2.1.

Study Area

The study area is within the Mississippi Bight located in the nGoM [Fig. 1(a)]. The Mississippi Bight coastal plain is broad and of low relief which allows large estuarine systems to intrude inland.^15,16 This complex coastal ecosystem is defined by a series of barrier islands that separate the estuarine system from the Gulf of Mexico. The barrier islands are separated by inlets and protect the shallow lagoons of Mobile Bay, MSS, Chandeleur Sound, and Breton Sound.³² Our study focused on the MSS and Mobile Bay which have water depths less than 6 m and a mainly diurnal tide of less than 0.6 m. The MSS is a shallow (average 3-m deep) elongated estuarine basin that connects to the GoM through a series of passes between five barrier islands, i.e., Cat, Ship, Horn, Petit Bois, and Dauphin islands [Fig. 4].^12,24,33 Although the geographical boundaries of the MSS are often a source of debate, the eastern and western boundaries are nominally Mobile Bay and Cat Island.³² Most of the fresh water fluxes into the MSS are due to the Pascagoula and Pearl rivers, however, other smaller rivers (i.e., Biloxi, Tchoutacabouffa, Jourdan, and Wolf), small bayous, and even Mobile Bay and the Mississippi River contribute to the fresh water inputs into the Sound.^12,17 The Mobile River may contribute fresh water to the eastern side of the Sound, while the western MSS may receive fresh water from the Mississippi River especially when the Bonnet Carré spillway is open^12,24,34 and also from other rivers through Lake Borgne and Pontchartrain. Based on salinity, Eleuterius¹² defined the Sound as vertically well-mixed from July to December, with vertical homogeneity reaching a peak in October which is the period of interest. Adjacent to the MSS is Mobile Bay, a wide, shallow, and highly stratified estuary. The main freshwater sources for Mobile Bay are the Mobile and Tensaw Rivers.¹⁴ A representative monthly average salinity in Mobile Bay was recorded as 13 psu during August–October 1968 and 4 psu during February–April 1969. The monthly average representative salinity in the MSS is generally higher with an average salinity of 16 psu in summer–fall months of 1980 and 13 psu in winter–spring months of 1984 with highest average salinities in the central part of the MSS and lower average salinities toward the western and eastern parts of the MSS.¹⁷

2.2.

Ocean Circulation Model

In this study, the results of a regional application of NCOM for the Gulf of Mexico at 1-km resolution were used. A daily assimilation cycle was used at the beginning of each daily simulation followed by 72-h forecasts. Three-hour model outputs of temperature, salinity, velocities, and surface elevation were produced by NCOM, and the solution only from the first 24 h of each 72-h forecast period was used in the analysis of this study. The model results for the entire GoM was the subset for the Mississippi Bight.

NCOM is a Boussinesq model that solves the hydrostatic primitive equations.^35–37 The spatial resolution of the model is 1 km in the horizontal, and the water column is resolved by 50 levels in the vertical. Locations with 250 m and shallower depths were resolved by 35 sigma levels, and additional 15 fixed $z$-levels were used below 250-m depth at deeper locations. The model incorporates a realistic bathymetry derived from the Naval Research Laboratory 2-min database. Atmospheric forcing is provided hourly from a 17-km resolution operational application of Coupled Ocean Atmosphere Mesoscale Prediction System (COAMPS^®).^38–40 Boundary conditions, i.e., temperature, salinity, velocities, and elevation, are provided from the global operational HYbrid Coordinate Ocean Model (HYCOM). In addition, tidal boundary conditions are gathered from the global Oregon State University Tidal Inversion Software solution.^41,42 In this specific version of NCOM, monthly climatological river forcing was used for the major rivers in the area, i.e., Mississippi, Pearl, Pascagoula, and Mobile Rivers. Model predictions were a 40-h moving average filtered to eliminate the tidal signal when comparing with the measurement data.

2.3.

