrievals by means of the ACOLITE processor [56]. The turbidity model has already been validated in different regions worldwide with accurate performance [46–48,57] and has previously been used in Mar Menor during an extreme weather event [13]. These methodologies are consistent and valid approaches for the assessment of suspended material or turbidity, which contribute towards achieving more precise performance worldwide [45,58].

**Figure 8.** Validation of water quality parameters obtained with Sentinel-2 (S2) and Landsat-8 (L8) during the four in situ campaigns carried out in March, July, August, and November 2021 for (**a**) chlorophyll-a concentration (chl-a, mg/m3) and (**b**) turbidity (FNU).

#### *3.3. Water Quality Monitoring*

Figure 9 shows the RGB composite scenes acquired on 9 and 24 June and 2, 14, 18, and 29 July 2021, whereas Figures 10 and 11 display the image-derived maps for turbidity and chl-a, respectively. The imagery corresponded to the months prior to the ecological crisis in mid-August 2021. Generally, the turbidity levels were low in the lagoon (<5 FNU), except in the western section on 24 June and 14 July 2021, indicating higher levels (~25 FNU). A turbid plume appeared near land where the Albujon watercourse flows into Mar Menor. The most common chl-a condition during this period was <1.5 mg/m3, while higher chl-a concentration (~2.5 mg/m3) was observed close to the turbid plume. Figures S1 and S2 indicate the available (cloud and sunglint-free) Sentinel-2 scenes in August 2021 for further evaluation. On 3 August, a turbid plume was observed in the western section close to the input of the Albujon watercourse with peaked levels ~20 FNU, whereas minimum turbidity was encountered in the rest of the lagoon. The chl-a concentration for this date seemed to increase in the western section, indicating maximum values within the lagoon. The turbidity maps depicted high and constant turbidity values ~20 FNU in August 2021, except a slight decrease on 18 August 2021 on the western side (Figure S2). Moreover, chl-a maps displayed higher concentrations compared with July and the beginning of the bloom during this month, in particular on 13 August 2021 with chl-a ranging from 4 to 9 mg/m3, a strong indicator of algal blooms in Mar Menor. Clear-water lagoon phases are characterized by chl-a concentrations ranging from 1 to 3 mg/m3 [59], as occurred during June and July

2021. However, these typical low chl-a values tipped rapidly towards more eutrophic conditions, with chl-a concentration higher than 3 mg/m3, such as last year's [18,19,21,60]. We reported that using the multisensor approach during the eutrophication episode in 2021, the beginning of the bloom (chl-a concentration higher than 3 mg/m3) was detected mainly in the western and southern section. This is critical information for early detection of the eutrophication processes, given that the massive mortality of fish and crustaceans occurred during the last weeks of August 2021 [21], as well as an opportunity to enhance emergency management response in early deterioration stages.

**Figure 9.** Sentinel-2 and Landsat-8 RGB (Red–Green–Blue) composite image acquired on (**a**) 9 June 2021, (**b**) 24 June 2021, (**c**) 2 July 2021, (**d**) 14 July 2021, (**e**) 18 July 2021, and (**f**) 29 July 2021.

After this event, the ecosystem equilibrium recovered slightly during September 2021 (Figure S3) as can be observed in the decrease in turbidity retrievals on the western side (Figure S4). However, the Albujon continued to discharge to the western section of the lagoon, indicating increased surface runoff and rising turbidity levels, and a plume was always close to this area. In addition, chl-a concentration gradually reached normal values <3 mg/m3 in some areas in September 2021, although on 7 and 12 September 2021 high chl-a concentrations of ~4–5 mg/m<sup>3</sup> persisted not only in the center and south but also in the northern and eastern sections of the lagoon (Figure S5). A minor "Cold Drop" occurred on 20–21 September 2021, but the cloud and haze coverage remained very high during the consecutive days, as can be observed on 22 and 27 September 2021.

**Figure 10.** Turbidity (FNU) from Sentinel-2 and Landsat-8 acquired on (**a**) 9 June 2021, (**b**) 24 June 2021, (**c**) 2 July 2021, (**d**) 14 July 2021, (**e**) 18 July 2021, and (**f**) 29 July 2021.

**Figure 11.** Chlorophyll-a concentration (Chl-a, mg/m3) from Sentinel-2 and Landsat-8 acquired on (**a**) 9 June 2021, (**b**) 24 June 2021, (**c**) 2 July 2021, (**d**) 14 July 2021, (**e**) 18 July 2021, and (**f**) 29 July 2021.

