Coastal gradients in False Bay, south of Cape Town: 1 what insights can be gained from mesoscale reanalysis?

9 Mesoscale datasets are used to study coastal gradients in the marine climate and oceanogra-10 phy in False Bay, south of Cape Town. Building on past work, satellite and ocean / atmos-11 phere reanalysis are used to gain new insights on the mean structure, circulation and meteoro-12 logical features. HYCOM v3 hindcasts represent a coastward reduction of mixing that en-13 hances stratification and productivity inshore. The mean summer currents are westward – .4 14 m/s along the shelf edge and weakly clockwise within False Bay. The marine climate is dom-15 inated by southeasterly winds that accelerate over the mountains south of Cape Town and fan 16 out producing dry weather. Virtual buoy time series in Dec 2012-Feb 2013 exhibit weather-17 pulsed upwelling in early summer interspersed with quiescent spells in late summer. Inter-18 comparisons between model, satellite and station data build confidence that coupled reanal-19 yses yield opportunities to study air-sea interactions in coastal zones with complex topogra-20 phy. The 0.083° HYCOM reanalysis has 16 data points in False Baythe embayment south of 21 Cape Town, just adequate to resolve the coastal gradient and its impacts on ocean productivi-22 ty. 23


Introduction
The coastal zone south of Cape Town, South Africa, is comprised of linear sandy beaches and a semi-enclosed bay surrounded by mountains (Fig. 1a, b). False Bay is southward facing and about 10 3 km 2 , with the Cape Peninsula to the west and Cape Hangklip to the east. The shelf oceanography exhibits a range of conditions from seasonally pulsed upwelling events (Shannon and Field, 1985;Lutjeharms and Stockton, 1991;Largier et al., 1992;Dufois and Rouault, 2012) to warm-water intrusions from the Agulhas Current, creating great biological diversity Griffiths et al., 2010). The upper ocean circulation tends to be northwestward and pulsed at subseasonal timescales by passing weather, shelf waves, warm rings and tides (Grundlingh and Larger, 1991;Nelson et al., 1991). Coastal winds and temperatures exhibit sharp gradients (Bang, 1971;Jury, 1991;Van Ballegooyen, 1991) depending on excursions of the subtropical anticyclone.
The high-pressure cells of the South Atlantic and South Indian Ocean tend to join in summer and produce dry weather and upwelling-favourable winds from the southeast that are shallow and diverted around the > 1000 m mountains of Cape Hangklip and the Cape Peninsula ( Fig. 1a, b). The winds accelerate off the capes and form shadow zones over leeward bays, creating cyclonic vorticity that enhances upwelling (Wainman et al., 1987;Grundlingh and Largier, 1991;Jacobson et al., 2014). Winds entering False Bay become channelled N-S and tend to induce standing clockwise rotors in the upper ocean (deVos et al., 2014), which are pulsed by geostrophic currents across the mouth.
With the passage of eastward-moving atmospheric Rossby waves across the southern tip of Africa at 3-20 d intervals (Jury and Brundrit, 1992), the subtropical ridge is replaced by coastal lows followed by downwelling-favourable northwesterly winds and frontal troughs that bring rainfall, stormy seas, onshore transport and mixing -most often in winter: May-September (Engelbrecht et al., 2011;Schilperoort et al., 2013;deVos et al., 2014;. The city of Cape Town, its 4 million residents and associated infrastructure have intensified anthropogenic pressure on the coastal zone. Sandy beaches there are vulner- able to sediment loss from rising seas, huge swell events and recreational use (Hughes and Brundrit 1991;Brundrit, 2009;Mather et al., 2009;Theron et al., 2010;Roux and Toms, 2013;Theron et al., 2014a, b;Fourie et al., 2015). Climate change has become manifested in longer summers and a southeastward shift in wind-driven upwelling, marine ecosystems and fisheries (Rouault et al., 2010;Lloyd et al., 2012;Blamey et al., 2012;Schlegel et al., 2017).