Ocean Color Satellite Imagery

Satellite derived chlorophyll-a was obtained from the MODIS-Aqua satellite. Quality controlled MODIS-Aqua Level-3 standard mapped image (SMI) chlorophyll-a monthly means and climatology were downloaded from the NASA-Ocean Biology Processing Group website) at 4-km spatial resolution.⁴³ Level-3 chlorophyll-a is derived using a combination of the OC3 algorithm from O’Reilly et al.⁴⁴ and the color index from Hu et al.⁴⁵ NASA’s l2-gen processing take care of calibration and atmospheric correction and level-2 flags are applied which eliminates suspicious or erroneous pixels. For more information about NASA’s ocean color data processing, calibration, and validation, refer to O’Reilly et al.⁴⁶ and Werdell and Bailey.⁴⁷

A 13-year monthly climatology for the month of October was developed by averaging all chlorophyll-a data for October months of each year from 2003 to 2015. The MODIS-Aqua October monthly mean chlorophyll-a anomaly fields over the Mississippi Bight for each year between 2003 and 2015 were calculated by taking the difference between the October chlorophyll-a monthly mean field of each year and the 13-year October monthly mean climatology as

2.4.

In Situ Measurements

The Mississippi Department of Marine Resources (MDMR) provides data from real-time hydrological monitoring stations operated in partnership with the U.S. Geological Survey (USGS) in the MSS.⁴⁸ The stations are maintained to USGS specifications⁴⁹ with calibration uncertainties not exceeding $\pm 0.2°\mathrm{C}$, for temperature, and the higher of $\pm 5\mathrm{mS}/\mathrm{cm}$ (roughly 0.01 psu) or $\pm 3\%$ for specific conductance at 25°C. Figure 1(a) shows the locations of the MDMR/USGS stations (black labels). The instruments are mounted near the sea-floor and measure temperature, salinity, and pressure (water level). The National Data Buoy Center (NDBC), a part of the National Oceanic and Atmospheric Administration's (NOAA) National Weather Service (NWS), both operates coastal marine stations in the area, and serves data from other organizations that continuously monitor meteorological and oceanographic conditions in the coastal GoM. NDBC performs automated quality control on measurements before distribution.⁵⁰ The locations of NOAA/NDBC stations used in this study area are shown in Fig. 1(a) (red labels). Stations in Alabama waters around the Mobile Bay estuary with continuous salinity records were used, i.e., Dauphin Island, AL (DPHA), Perdido Pass, AL (PPTA), Middle Bay Light, AL (MBLA), Bon Secour, AL (BSCA), and Cedar Point, AL (CRTA). Wind data from an offshore buoy 44-nm southeast of Mobile Bay Main Pass near Orange Beach (OB), AL, at 28-m depth, were also analyzed. Salinity data from The University of Southern Mississippi buoy (NDBC Station 42067) and current data from the FOCAL (Fisheries Oceanography in Coastal Alabama) mooring, both at the 20-m isobath on the inner-shelf, were also used. Table 1 provides the coordinates, names, and measurements available from each station. River discharge data were obtained from four USGS stations from the Alabama, Mobile, Pascagoula, and Pearl Rivers. Note that the river discharge stations are not included in Fig. 1, but the coordinates can be found in Table 1.

Table 1

List of station names, coordinates, and measurements used for salinity (S), temperature (T), water level (WL), wind (W), currents (C), and river discharge (R).

3.1.

Navy Coastal Ocean Model Salinity and Currents

NCOM model predictions for sea surface salinity (SSS) and surface currents covering the Mississippi Bight and Sound were used to understand the dynamic processes and chronological meteorological and oceanographic events that forced offshore saline waters toward the MSS during October 2015. NCOM surface salinity in Fig. 3 shows the inflow of saline shelf waters into the Mobile Bay and MSS through the multiple barrier island inlets from October 19 to 28, 2015. Due to strong easterly and southeasterly currents, low-salinity estuarine waters were predicted to be trapped inside the Sound until the passage of Patricia’s remnants on October 27, 2015.

Fig. 3

NCOM predictions of SSS and surface currents in the study area on (a) October 18, 2015, (b) October 20, 2015, (c) October 22, 2015, (d) October 24, 2015, (e) October 26, 2015, and (f) October 28, 2015.

