State Key Laboratory of Marine Environmental Science, College of Ocean and Earth Sciences, Xiamen University, Xiamen 361002, China
2.
Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), Zhuhai 519082, China
3.
Center for Remote Sensing, College of Earth, Ocean and Environment, University of Delaware, Newark, DE 19716, USA
4.
Joint Center for Remote Sensing, University of Delaware-Xiamen University, Newark, DE 19716, USA and Xiamen 361002, China
Funds:
The National Key R&D Program of China under contract No. 2019YFA0606702; the National Natural Science Foundation of China under contract Nos 42176222, 91858202, 41630963 and41776003; the National Science Foundation under contract No. NSF-IIS-2123264; and National Aeronautics and Space Administration under contract No. NASA-80NSSC20M0220.
The Agulhas system is the strongest western boundary current system in the Southern Hemisphere and plays an important role in modulating the Indian-to-Atlantic Ocean water exchange by the Agulhas leakage. It is difficult to measure in situ transport of the Agulhas leakage as well as the Agulhas retroflection position due to their intermittent nature. In this study, an innovative kinematic algorithm is designed and applied to the gridded altimeter observational data, to ascertain the longitudinal position of Agulhas retroflection, the stability of Agulhas jet stream, as well as its strength. The results show that the east-west shift of retroflection is related neither to the strength of Agulhas current nor to its stability. Further analysis uncovers the connection between the westward extension of Agulhas jet stream and an anomalous cyclonic circulation at its northern side, which is likely attributed to the local wind stress curl anomaly. To confirm the effect of local wind forcing on the east-west shift of retroflection, numerical sensitivity experiments are conducted. The results show that the local wind stress can induce a similar longitudinal shift of the retroflection as altimetry observations. Further statistical and case study indicates that whether an Agulhas ring can continuously migrate westward to the Atlantic Ocean or re-merge into the main flow depends on the retroflection position. Therefore, the westward retroflection may contribute to a stronger Agulhas leakage than the eastward retroflection.
Figure 1. The climatology of mean kinetic energy (MKE) (a) and eddy kinetic energy (EKE) (b) in the Agulhas system during 1993 to 2018, which is computed using the geostrophic velocity from the gridded L4 satellite altimetry product, the difference between MKE and EKE (c) and the time series of EKE and MKE in the black rectangular box (10°-18°E,35°-42°S) in c, which denotes the region where MKE is much less than EKE and is roughly the origin of Agulhas leakage as well (d). The unit for EKE, MKE and their difference is m2/s2.
Figure 2. The climatology of absolute dynamic topography (ADT) (colored shading, unit: m), and the corresponding geostrophic velocity (black vectors) during the period of January 1993–December 2018. The thick black line is the mean location of the Agulhas current jet path derived for each available longitudinal grid (with (1/4)° increment). The grey lines are all the identified Agulhas current jet path. The dashed black line at 26°E approximately divides the Agulhas current jet path into upstream and downstream sections. It is noted that the variation of the paths gets much higher downstream.
Figure 3. Normazlied monthly time series for various physical factors. a. AC strength integrated along the upstream AC jet path (black line) and its 13-month moving average (blue line). b. Same as a, but integrated along the downstream AC jet path. c. Spatial standard deviation along the downstream jet path, the black and the blue lines represent monthly and 13-month running mean respectively. d Longitudinal displacement of the retroflection point, which is defined as the westernmost longitude of the AC jet path. The black , blue and red lines denote the monthly, the 13-month running mean, and the 5-year low-pass filtered displacements, respectively. The grey dashed lines indicate ±1.5 standard deviations from the mean displacements, which is the thresholds for composite analyses. The normalization of the time series is achieved following the formula $ i=\dfrac{x-\mu }{\sigma } $, where $ \mu $ is the climatological mean and $ \sigma $ the standard deviation of $ x $.
Figure 4. The ETOPO1 bottom topography in colored shadings (unit: m) overlapped with the 21 westernmost jet paths (a) and the 24 easternmost jet paths (b), with ±1.5 standard deviation of the retroflection longitudes as selecting criteria, respectively. The black crosses denote the final retroflection points for each path according to our algorithm.
Figure 5. ADT anomalies on the north side are associated with the retroflection position.The composites of the ADT (unit: m) for the 24 most eastward (a) and the 21 most westward (b) retroflection months, and the black vectors denote the corresponding surface velocity. The black rectangular box is approximately where a cyclonic feature coincides with the westward elongation of the AC jet path in the westward composite. c. The blue curve is the 15-d running average of the spatial mean ADT (unit: m) within the black box in a, 45 d before the most eastward retroflection events until 105 d afterwards; and the red curve is the same but referenced to the most westward retroflection events. The grey shading indicates the composited differences between the westward and the eastward retroflection events are significant at the 95% confidence level of t-test.