Coastal embayments tend to be productive and False Bay is no exception. Brown et al. (1991) reported an average chlorophyll concentration of 4 mg m −3 in the euphotic layer, which varies from summer to winter: 5.5 vs. 2.1 mg m −3 (Giljam, 2002). Nutrients enter the southern coastal zone via runoff and municipal waste streams (Parsons, 2000;Taljaard et al., 2000). Although numerous small rivers drain into False Bay, the nutrients supplied by upwelling exceed those from terrestrial sources (Taljaard, 1991;Giljam, 2002). Coastal and offshore waters show healthy rates of exchange, particularly during stormy spells that induce surf-zone currents and a dissipation of the thermocline.
Our understanding of the physical oceanography south of Cape Town has benefited from studies of the upper ocean circulation (Botes, 1988), the wind field and the variability of sea temperatures (Dufois and Rouault, 2012). Yet many processes governing intra-seasonal variability remain obscure (Wainman et al., 1987). There is a lack of consensus on the mean seasonal circulation (Grundlingh et al., 1989;Taljaard et al., 2000), despite ample knowledge of the air-sea interactions. To overcome the limited scale and brevity of measurement campaigns, modelling efforts (e.g. Penven et al., 2001) have elucidated coastal features over a longer period. Hydrodynamic simulations with temporal forcing by Nicholson (2011) gave promising results, and Coleman (2019) recently modelled the circulation south of Cape Town forced with daily data from the HYbrid Coordinate Ocean Model (HYCOM; Cummings and Smedstad, 2013) and the Weather Research and Forecasting model (WRF; Skamarock et al., 2008). Coleman (2019) found sheared clockwise circulations during summer and favourable validations for mean currents and thermal stratification in False Bay.
Given the above history of scientific endeavours, the objective of this work is to embark on a new mission to utilize the global ocean data assimilation system to describe the spatial pattern and temporal variability of the marine environment. We demonstrate that mesoscale reanalysis offers valuable new insights into the coastal gradient in summer climate and physical oceanography south of Cape Town.

Data
Marine climate variability is described using weather and wave reanalysis products at 20-30 km resolution, namely CFSr2, ECMWF and WAVEWATCH III (Saha et al., 2014;Dee et al., 2011;Tolman, 2002;respectively). Coastal gradients are described using 4 km resolution satellite visible and infrared products (Reynolds et al., 2002) and IMT station observations in western False Bay. Table 1 lists abbreviations and dataset attributes.
The mesoscale oceanography of False Bay, south of Cape Town, is studied with HYCOM v3.1 reanalysis (Cummings and Smedstad, 2013;Metzger et al., 2014), which assimilates microwave, infrared and visible measurements from multiple satellites, calibrated with in situ observations. Climatology, persistence and model-calculated fields are used to quality control and nudge the incoming data, within static 0.033 • resolution GIS fields that include bathymetry, surface rough- ness, etc. Running in parallel with the ocean model are operational atmosphere and land models that deliver coupled information on momentum, heat and water fluxes, and feedbacks (Table 1). In the 41-layer 0.083 • HYCOM v3.1 hindcast employed here, NAVGEM v1.4 3-hourly 0.176 • resolution atmospheric data provide background initialization for kinematic and thermodynamic fields derived from satellite and in situ measurements, continually assimilated over a rolling 5 d window . A hydrological sub-model assimilates satellite rainfall-soil moisture and predicts runoff, which is blended with satellite salinity measurements (Table 1). Regional validations have been done for the HYCOM reanalysis, and errors for key variables are < 10 % Metzger et al., 2017). Local validations are reported here in Fig. A1. Hindcasts differ from operational forecast simulations in that the rate of change and evolution of spatial structure is known; the rolling 5 d analysis window has overlapping temporal information to ensure a close fit to environmental conditions. This is crucial for infrequent zenith altimeter data which prescribe the currents. Post-2008 reanalysis better represents the nearshore oceanography due to finer microwave footprints that reach the coast. HYCOM reanalysis fields of near-surface sea temperature, salinity, currents and mixed layer depth (MLD) are analysed as mean maps and sections. We focus on the summer of December 2012 to February 2013, which coincides with VIIRS reflectance, Jason-1 and Jason-2 altimeter, and ASCAT-A and ASCAT-B scatterometer coverage that better constrains the physical oceanography. Cross-correlations between the various surface ocean and atmosphere parameters are studied in this 90 d period. Other motivations for our study period include summer's marine productivity (Pfaff et al., 2019) and Table 2. Relative influence of surface weather observations in model assimilation, with grey land mask. Stations reporting (in 2020) land-based private and official and marine-based online and offline . Curved line is routine aircraft wind-temperature profile. Sources: NASA, NOAA, MADIS, WunderMap. the variety of conditions attributable to pulsed upwelling and shelf wave events.