Velocity measurements (not shown here) at an inner-shelf mooring FOCAL station, located $\sim 20\mathrm{km}$ southwest of the Mobile Bay Main Pass [see Fig. 1(a) for location], showed that the currents were directed onshore, mainly in the NW direction from October 18 until the instrument was buried due to Patricia’s remnants on October 27, 2015. NCOM results in Fig. 3 show an inflow of high-salinity shelf waters via the Horn Island Pass before October 19, 2015 [Fig. 3(a)]. The model results show that higher salinity shelf waters reached the southern coastline of almost all barrier islands, entered the estuarine system mainly via the Mobile Bay Main Pass, mixed with estuarine waters, and then flowed west into the MSS via Grant Pass (Pass aux Herons) [Figs. 3(b)–3(e)]. Because of the wind reversal associated with Patricia’s remnants, estuarine waters were expected to be flushed out of the MSS and Mobile Bay as predicted by the model on October 28, 2015 [Fig. 3(f)]. Satellite imagery following passage of Patricia’s remnants (not shown here) also showed high-chlorophyll-a concentrations due to strong mixing, sediment resuspension, and increased biological activity in the freshwater plumes on the inner-shelf.⁵¹

Hovmoeller diagrams of NCOM SSS (0 to 2 m) along three chosen transects [Fig. 4(a); dashed lines] around the barrier island inlets are shown in Fig. 4. Figure 4(a) shows a zoomed-in map of the MSS barrier islands and passes. The southern-most transect is outside the Sound along 30.2°N and is aligned with Pelican, Petit Bois, and West Ship islands [Fig. 4(b)]. The salinity along the eastern side of the transect (${<88°40}^{\prime }\mathrm{W}$) shows salinities of inner-shelf waters above 32 psu, reaching and exceeding 34 psu. Lower salinity waters were predicted by the model to flush out of the Mobile Bay between October 11 and October 18, 2015, and later from October 27 to November 3, 2015, after the passage of Patricia’s remnants and a subsequent cold front on October 30, 2015. The flushing of lower salinity water from the Sound starting from October 27, 2015, was predicted to be stronger from the inlets around the Petit Bois Island most likely via both the Horn and Petit Bois passes. Lower salinity estuarine waters were predicted to continuously flush out of the Ship Island Pass during the entire month of October, while the intensity of flushing decreased between October 18 and 27, 2015.

Fig. 4

(a) Map of the MSS showing the major barrier islands and passes. NCOM predictions of sea surface salinity in between October 11 and November 11, 2015, through transects, (b) outside the Sound along 30.2°N(30°12’N), (c) along the passes at 30.24°N(30°14’24”N), and (d) inside the Sound along 30.3°N(30°18’N). Transect locations are shown in (a) as dashed lines.

Figure 4(c) shows the NCOM SSS through a transect north of the Mississippi barrier islands at 30.24°N crossing Dauphin Island and Fort Morgan, AL. This transect may be considered as the transition in between estuarine waters and shelf waters because it shows the signature of both lower salinity estuarine waters ($<25\mathrm{psu}$) and higher salinity shelf waters ($>34\mathrm{psu}$). The model predicted strong northward transport of high-salinity waters through Mobile Bay Main Pass starting from October 18, 2015, and the inflow associated with this transport seems to prevent the flushing of Mobile Bay estuarine waters onto the shelf. The diagonal pattern of low-salinity waters from October 18 to October 27, 2015, shown in Fig. 4(c), suggests that the blockage of outflow at the Mobile Bay Main Pass and the strong easterly winds and currents caused the estuarine waters to be transported west from Mobile Bay toward the MSS. Therefore, it is likely that the low-salinity waters flushing out of the Horn Island and Petit Bois passes on October 27, 2015, could possibly be Mobile Bay estuarine waters transported westward during the 10-day period leading to the Patricia’s remnants passage over the study area.

The last transect is inside the MSS and Mobile Bay at 30.3°N. Low-salinity ($<30\mathrm{psu}$) estuarine waters dominated throughout the period as shown in Fig. 4(d). The model predicted high-salinity shelf waters reaching this transect north of Mobile Bay Main Pass starting from October 19, 2015, with increasing intensity toward October 27, 2015. The salinity through this transect also showed an earlier high-salinity inflow via the Petit Bois Pass mixed with estuarine waters. These relatively high-salinity waters ($>30\mathrm{psu}$) were observed to be transported west between October 18 and October 27, 2015, possibly due to westward transport of Mobile Bay estuarine waters.

3.2.