Figure 6. Wind stress curl anomalies at the north flank lead to the retroflection position shift. a. The wind stress curl anomalies (unit: 10-7 N/m3) corresponding to the composite of the 24 most eastward retroflection events. b.Same as (a) but composite of the 21 most westward retroflection events. The black rectangular is the same as in Fig. 5b, and the grey box is chosen to represent an area of maximal negative wind stress curl anomaly. c. 15-d moving average of the spatial mean wind stress curl in the black rectangular box to the north, from 15 d before the event till 15 d afterwards. d. Same as c but for the mean wind stress curl in the grey box to the south. The grey shading indicates the composited differences between the westward and the eastward retroflection events are significant at the 95% confidence level of t-test.
Figure 7. The mean ADT derived from altimetry data (unit: m) in 2017 (a) and 2002 (b). Notice the cyclone to the north of the retroflection point in 2002.
Figure 8. Configurations of the wind forcing for numerical experiments. The anomaly of the ERA5 wind stress curl and 10 m wind stress vector in 2017 (a) and 2002 (b), referenced to the 1979 to 2019 climatology. The difference between a and b is shown in c. d. Non-dimensional weighting function, which will be multiplied to the wind stress difference (2002–2017) and then added to the case1 forcing to produce the perturbed run (Case 2). The cyan box is the same as the black box in Fig. 6a and 6b, and the dark and light green contours represent the mean SSH to indicate the retroflection position for the year of 2017 and 2002 respectively.
Figure 9. Modeled longitudinal position of the retroflection. Mean SSH (unit: m) in the retroflection area from control run forced by 2017 mean wind stress (a) and the ensemble of the 3 perturbed runs forced by 2017 wind stress (b) plus a local 2002 anomaly around the retroflection area.The black dashed curve represents the 0.5 m SSH isoline for the control run, and the cyan curve represents the 0.5 m SSH isoline for the ensembled perturbed runs. c. the probability distribution of the retroflection longitudes for the first 3 model years, the black curve and circles represent the control run, while the magenta, the green and the red represent the 3 perturbed runs. d. Same as c, but the result from the latter 2 model years.
Figure 10. Time evolution of Agulhas ring shedding process for the westward retroflection events. The composite of 15-d mean ADT leading to extreme westward retroflection events (unit: m). a. Mean ADT over the period 30 d to 15 d before the extreme eastward retroflection events. b and c. Same as a, but from 22 d to 7 d, and 15 d to 0 d before the extreme westward retroflection events respectively. d. Mean ADT from 8 d before the event to 7 d after the event. e and f. Same as d, but for the mean ADT from the day of the event to 15 d afterwards, and from 16 d to 23 d after the event respectively. Red solid contours include areas of negative ADT value. Black dashed lines indicate the 0.7 m ADT value, which has been usually used to estimate the retroflection position.
Figure 11. Time evolution of Agulhas ring shedding process for the eastward retroflection events. The composite of 15-d mean ADT leading to extreme eastward retroflection events (unit: m). a. Mean ADT over the period 30 to 15 d before the extreme eastward retroflection events. b and c. Same as a, but from 22 to 7 d, and 15 to 0 d before the extreme westward retroflection events respectively. d. Mean ADT from 8 d before the event to 7 d after the event. e and f. Same as d, but for the mean ADT from the day of the event to 15 d afterwards, and from 16 d to 23 d after the event respectively. Red solid contours include areas of negative ADT value. Black dashed lines indicate the 0.7 m ADT value, which has been usually used to estimate the retroflection position.
Figure 12. Westward retroflection leads to warming in the South Atlantic (SA) and cooling in the Indian Ocean (OI), corresponding to the enhanced Agulhas leakage. Composited monthly SSTa differences between the westward and the eastward retroflection events from OISST dataset (a) and EAR5 dataset (b). c. Same as b, but with SSTa lag retroflection events for one month. The grey contours include the areas where the differences are significant at 95% confidence level of t-test. The black solid and dashed rectangular boxes denote the South Atlantic region and the Indian Ocean region, respectively.
Figure 13. Significant differences between the westward and eastward retroflection events in ALSST. (a) Lead-lag compositions of ERA5 ALSST for the westward retroflection events (red crosses) and for the eastward retroflection events (blue circles). The ALSST is computed from the difference between the SA (solid box in Fig. 12) mean SST anomalies and the IO (dashed box in Fig. 12) mean SST anomalies. (b) The ALSST differences between the westward and the eastward retroflection differences are statistically significant at 95% confidence level of t-test, whereas the values overlapped by the blue dashed bars are insignificant.
Figure 14.
Figure 15. Time evolution of Agulhas Ring shedding processes for the extreme early (eastward) retroflection during November 2000 to February 2001 recorded by Dencausse et al. (2010a). a. November 2000; b. December 2000; c. January 2001; d. February 2001.