The climate of desiccation in the SW Cape

Abstract. Hydro-meteorology conditions in the Southwest Cape of South Africa are analyzed for historical trends in satellite and station measurements. Results show an increase of coastal upwelling, low-level subsidence and shorter winters. The shearing by offshore easterly winds causes a circulation over the SW Cape which entrains dry air from the south coast upwelling zone. Potential evaporation exceeds precipitation and streamflow discharge has declined particularly northwest of the Hottentots Holland mountains. Many of Cape Town's water reservoirs are drying up, and show steep in-creases in surface temperature (+.2 C/yr) and browning of perimeter vegetation. The unfavorable wind shear is compounded by negative sensible heat flux and a capping inversion, so alongshore winds and mountain-top clouds divert seaward, desiccating the upper Berg River catchment.


Introduction
The southwestern (SW) Cape of South Africa (34S, 19E) lies at the transition between the subtropical easterly and mid-latitude westerly wind regimes, and has a semi-arid, rainy-winter climate.
The adjacent interior (Karoo) has sparse vegetation fronted by coastal mountains over 1000 m.Past research on climate change at country-scale has found a ~0.02°C/yr increase of temperature (Kruger and Shongwe 2004;Morishima and Akasaka 2010) consistent with global averages, and mixed trends in rainfall (Tadross et al 2005;MacKellar et al 2014).Yet, water resources within reach of Cape Town's 4 M people (west of 20E) have dwindled to unsustainable levels (13% of storage capacity = 217 M m 3 < www.dwa.gov.za/Hydrology/> as of April 2018).Water supplies are so scarce that numerous engineering projects are underway (Muller 2017), in addition to rationing of 50 L/day/person.Here an analysis of meteorological factors underlying the water deficit is conducted to promote awareness and strategic planning.During this era, global greenhouse gases have risen 2 ppm/yr, national industrial emissions exceeded 600M T/yr (Bekker et al 2008;DEAT 2009), and continental agricultural emissions reached 1350M T/yr (Semazzi and Song 2001;Sinha et al 2003).

Data and Methods
The SW Cape of South Africa (SA) has a dense network of rainfall, streamflow and potential evaporation stations maintained by the Dept of Water Affairs (DWA).Here monthly long-term records with > 90% completeness are obtained in the Upper Berg River catchment, (lat/lon): -33.65/19.01; -33.83/19.08; -33.93/19.06; -34.32/18.99.High resolution NOAA MODIS satellite data are analyzed for land surface temperature and vegetation color.The dispersion of urban emissions is represented by OMI satellite NO2 measurements (Bucsela et al 2013)  Monthly averages over the key area: 34.5-33.5S,18.5-19.5Eare used to calculate trends by linear regression, as in (Jury 2013(Jury , 2018)).The rate of change or slope is evaluated across the field (for maps and sections) and key area (for time series) over the period 1980-2017 using the above reanalysis datasets.Statistical evaluations use the Pearson's 2-tailed t-test and degrees of freedom.
With a DF = 37, 90% confidence is achieved by r > |0.27| or r 2 > .06.Some records are longer: the DWA stations 1956+, GPCC7 and CRU4 rainfall / PDSI 1901+; while some are shorter: MODIS satellite data 2000+.Ranking the satellite land surface temperature, a hot dry spell of 1-8 January 2011 is analyzed using reanalysis fields and SA Weather Service hourly observations and radiosonde profiles from Cape Town airport (CPT).The mesoscale structure of SW Cape hydrology is linked to the uptake of climate change from rising greenhouse gases, and natural multi-year fluctuations (Poccard et al 2000).

SW Cape local and regional trends
The SW Cape urban NO2 emissions (Fig 1a range typical of winter, there was a reduction from 18 to 12 months.In the 30-32.5Crange typical of summer, there was huge increase from 3 to 24 months.The histogram indicates a desiccating regime, which is placed in context below.