In situ Measurements in the Mississippi Sound and Mobile Bay

Figure 5 shows 40-h moving average filtered wind measurements at the OB buoy [Fig. 1(a); OB], water level, and salinity measurements (blue lines) at the Dauphin Island station [Fig. 1(a); DPHA] compared to the model predictions (red lines) during October 2015. Figure 5(a) and 5(b) show that the variability and magnitude of the East–West(u) and North–South(v) components of wind forcing from the COAMPS solution, used to force NCOM in October 2015, compare well with the measurements on the shelf at the OB buoy station southeast of the Mobile Bay Main Pass. The correlation coefficients between the u(E–W) and v(N–S) components of COAMPS wind variation and the measurements in October 2015 were $r=0.939$ and $r=0.967$, respectively. Such high correlations indicate that the model forcing captured the wind variability on the shelf while the root mean squared error of modeled winds compared to the measurements is $5\mathrm{m}/\mathrm{s}$. Starting from October 16, 2015, the wind was predominantly southeasterly and easterly until a strong wind reversal happened due to Patricia’s remnants on October 27, 2015. The cross correlation of the model wind results with the measured winds showed that the u-component (E–W) of the model winds has a lag time of 2 h in regard to the field measurements, whereas the v-component (N–S) has no lag.

Fig. 5

Measurements (blue) versus NCOM model predictions (red) of (a) east–west wind speed at the OB buoy, (b) north–south wind speed at OB buoy, (c) water level at Dauphin Island station (DPHA) above Mean Lower Low Water (MLLW), and (d) salinity at Dauphin Island station (DPHA) in October 2015. Shown measurements and model results were 40-h moving average filtered.

Figure 5(c) shows that the timing of model predictions for the water level peak and flushing was close to the measurements, with the peak magnitude matching the observed at the Dauphin Island station. The water surface elevation increased inside the MSS and Mobile Bay because of the trapping of estuarine waters combined with the surge. The salinity prediction of the model for October 2015 [Fig. 5(d)] was also in line with the measurements at Mobile Bay Main Pass (DPHA) with a correlation coefficient of $r=0.908$ for the month of October. Measurements showed that the salinity began to increase on October 18, 2015, and peaked at 33 psu during Patricia’s influence on October 27, 2015. Salinity then decreased with Patricia’s wind shift followed by several cold fronts to a value of 22 psu by early November 2015. The model salinity predictions were closer to the measurements between October 1 and 17 after which the model started consistently over-predicting the salinity. During the high-salinity event, the model consistently predicted between 1 and 2 psu above the measurements. Model-data comparison for winds indicate that the easterly winds are stronger in the model possibly causing a stronger inflow of estuarine waters into the estuarine system via Mobile Bay Main Pass as shown in Fig. 4(c). On October 27, 2015, the model salinity peak was 34 psu, compared to 33 psu measured at DPHA station. The root mean squared error for the modeled salinities at DPHA station compared to the measurements for the month of October 2015 was 1.77 psu while a comparison of modeled and measured salinities at the USM buoy at 20-m isobath [Fig. 1(a); USM] showed that the model results were within 0.5 psu of the measurements. Overall, the model predictions were in agreement with the measurements in capturing the trends during the time period of the inflow event in October 2015 validating the model’s capability to represent the estuarine-shelf exchange at the barrier island inlets. The discrepancies between model and measurements for high-salinity values during the inflow event may be because the ambient salinity on the shelf or in the estuaries may be higher in the model causing higher salinities predicted by the model during the inflow event. In addition, the horizontal diffusion or the vertical mixing of the 1-km model at these small scales for the shallow waters of the nGoM may result in such discrepancies between the model and the measurements at the mouth of Mobile Bay.