In situ measurements over the coast and shelf south of Cape Town are made by numerous government agencies: South African (SA) Weather Service, Department of Environmental Affairs, Institute of Marine Technology (IMT), Council for Scientific and Industrial Research -Marine Division, SA Department of Water Affairs, and SA Hydrographic Department, with data operationally reported and subsequently archived at the SA Data Centre for Oceanography. The University of Cape Town Oceanography Department hosts short-term projects and regional ocean numerical modelling.
Evaluating the "influence" of surface reports in operational data assimilation (Table 2), values of ∼ 24 % in False Bay contrast with ∼ 90 % inland. This trend continues for upper ocean T /S observations that are nearly 4 times greater in Table Bay than False Bay (WOA, 2013). Hence our analysis of marine conditions over the shelf south of Cape Town relies more on satellite and model than in situ observations.
Comparisons of HYCOM reanalysis ocean data with daily gauge and radiometer measurements show reasonable agreement (cf. Fig. A1a, b) in the period 2008-2015. The sea surface height comparison has a 24 % fit with discrepancies attributable to coastal tide residuals and non-co-location. Sea temperatures have a 38 % fit and diverge in warm spells, the model tending to over-estimate. Comparisons of ECMWF-5 reanalysis and Simonstown station hourly weather data in the period December 2012-February 2013 (cf. Fig. A1c, d) are good for pressure (88 %) and wind speed (62 %) but lower for air temperature (21 %) presumably because the 0.3 • reanalysis has contributions from land. Coleman (2019) reports similar validations for the summer of 2010.
The HYCOM reanalysis has limited atmospheric outputs, so to evaluate the wind circulation south of Cape Town, the WRFv3.8 model (Skamarock et al., 2008) is used to downscale ECMWF fields, as in the simulations of Coleman (2019). The WRF model resolution of 0.1 ‰ complies with the HYCOM reanalysis and uses default schemes for boundary layer, flux transport, radiative transfer and surface coupling. We focus on the nature of horizontal flow over False Bay during summer December 2012-February 2013, a period of "near-normal" climate (e.g. sea level air pressure anomaly ∼ 0 hPa).

Summer climate and weather
We first consider the coast and climate before analysing the shelf and ocean. Warm dry weather and sparse vegetation characterize summer (Fig. 2a, b). Satellite land surface temperatures exhibit sharp gradients from the Cape Flats (40 • C at 34 • S) to cool southern coasts (25 • C at 34.4 • S), similar to Tadross et al. (2012). Little rainfall occurs in summer, so terrestrial vegetation is depleted and ocean salinity is controlled by evaporation and currents, not terrestrial run-off. Figure 2c and d illustrate the spatial pattern of ECMWF WRF-downscaled surface winds over the False Bay region in morning and afternoon. The mean southeasterly winds pass Cape Hangklip and reach 9 m s −1 in mid-bay. The flow acceleration is attributed to the following: (1) orographic channelling (Venturi effect), (2) vertical constraint by trade wind inversion, and (3) sinking motion from declining coriolis and sensible heat flux (cf. Jury and Reason, 1989). Summer winds are characterized by a low-level wind jet over False Bay, seen in earlier aircraft surveys (Jury, 1991), which is embedded in a shallow moist layer (cf. Fig. A2a). Diurnal variability is of high amplitude as evident below.