Temporal characteristics
In this section, a temporal analysis is presented for station observations in the key area.Ensemble   2. Desiccation is most significant in the December to April months .
) illustrate the atmospheric impact of 4 M residents and their resource needs.Higher pollution concentrations at 33.75S, 18.75E extend northward in a broad arc, according to OMI satellite measurements averaged 2010-2017.Values > 10 μg /m 3 are consistent with in-situ data (SOER 2017).The wind and rainfall trend map (Fig 1b) illustrates a cyclonic circulation and drying to the northwest of the mountains.SST trends (Fig 1c) are positive offshore but negative nearshore (cf.Rouault et al 2010), and point to intensification of coastal upwelling by the southeasterly winds 1980-2017.The offshore Ekman transport and cyclonic wind shear lifts cool water over the shelf, with desiccating effect as seen below.Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-223Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 2 May 2018 c Author(s) 2018.CC BY 4.0 License.Land surface temperatures (Fig 2c) display increases of 0.1 C/yr (Fig 2a) in the MODIS era (2000-2017).Values are highest at the Berg River dam near Franschoek (0.2 C/yr), ~10 times above global and national means (Kruger and Shongwe 2004).Such a high rate of warming is caused by receding water around the perimeter of the reservoir, and heating of bare soil during periods of high solar radiation (Appendix 1 photo; cf.Earth observatory 2018).In conjunction, the vegetation color fraction has declined around the Berg River dam near Franschoek (Fig 2d).A 'browner' surface is also found in the Theewaterskloof Reservoir (-.02 /yr) and west of the Hottentots Holland mountains.Yet greening is noted in the MODIS era (2000-2017) to the east of 19.1E (+.01 /yr), and Chirps2 rainfall trends are weakly positive there (Fig 2e) due to the up-slope of easterly winds.The main hydrological concern is the diminishing rainfall northwest of the Hottentots Holland, across the populated Cape Flats.The annual rain trend is -.5 mm month -1 /yr in the vicinity of Wellington (33.64S, 19.00E).The histogram of NOAA infrared vegetation temperature, comparing the 1982-88 and 2010-16 era (Fig 2f), illustrates shifts in both cold and warm seasons.In the 12.5C rainfall trends per month (Fig 3a) are downward except for small rises in August and November.Months with significant declines include January, March, October and December.Months with the largest drying trends are May and September, at the beginning and end of winter.Streamflow discharge shows a downward trend in the Upper Berg River catchment (Fig 3b).Of note are the low summer flows in the early 1970s that culminated in three dry years: 1978-1980.Streamflow discharge recovered over a lengthy era from 1981 to 2014, but then declined in both winter and summer to almost zero by 2017 (end of record).
Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-223Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 2 May 2018 c Author(s) 2018.CC BY 4.0 License.A longer term perspective is offered by CRU4 PDSI anomalies (Fig 3c) that show a persistent downward trend through the 20 th century, consistent with Wolski (2018).The 1970s dry spell recurred in the most recent decade (2010+).Apart from the 1950s and 1980s, the water budget has shown deficit conditions: potential evaporation exceeded precipitation.The downward trend in PDSI is 17% of total variance.The DWA pan evaporation measurements (Fig 3d) have a weak upward trend since 1956.There is a large annual cycle that peaks in summer each year, when desiccating weather prevails.Seasonal variability tends to obscure the climate change signal, so individual monthly trends are considered.The linear upward trend in potential evaporation accounts for 15% of variance from January to March with a summer-time slope of 0.15 mm day -1 /yr in the period 1956-2017.Similarly, the satellite net OLR shows positive trends in the key area (reduced cloud cover), as listed in Table . A surprizing result is that subsident trends within the easterlies are strongest south of the Langeberg mountains (20-24E).Air temperature trends are relatively weak at .01 C/yr in the 1000-900 hPa layer.Warming increases westward consistent with a change from moistening 21-24E to drying 16-19E.Turning attention to the shelf oceanography, it is evident that upwelling has intensified (-.02 C/yr) from surface to 40 m depth to the east of 19E, despite a warming trend elsewhere in the period 1980-2016.The warm air / cool sea trend causes sensible heat flux (Qh) to decline -.2 W m -2 /yr.Hence the lower atmosphere is stabilized along the South Cape coast.As a secondary consequence, the subsidence inversion strengthens and caps the surface easterly flow, as outlined inJury and Reason (1989).The Froude number, calculated from F = U ∕ NH, offers a way to determine whether the airflow will ascend the SW Cape mountains (F≥1) or go around them (F<1).With a westerly wind of U = 10 m/s, a mountain height of H = 10 3 m, and a winter-time unstable lapse rate of N = Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-223Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 2 May 2018 c Author(s) 2018.CC BY 4.0 License.