Surface salinity measurements at the USM buoy (results not shown here), located at the 20-m isobath [Fig. 1(a); USM], showed the salinity at the inner-shelf exceeded 30 psu in July 2015 and stayed over 30 psu until the end of the calendar year. Moreover, the USM buoy showed the salinity in October remained over 33.5 psu for the majority of October 2015 until a 2-psu drop from 34.5 to 32.5 psu due to Patricia’s remnants moving over the area. Figure 6 shows salinity measurements (blue) recorded at 30-min (DPHA, CRTA) or 6-min (M14) frequency at stations in the study area along with the 36-h moving averaged salinity variations (red) to filter out the tidal fluctuations. Figure 6(a) shows salinity at the Dauphin Island station [Fig. 1(a); DPHA] recorded every 30 min (blue) between 2011 and 2015, and Fig. 6(b) shows the salinity for 2015 only. Salinity increased during the spring and summer for all years and peaked during the fall months. It is seen that the salinity rarely exceeded 30 psu between 2011 and 2015. In fact, salinity only exceeded 30 psu for a combined duration of less than 12, 9, 6, 3, and 23 days during 2011, 2012, 2013, 2014, and 2015, respectively. Therefore, we choose 30 psu as a high-salinity threshold at this station and north of the barrier islands. Salinity exceeded this threshold at DPHA only episodically and for short intervals. In fact, fall 2015 was the only period when the salinity exceeded and persisted over 30 psu for an extended period of time (7.5 days) and over 29 psu for 11 straight days. Figure 6(b) shows that salinity records at Dauphin Island station, DPHA, remained over 30 psu for most of the second half of October. A similar persistent salinity increase was also observed in other stations in the MSS.

Fig. 6

Salinity measurements (raw data in blue and 36-h filtered data in red) at (a) DPHA station from 2013 to 2015 and a subset for only 2015 data for stations at: (b) DPHA, (c) CRTA, and (d) M14.

The salinity measurement at CRTA station [Fig. 1(a)] on the eastern side of the MSS close to Mobile Bay is shown in Fig. 6(c). Since this station is north of the barrier islands and inside the MSS, the peak salinity was lower than the DPHA station due to the proximity to low-salinity freshwater sources and mixing inside the estuarine system. However, while the salinity at this station reached 25 psu episodically, it exceeded 25 psu in October 2015, and similar to DPHA, it stayed over this value for more than a week. Figure 6(d) shows the salinity measurements at MDMR/USGS station M14 which is on the western end of the MSS [Fig. 1(a); M14]. The salinity exceeded 30 psu in mid-October and stayed above 30 psu for the same late-October duration shown in the other stations. The fact that similar salinity fluctuations were seen not only at the DPHA but also at CRTA and M14 at both ends of the MSS indicates that the inflow of saline waters was a system-wide event impacting the entire MSS.

Since high-salinity events in the MSS are indicative of intrusions of high-salinity shelf water, these events have important implications for potential transport of larvae, pollutants, and toxic algae into the MSS. An important question then is to understand whether the high-salinity signal observed in October 2015 is atypical or not. To do that, the October 2015 event was compared with October conditions in other years. An anomaly study on salinity measurements was conducted for these purposes. A 9-year monthly climatology of salinity values from 2007 to 2015 was generated using salinity from near-bottom temperature and conductivity measurements at MDMR/USGS stations. Monthly mean salinities were calculated for the entire time series at each station. Monthly anomalies were calculated as the difference between the monthly climatology [$\mathrm{clim}(S)$, average from 2007 to 2015] and the monthly mean value of each year (${⟨S⟩}_{\mathrm{mth}}$). Figure 7 shows the salinity anomalies for the last 5 years (2011 to 2015) of the 9-year period, during which interannual fluctuations were apparent. Elevated salinity shown as positive salinity anomaly peaks in late summer and fall months (August to October) in 2014 and 2015 at all stations, and early summer months (May to July) in 2011 and 2012 at most stations. The highest positive salinity anomaly was seen in October 2015 (shown in red) at all MDMR/USGS stations across the MSS. Monthly anomalies for temperature and water surface elevation measurements at MDMR stations (not shown here) were also calculated. Temperature anomalies showed no significant difference between years at all stations. October 2015 had one of the highest sea surface elevation anomalies (up to 50 cm higher water level) during the 2011 to 2015 time frame due to the passage of Patricia’s remnants.

Fig. 7

Monthly salinity anomalies in psu from 2011 to 2015 at the MDMR stations in the MSS.