Time-consuming and costly on-site measurements are regularly carried out to determine the water quality status in the lagoon; nevertheless, these observations are not able to address the heterogeneity and complexity of the spatial distribution within Mar Menor. In fact, in situ data might lack samples from the peak of the bloom or high turbidity levels due to the sparsely distributed single sampling sites. Field campaigns may not have adequately retrieved maximum concentration in chl-a if in situ measurements were not correctly spotted in Mar Menor [35]. The combined satellite data series characterized the dynamic nearshore patterns and fine-scale bio-optical gradients across this complex coastal interface. Satellite maps offered a synoptic perspective of the entire lagoon, detecting higher and lower turbidity and chl-a concentration over the study area. Interestingly, while maximum turbidity levels across the study site were typically located in the western section associated with the drainage of the Albujon, highlighting the impact of hydrological inputs and discharge from this canal, minimum levels were observed on the eastern, northern, and southern sides and along the barrier beach "La Manga". Our results also present the highest chl-a concentrations along the western coastline, detecting a change due to a proliferation of phytoplankton in early August 2021. The IEO-CSIC suggested that this eutrophication event was due to an excess of nutrient availability flowing from the Albujon [21]. Therefore, the abrupt deterioration of the water quality in Mar Menor reached a stage of severe eutrophication that resulted in an ecological collapse in mid-August, showing a gradual recovery during September 2021 before the minor "Cold Drop" event on 20 September 2021. Among all the wadis transporting materials, water, and nutrients from agricultural run-off, the Albujon is the principal collector of the Campo de Cartagena drainage basin, subjecting the lagoon to nutrient and sediment runoff from the agricultural landscape [11,12,15,16]. These results sustain that in order to remedy the ecological collapse of the lagoon, it is crucial to design and implement environmental strategies and policies [22], in particular those that focus on limiting the suspended material discharged from the Albujon to regulate the massive proliferation of phytoplankton and the eutrophication pressure favored by agricultural dumping [25,61].

#### *3.4. An Early Warning Tool with High Spatial Resolution*

The complex distribution and variability of the lagoon can be observed in detail in all the satellite-derived products presented in this study, in particular for the turbid plume located in the western part. These turbid features are usually small in dimension; therefore, detecting them by means of traditional ocean color sensors at lower spatial resolution can be challenging. We recommend using Landsat-8 and Sentinel-2 missions in tandem to improve the monitoring and control of Mar Menor. The multisensor methodology might enhance previous studies that attempted to map water quality using coarser spatial resolution imagery at 300–1000 m [24–26]. Additional evaluation of previous months in March 2021 also highlighted the importance of our methodology for studying the impact of weather events on the coastal lagoon. Figures S6 and S7 show the RGB composite images and turbidity levels on 11 and 12 March 2021 and on 21 and 28 March 2021, before and after a severe winter storm, respectively. The maps corresponding to 11 and 12 March 2021 presented minimum turbidity levels (<4 FNU) in front of the Albujon. This cycle was occasionally disrupted by the intense winter storm resulting in increased inputs of terrestrial discharges into the entire lagoon. The high resuspension of materials can be observed in both the RGB and turbidity maps after the storm, in particular on 21 March 2021 along the western coastal region with turbidity >50 FNU. A zoom on 21 March 2021 corresponding to the southeastern shore of Mar Menor showed the high variability of the turbidity patterns (Figure S8). Turbidity generally decreases seaward in the lagoon and extreme events, such as storms, can increase turbidity 5-to-10-fold, altering the water quality distribution in the system. In particular, "Cold Drop" events can dramatically alter the ecological status of the lagoon with turbidity levels increasing by more than a factor of five [13]. Previous studies have already indicated that finer spatial resolution is needed to comprehensively determine these complex spatial and temporal features [62].

With three-to-four-day revisits allowed by combined Landsat-8 and Sentinel-2 datasets, the managers, end users, and coastal science community will take advantage of these synoptic, improved, consistent, and high-quality products. This information may be critical for operational purposes in the context of the EU WFD [32], from which early warning systems can be implemented. Although work remains to be performed towards improving and developing advanced sunglint, atmospheric, and bio-optical algorithms for both Sentinel-2 and Landsat-8, it is the ideal moment to examine, exploit, and maximize these merged datasets in Mar Menor. Particularly during ecological crises, such as the one explored in this study, this information is crucial to assess appropriate measures to be taken in coastal and inland water ecosystems. However, research must continue to enable retrievals in extremely contaminated sunglint scenes during eutrophic/turbid conditions in summer periods, as demonstrated in this study (Figure 6), as well as the analysis of other biogeochemical variables, such as Colored Dissolved Organic Matter (CDOM). In addition, as shown in recent research by Wójcik-Długoborska [63], turbidity measurements in the field may differ from those taken in the laboratory and thus provide different correlations between reflectance and the true value of turbidity. Therefore, we intend to focus additional research on this aspect during the coming field campaigns. Future studies will also be carried out to evaluate the entire Sentinel-2 and Landsat-8 series to assess the seasonality of these events and to identify possible common factors, which can be monitored or used as warning systems in the future. The improved resolution afforded by the combined time-series products offers additional insights into processes over weekly or subweekly timescales; nevertheless, these results emphasized the need for enhanced temporal coverage space-based datasets in dynamic coastal environments. With these three satellites now operating, Landsat-9 already in orbit, planned missions launching shortly (e.g., Sentinel-2C/D), and continuously improving atmospheric and sunglint correction techniques, the accessible record of high-to-moderate spatial resolution imagery will provide even more robust water quality monitoring in complex inland and coastal environments.
