Fig. 1

Bottom topography of the South China Sea (shading), the annually-averaged velocity at 1000 m depth (red vector, units: cm•s-1) and the Argo trajectories (black lines). The release positions are indicated by green dots, and their end positions are, red rectangles"

Fig. 2

Schematics of one float cycle for (a) conventional Argo floats and (b) Deep Argo floats during the test phase analyzed for this study. A conventional Argo cycle includes descent, parking, deep profiling, and surface telemetry. The schematic indicates the times when descent starts (DS), descent ends (DE), deep descent starts (DDS), ascent starts (AS), ascent ends (AE), and transmission ends (TE) (the definitions are based on Ollitrault et al, 2013). A Deep Argo cycle contains similar phases but with a slight change in the sequence, since the float first descends from DS to the ocean bottom (DE), and then rises to a parking pressure (DAE) for several days before rising to the surface, starting from AS. Ascent accelerates (at AA) at approximately 2000 m depth, and the float reaches the surface at AE"

Tab. 1

Information of core Argo floats trajectories"

Fig. 3

The EOF mode-1 for ECCO2 50 m currents (a), TOPEX geostrophic currents (b) and Panel (c) denotes the PC-1 for ECCO2 (black curve) and TOPEX (gray curve)"

Fig. 4

The averaged velocity at mid-layer (1000 m) depth of the SCS: (a) winter and (b) summer. The release positions of synthetic floats are indicated by “+”. These location (Ⅰ: 16°12′N, 115°E; Ⅱ: 11°N, 110°30′E) are used in the following section"

Fig. 5

Lagrangian motions of 100 synthetic conventional Argo floats in 100-day model simulations, which are released at (a) Ⅰ and (b) Ⅱ (shown in Fig. 4). Gray and black Argo floats are advected by the winter and summer mean currents, respectively"

Fig. 6

PDFs of the simulated conventional Argo floats at the end of 100 days. Panels (a~f) correspond to the float series: 2902691, 2902692, 29026993, 2902694, 2902695, 2902697, and 2902698, which are listed in Table 1. The starting positions of the real conventional Argo floats\releasing positions of synthetic floats are marked by green dots, and the end points are indicated by red squares"

Fig. 7

The Lagrangian trajectories for a simulated deep Argos array. The resolution of the array is 2°×2°×30 day. The releasing positions of synthetic floats are marked by red dots. The trajectories are marked with colored dotted lines indicating the running time and overlapped by the 2000 m isobath that is marked by gray line"

Fig. 8

The mean amplified random walk scales of synthetic floats as a function of run duration. The scale is calculated by the mean distance between each synthetic float and the center of mass of the ensemble of floats. The results of conventional Argo simulation are shown in black solid curve, and deep Argo simulation results are shown in dashed curve with the original random walk curve (without the ocean currents) indicated in gray"