Time series of CFSr2 winds over the coast and shelf (Fig. 2e, f) show a meridional component that is positive and steady except for brief reversals at the end of December 2012 and February 2013. The zonal wind component is negative and fluctuating particularly in mid-January 2013. The coastal gradient is small for mean meridional flow: shelf of V = 3 m s −1 vs. coast of 1.6 m s −1 . However, the standard deviation of zonal winds is shelf of U = 6.6 m s −1 vs. coast of 2 m s −1 . During spells of strong easterlies from transient anticyclones, the wind vorticity contribution to coastal upwelling is dominated by the gradient of ∂U/∂y.
Time series of 6-hourly CFSr2 thermal variables (Fig. 2g, h) show large land-sea differences, as expected. Coastal air temperatures fluctuate diurnally from 15-35 • C while shelf temperatures rise gradually from 18 to 21 • C over the summer. Standard deviations vary from shelf 0.4 • C to coast 5.7 • C. The landward increase of temperature drives a sea breeze contribution to the mean meridional flow. The CFSr2 surface heat fluxes show diurnal amplitude 0-300 W m −2 over the coast but stay in the range 50-100 W m −2 at the shelf edge. Hence the 0.2 • CFSr2 captures the coastal gradients that govern the shelf oceanography, with attributes consistent with NAVGEM v1.4 that underpin the HYCOM reanalysis.
Large swings in the meridional wind (cf. Fig. 2f) are accompanied by dips in air pressure < 1005 hPa on 27 December, 29 January, and 9 and 17 February 2013, which identify coastal low passage associated with trapped shelf waves. In the 27 December and 17 February cases, the wind reversed from 15 m s −1 SE (before) to 12 m s −1 NW (after passage of the coastal low). CFSr2 wind vorticity and sub-surface vertical motions in False Bay (point 3) changed from −5×10 −4 s −1 / +0.7 m d −1 (before) to +4×10 −4 s −1 / -0.8 m d −1 (after passage of the coastal low) and sea temperatures dipped < 15 • C. Yet much of the time environmental changes are inhibited in semi-enclosed False Bay, thus sustaining productivity.

Shelf oceanography
In this section we characterize the shelf oceanography south of Cape Town. The shelf edge has cooler waters and lower salinity due to upwelling (Fig. 3a). Equatorward winds drive surface currents into False Bay, trapping a warm salty zone against the north coast > 35.3 g kg −1 (Fig. 3a), consistent with Dufois and Rouault (2012). The summer water flux is negative across the region during summer (Fig. 3b), as evaporation of 4-6 mm d −1 exceeds precipitation of 1-2 mm d −1 . Fast and divergent winds desiccate the Cape Peninsula in contrast with orographic lifting over the eastern mountains. Hence the P-E field varies from neutral inside False Bay to strongly negative west of Cape Town. False Bay has a narrow exposure to the Southern Ocean. SW swells of ∼ 3 m tend to refract into the bay producing greater energy on the east side (Fig. 3c). The ocean mixed layer depth ranges from < 10 m inside False Bay to > 50 m outside, due to kinematic exposure and thermal stratification. Mean currents (Fig. 3d) are weak in the northern half of False Bay but westward at the shelf edge and drawn into the Benguela Current.  Table 2 statistics and the depth section in Fig. 4a-d. Winds and currents are sheared into clockwise gyres that increase water residence time enabling nutrient buildup and phytoplankton blooms within False Bay (satellitederived chlorophyll > 10 mg m −3 ; Fig. 3e). Month-to-month changes in productivity relate to wind angle, intensity of pulsed upwelling (cf. Fig. A2b) and prevalence of rotary circulations.  Shelf-edge upwelling creates a wedge of 12 • C, 34.7 g kg −1 waters below 60 m. Zonal currents are weak inshore and strongly westward at shelf edge above 20 m. HYCOM meridional currents reveal an overturning circulation, with deeper offshore flow and very shallow onshore flow. HYCOM daily time series at three points along 18.6 • E exhibit pulsing and cooler fresher conditions in the south compared with the north (Fig. 4e-g). There is a strong gradient in zonal currents from −5 m s −1 at shelf edge to zero at the coast.