((g ∕ θo)(dθ ∕ dz)) 0.5 ~ 10 -2 , the airflow lifts over the mountains causing stratiform rainfall.Under the summer-time easterly wind regime, the U and H values are the same.But surface cooling and lowlevel warming induce a stable lapse rate of N ~ 2 10 -2 , so F = 0.5.The airflow is diverted around the mountains instead of going over the Hottentots Holland; leaving them sunny and prone to desiccation.As the sub-tropical ridge shifts poleward, there is a broad zone of intensifying easterly flow from 32-40S, 0-40E (Fig 4e).Evidence of Venturi acceleration at the southern tip of Africa (35S, 20E) emerges in zonal wind trends of -.05 m s -1 /yr (1980-2017).The associated easterly shear spins-up a wind rotor over the SW Cape (cf.Fig 1b) with desiccating consequences.The change of winds from easterly to southeasterly can be traced to three influences: firstly the change in coastal orientation, secondly the day-time thermal gradient / seabreeze, and thirdly the land-friction / Ekman spiral.Seabreeze forcing is: dV = ((g H ∕ θ) dθ ∕ dy) dt with g gravity, H surface layer ~ 30 m, potential temperature θ ~ 300K, dθ ~ 3K sea-land increase, and dy, dt are length, time scales (~ 2 10 4 Ranking the MODIS day-time land surface temperature record in the SW Cape(33.5-34.5S,18.5-    19.5E), the period 1-8 January 2011 is the hottest case(42 C)  in the period 2000-2017.The map of surface temperatures (Fig 5a) reveals a narrow strip of cool conditions along the windward coastal promontories such as Cape Point (20 C).However across the Cape Flats and interior valleys (Berg, Brede) the day-time land surface temperatures exceed 50 C! The air-flow at 850 hPa (Fig 5b) shows Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-223Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 2 May 2018 c Author(s) 2018.CC BY 4.0 License.cyclonic curvature and > 5 m/s easterlies over the ocean to the south.The vegetation color fraction (Fig 5c) diminished 0.1 from the end of December to mid-January 2011 in the Upper Berg River catchment, west of Hottentots Holland mountains.A vertical section analysis shows a strong easterly 'jet' capped by an inversion (Fig 5d) in the period 1-8 January 2011.In the offshore zone (35S), the near-surface winds were 10 m/s and temperatures were < 20 C.During this dry spell, coastal upwelling caused SST < 15 C (Fig 5e).A consequence of warm air overlying cool sea is thermal stability, which inhibits the inland penetration of moisture (cf.Fig 5b).The CPT weather station reported desiccating weather conditions 3-6 January (Fig 5f) characterized by Tmax = 34 C and Tdew = 16 C.Southerly winds strengthened 3-4 January to 35 km/hr and then abated following passage of the South Atlantic high pressure cell.CPT radiosonde profiles on 6 January 2011 (Appendix 2) describe characteristics in the 400-700 m subsidence inversion: 110º / 14 kt winds, 32.6 C temperature, -14.6 C dewpoint, and specific humidity 1.36 g/kg due to entrainment of dry air from aloft (cf.500 hPa sinking motions in Fig 5e).Many DWA stations near Wellington recorded potential evaporation > 12 mm/day!4. Summary Hydro-meteorology conditions and trends in the Southwest Cape of South Africa have been studied using historical satellite and station measurements, and gridded reanalysis fields.Results show an increase of coastal upwelling, low-level subsidence and shorter winters.The sub-tropical ridge has shifted poleward causing an increase of easterly winds along 35S.The wind shear induces a cyclonic rotor over the SW Cape, which entrains dry air from the interior Karoo and south coast upwelling zone.Precipitation has declined particularly northwest of the Hottentots Holland mountains and many water reservoirs show steep increases in surface temperature (+.2 C/yr) and browning of perimeter vegetation since 2000.The unfavorable wind shear is compounded by negative sensible heat flux and a capping inversion, so winds and mountain-top clouds divert seaward.The regime shift from mid-latitude westerly winds and winter rainfall, to sub-tropical easterly winds and summer dry spells has depleted water resources.Hence recycling, desalination, importation, and aquifer Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-223Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 2 May 2018 c Author(s) 2018.CC BY 4.0 License.extraction projects are underway (EWN 2018), and water conservation has become rooted in popular thinking and community awareness.The above results contribute to this educational initiative, and suggest that water deficits in the SW Cape are here to stay.

Table 1
Datasets used in the trend analysis.References are listed in text; web sources are given in Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-223Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 2 May 2018 c Author(s) 2018.CC BY 4.0 License.