Figure 8(a) shows the October in situ mean salinities and variability from 2007 to 2015 at the MDMR/USGS stations and the NOAA/NDBC stations in the study area. October mean salinities were the highest in 2015 at all stations. This indicates that October 2015 had more saline offshore water inflow into the MSS and Mobile Bay and toward the station locations. The highest salinities of October 2015 were measured at the MSS stations; M2, M6, M7, M14, DPHA, and PPTA. Stations M2, M6, M7, and M14 were either near the barrier island inlets (M6, M14) or relatively away from the coastline and freshwater sources (M2, M7).

Fig. 8

(a) October mean salinities from 2007 to 2015 at the MDMR/USGS and NOAA/NDBC stations. Number of days when the October mean salinities exceed (b) 30 psu (c) 26 psu at all the stations (gray areas indicate no data). (d) Study area showing station locations.

The DPHA station at Dauphin Island is exposed to saline offshore waters via the exchange through the Mobile Bay Main Pass, while PPTA at Perdido Pass is already located at the Gulf of Mexico coastline directly exposed to the inner-shelf waters. In addition to examining monthly mean salinity, it is also important to know for how long salinity exceeded a certain threshold. Figure 8(b) shows the number of days in October for which salinity exceeded 30 psu. It is clear that 2015 had the most number of days, especially at those stations with the highest monthly mean salinities mentioned above. For M6 and PPTA, the salinity exceeded 30 psu for over 25 days in October 2015, followed by M2 where the salinity exceeded the 30 psu threshold for at least 15 days. At M7, M14, and DPHA, measured salinity exceeded 30 psu for at least 10 days in October 2015 and did not exceed this threshold in any of the other years with available salinity measurements.

Some stations, e.g., M3, M4, M8, M10, M11, and MBLA, were never exposed to salinities above 30 psu due to proximity to freshwater sources, therefore, a lower salinity threshold of 26 psu was also used at all stations between 2007 and 2015. Figure 8(c) shows that the salinity values exceeded 26 psu at least at one station each year, but the exceedance frequency was the highest in 2015. Measured salinity exceeded the 26-psu threshold more frequently in October months of 2014 and 2015. It was found that the salinity exceeded this value for most if not all days in the month of October at the only open water station PPTA. The exceedance ratio was also high at M6 and M7, followed by M2, DPHA, and M14. While the salinity never exceeded 30 psu at M1 and M5, it exceeded 26 psu for more than 2 weeks at M1 and M5. Figure 8 highlights that October 2015 was different from earlier years and that the salinity in all these coastal stations were higher than usual within the 9-year time period of 2007 to 2015.

3.3.

Ocean Color Data

The analysis using the NCOM salinity and the salinity exceedance and anomaly analysis on the in situ measurements demonstrated that October 2015 had a high-salinity event at multiple stations across the system from Mobile Bay to western MSS. Satellite ocean color chlorophyll-a imagery was used to determine the corresponding surface, or more precisely over the first optical depth, biological response associated with the anomalous surface salinity conditions over the broader region. MODIS-Aqua October monthly chlorophyll-a anomaly fields were calculated for the Mississippi Bight from 2003 to 2015 and shown in Fig. 9. The October 2015 monthly chlorophyll-a anomaly has a negative anomaly (less chlorophyll-a than the monthly climatology) across the entire Mississippi Bight except very near the Mississippi river outlets in the Bird Foot Delta. The only other year with negative chlorophyll-a anomaly to such a great extent was 2003. The other years at least have a positive anomaly either outside the MSS on the shelf or inside the MSS. The reason for most years to have positive chlorophyll-a anomaly either inside the MSS or just south of the barrier islands is probably high chlorophyll-a associated with freshwater sources inside the MSS and plume waters coming out of the estuarine system. Figure 10 shows the October monthly chlorophyll-a anomaly of each year along transects (Fig. 1; red lines) inside the MSS [Fig. 10(a)] north of the barrier islands and outside the MSS [Fig. 10(b)]. Ocean color data clearly shows that October 2015 has the largest negative chlorophyll-a anomaly both inside and outside the MSS. It is important to notice that 2015 is not the only year with such high-negative chlorophyll-a anomaly ($-4.0$ to $-5.0\mathrm{mg}/{\mathrm{m}}^3$) at both transects. October 2003 had a similar chlorophyll-a anomaly.

Fig. 9

October monthly mean MODIS-Aqua chlorophyll-a anomaly in the Mississippi Bight from 2003 to 2015.