Statistical analysis is given in Table 3 and reveals that inshore (point 1 at 34.1 • S) sea temperatures are more sensitive to waves than winds, and offshore (point 3 at 34.3 • S) sea temperatures follow zonal winds more than currents. We note that offshore and inshore temperatures are uncorrelated, and offshore salinity is negatively related to inshore temperature. Coastal and shelf-edge salinity are correlated, and inshore salinity responds to zonal currents (−r). HYCOM zonal currents inshore and offshore associate similarly with winds at 1 d lead and -being correlated with each other -suggest that Ekman transport frequently overrides the clockwise gyre.
Time series of WAVEWATCH III swell characteristics at coast and shelf-edge virtual buoys are given in Fig. 4h-j. Swell heights offshore (point 4 at 34.4 • S) oscillate around 2 m except for a spell of stormy seas at the end of December 2012. Nearshore swell heights (point 1 at 34.1 • S) remain near 1 m after attenuation. Southwest swell directions prevail offshore with an occasional swing to southeast. Inshore directions refract to southerly and show little change. Swell periods from 9 to 13 s tend to "bunch" inshore < 8 s. The 25 km W3 reanalysis captures the coastal gradient in swell properties (Johnson et al., 1999), but finer resolution or downscaling would be ideal.  This case study highlights the role played by False Bay at the interface between the Benguela and Agulhas ecosystems. Initially the bay is dominated by cool upwelling waters that are well mixed, but through the summer a warm surface layer forms near the coast -improving productivity (cf. Fig. 3e). The leakage of nutrient-rich upwelled waters back into False Bay is evident (cf. Fig. 4a). Stormy spells can disrupt the coastal gradient (in early January 2013; cf. Fig. 4e, h) that is typical of summer.

Conclusions
Mesoscale datasets were employed to study the marine climate and physical oceanography of False Bay near Cape Town during summer 2012-2013. The 0.083 ‰ HYCOM v3.1 reanalysis offers new insights into the spatial and temporal nature of air-sea interactions, and consistently represents a coastward reduction of mixing that enhances thermal stratification (cf. Figs. 3c, 4a, b). Cross-coast gradients are particularly strong for zonal wind and current, temperature and salinity, and wave height. The reanalysis circulation obtains westward flow across the mouth (−0.4 m s −1 ) and a weak clockwise gyre in mid-bay (cf. Fig. 3d) that sustains productivity (Fig. 3e). The mesoscale features seen here are consistent with Coleman (2019), whose high-resolution model assimilated the very same HYCOM and ECMWF-WRF data. Under summertime southeasterly winds, the clockwise gyre in False Bay was modelled to have inflow/outflow of ∼ 0.2 m s −1 on the upper-west/lower-east side, and a sea temperature increase of ∼ 5 • C from deep offshore to surface inshore. These features (∂V /∂z, ∂T /∂y) are reflected in the HYCOM reanalysis (cf. Fig. 4d, e) and in Coleman (2019, Figs. 6-22, 6-26 therein). Cool nutrient-rich Benguela waters enter False Bay from its southwest corner and infiltrate the bay via the clockwise circulation (cf. Figs. 3d, 4c). The cool wedge underlies a warm layer near the coast during summer (cf. Fig. 4a) that improves productivity.
Temporal variability during summer is dominated by SE winds that accelerate near Cape Hangklip and fan out across False Bay, promoting dry weather. Virtual buoy time series in December 2012-February 2013 exhibit weather-pulsed upwelling, and station intercomparisons build confidence that coupled reanalyses yield opportunities to study air-sea interactions in coastal zones with complex topography. Yet our 0.083 • reanalysis has 16 data points in the embayment south of Cape Town. Finer downscaling could propagate ambiguities from microwave radiometers. Thus we propose that current technology allows many questions to be answered, from coastal processes to climate change. Longer summers in Cape Town could see a shift in resources from land to sea. This sentinel for global impacts on sustainable development needs ongoing scientific assessment in support of holistic management.