Fig. 10

October monthly mean MODIS-Aqua chlorophyll-a anomaly ($\mathrm{mg}/{\mathrm{m}}^3$) from 2003 to 2015 along a transect (a) inside the MSS and (b) outside the MSS. Transect locations plotted in Fig. 1(a) as dotted red lines.

While chlorophyll-a anomaly is generally positive inside the MSS due to freshwater dominance that brings sediment and nutrients, it is generally negative outside the MSS due to offshore water dominance. October 2007 and 2013 had low-chlorophyll-a anomalies ($-1.0$ to $-3.0\mathrm{mg}/{\mathrm{m}}^3$) inside the MSS similar to October 2015, but not necessarily as negative a chlorophyll-a anomaly as October 2015 outside the MSS. So, the negative chlorophyll-a anomaly outside the MSS may be attributed to the fact that the shelf was covered by low-chlorophyll-a/less turbid offshore saline waters outside the MSS in October 2015 and the estuarine low-salinity high chlorophyll-a, and turbid waters were not flushed out of the MSS preventing the formation of high-chlorophyll-a plumes during most of the month.

An ANOVA was performed on the October monthly anomaly values to see the difference in chlorophyll-a anomaly by year followed by a multiple comparison test on the monthly mean chlorophyll-a anomaly to determine which years had a chlorophyll-a anomaly distribution either similar to or significantly different from October 2015 inside and outside the MSS [see Fig. 15 in the Appendix]. Outside the MSS, the October 2003 anomaly was the only one similar to October 2015 and all the other years were significantly different. Inside the MSS, the multiple year comparison showed that October 2003 and October 2007 were similar to October 2015 with low chlorophyll-a. The other 10 years were found to be significantly different and have higher chlorophyll-a than 2015, but the October 2013 results had the closest anomaly means [Fig. 15 in the Appendix) and was similar to October 2015 in transect lines both inside and just outside the MSS.

3.4.

Meteorological and Hydrological Data

It is important to understand the forcing mechanisms that generated the high-positive salinity anomaly along with the low-negative chlorophyll-a anomaly in October 2015. For this reason, the meteorological data (i.e., wind, water level) and hydrological data (i.e., discharge from freshwater sources into the MSS and Mobile Bay) were analyzed to understand the balance between high-salinity/low-chlorophyll-a offshore waters and the low-salinity/high-chlorophyll-a freshwater sources.

The role of wind forcing was assessed using a correlation analysis between the wind measured at the OB Buoy [Fig. 1(a); OB] and the adjusted water level (i.e., inverse barometer effects removed) and salinity data at the Dauphin Island station [Fig. 1(a); DPHA]. The component of the wind forcing with the maximum correlation was determined by calculating a lagged correlation between wind vector component contribution at 5-deg intervals and the scalar parameter of interest (water level or salinity) during the month of October. The highest correlations between the wind component and water level were along the orientation of 290 to 330 deg/110 to 150 deg (i.e., NW/SE axis) with $r$-values of 0.90 to 0.93 with lags of 12 to 23 h. Similarly, the highest correlations between the wind component and salinity were along the orientation of 315 to 350 deg/135 to 170 deg (i.e., NNW/SSE axis) with $r$-values of 0.80 to 0.92 corresponding to lags of 0 to 2 h. These results are consistent with the combined effects of coastal Ekman circulation driving coastal set-up and set-down via along-shelf wind forcing, as well as direct wind forcing from the N/S component in the shallower coastal areas where the Ekman boundary layers would be expected to overlap (i.e., water is being directly pushed onshore contributing to the coastal set-up or set-down). The wind-driven changes in coastal water level result in estuarine-shelf exchange that alter the estuarine salinity as observed. Salinity is somewhat more sensitive to the N/S component as indicated by the more NNW/SSE orientation and shorter lag time (i.e., direct wind response would be expected to be faster than the local Coriolis timescale). This correlation analysis shows that the wind forcing was the primary driver of the low-frequency salinity variability during this period and that wind generally from the southeast quadrant is favorable for high-salinity intrusion events.

The correlation analysis (above) showed positive correlation between wind and salinity in the 40- to 215-deg interval with decreasing correlation for directions larger than 180 deg. Given that southerly to easterly winds could drive salinity intrusions, the wind from the 45- to 180-deg direction is considered favorable for forcing the transport of saline offshore waters toward all the barrier islands around the MSS including the N–S oriented Chandeleur Islands. In particular, 13 years (2003 to 2015) of wind data for October at the Dauphin Island station (Fig. 1; DPIA) were analyzed. An important factor to consider is the wind persistence for a nonstop consecutive time. Consecutive winds in certain directions will force saline offshore waters toward the MSS and Mobile Bay, as well as prevent estuarine waters from flowing out of the MSS and Mobile Bay. We found that the number of consecutive hours in October since 2003 with the winds within the 45- to 180-deg interval was the highest in 2015 with 204 consecutive hours. This is approximately an 8.5-day period between October 19 and 27, 2015. October 2013 winds follow with 143 consecutive hours, October 2004 with 132, and October 2007 with 122 consecutive hours of wind within the favorable 45 to 180 deg interval. The October monthly chlorophyll-a anomaly for 2003 and 2007 was statistically similar to October 2015, however, the October 2013 salinity anomaly was not as high as that of 2015. Unfortunately, there were no salinity measurements available from 2007 to compare.

It is not only the persistence of wind from favorable directions, but also the strength of the wind that will impact the intensity of forcing. Therefore, wind roses were created for those periods when the wind at the DPIA station was consecutively within the 45- to 180-deg interval to visualize the directional spread within the interval along with the wind speed as shown in Fig. 11. The time period in which the wind was persistently within the 45- to 180-deg interval is shown below the wind rose of each year. The wind speed exceeded $15\mathrm{m}/\mathrm{s}$ only in October 2015 (Fig. 11; at the bottom), 2004, and 2006; and for all these years, wind was from E-SE for more than 50% of the favorable wind window. While the wind speed was high in 2006, the number of consecutive hours was lower (64 h). On the other hand, 2004 winds were strong, but Figs. 9 and 10 indicate that this was a high-chlorophyll-a year possibly due to high-freshwater discharge both from the Mississippi River onto the shelf and/or from other sources into the MSS.

Fig. 11

October wind roses from 2003 to 2015 during the time windows when the wind was uninterruptedly from the [45-180°] direction interval.

Figure 12(a) shows the river discharges measured at USGS stream gauges at Pearl, Pascagoula, Alabama, and Mobile Rivers (see Table 1 for station locations). In 2015, the river discharge was lowest in August, September, and November at all river systems in the area. Both the Pearl and Mobile Rivers are good indicators of the variability of freshwater input intensity in our region because of their discharges into the MSS and Mobile River, respectively. Therefore, the October discharges from 2007 to 2015 for those two rivers were analyzed in Figs. 12(b)–12(i). October 2009 was not shown in Fig. 12 due to the extremely high-discharge offsetting the $y$-axis. Measurements from the Pascagoula and Alabama rivers were not shown in Fig. 12 due to their similarity to the Pearl and Mobile Rivers’ discharges, respectively.

Fig. 12

(a) Calendar year 2015 discharge measurements at Pearl, Pascagoula, Alabama, and Mobile Rivers. (b)–(i) October discharge measurements at Pearl River (blue) and Mobile River (green) from 2007 to 2015. October 2009 is not shown due to extremely high-discharge offsetting the $y$-axis.

The discharge at Pearl River was seasonally low and around $50{\mathrm{m}}^3/\mathrm{s}$ for almost all years and was generally less than $250{\mathrm{m}}^3/\mathrm{s}$ for the Mobile River. The Mobile River showed higher fluctuations within October 2015, however, discharge variations were similar for both rivers. At Pearl River, October 2007 and 2010 had the lowest discharges, along with 2015, until the discharge peaks after passage of cyclone Patricia’s remnants in late October 2015. At Mobile River, the discharge was very low between October 19 and 26, 2015, but the lowest October discharge year was 2010, followed by 2011. While high-freshwater input into the Mobile Bay and MSS is expected to decrease the likelihood of having elevated salinities inside the estuarine system, similarity in the intensity of discharge for all years implies that the low rainfall and low discharge in the area were not a significant contributing mechanism for the increased salinities inside the MSS and Mobile Bay while the wind over the area is the primary driver for the inflow event.