Adaptive Sampling of the Upper Ocean by Autonomous Floats During Atmospheric River Precipitation

March 5, 2026

A paper titled “Adaptive Sampling of the Upper Ocean by Autonomous Floats During Atmospheric River Precipitation” was recently published in Geophysical Research Letters. This study was led by Donata Giglio (University of Colorado, Boulder), with contributions from Jacopo Sala (University of Colorado, Boulder), John Gilson (Scripps Institution of Oceanography, University of California, San Diego), Lauren Hoffman (Scripps Institution of Oceanography, University of California, San Diego; Université catholique de Louvain), Brian Kawzenuk (Center for Western Weather and Water Extremes, University of California, San Diego), Bill Katie-Anne Mills (University of Colorado, Boulder), Sarah G. Purkey (Scripps Institution of Oceanography, University of California, San Diego), Megan Scanderbeg (Scripps Institution of Oceanography, University of California, San Diego), Aneesh C. Subramanian (University of Colorado, Boulder), Anna M. Wilson (Center for Western Weather and Water Extremes, University of California, San Diego), F. Martin Ralph (Center for Western Weather and Water Extremes, University of California, San Diego). The work supports the Novel Observations Priority identified in CW3E’s Strategic Plan by developing a new adaptive sampling technique for a critical part of the global observing system that can be used during atmospheric rivers.

This study presents the first targeted use of autonomous ocean floats to adaptively sample the upper ocean response to Atmospheric River (AR) precipitation events. This adaptive sampling was conducted in coordination with the Argo program and the Atmospheric River Reconnaissance Program. Normally, profiles collected by the floats (as part of their mission) are too sparse in space and time to capture rapid changes in upper‑ocean properties driven by rainfall. To overcome this limitation, the authors developed an adaptive sampling strategy that evolved with the precipitation forecast and enabled high‑resolution observations of temperature, salinity, and stratification changes in the northeast Pacific during AR-related rainfall (Fig. 1a,b).

Observations reveal that ARs can temporarily cool and freshen the upper ocean, with stronger effects when winds at the surface are weak (Fig. 1c,d, Fig. 2). During winter months, ARs supply most of the region’s precipitation and are the primary driver of ocean freshening within the top 25–30 meters of the water column (Fig. 3). Ocean observations during AR events help improve our understanding of air‑sea interactions during rainfall and offer a pathway toward a better representation of coupled processes in numerical models, with potential implications for the forecasting of Atmospheric Rivers, which is essential for effective water management and flood risk reduction along the coast.

Figure 1. (a, b) Integrated vapor transport, (c, d) Global Precipitation Measurement precipitation (gray, right y‐axis) and ERA5 surface wind speed (red, left y‐axis) at the location of floats used in this experiment (blue stars in panels (a, b)) to measure salinity changes during atmospheric river precipitation. In panels (a, b), gray contours are shown every 250 kg/m/s. In panels(c, d), the light‐gray shading shows the 10th–90th percentile range across the 1,000‐member precipitation ensemble generated from IMERG random errors, and the solid dark‐gray line shows the ensemble mean. Figure 2 from Giglio et al. (2025.)

Figure 2. Salinity in the top ≈25 m of the ocean (blue to red lines, left y‐axis), during atmospheric river precipitation events in (a) January and (b) February 2022. Salinity is based on observations from profiling Argo floats operating with an adaptive sampling strategy (solid lines, with error bars from the Argo files; Wong et al., 2020) and a General Ocean Turbulence Model simulation (dotted lines for the ensemble mean and shading for the 10th–90th percentile range across precipitation ensemble members in Figures 1c and 1d). The gray line is the ensemble mean Global Precipitation Measurement precipitation (righty‐axis), as in Figures 1c and 1d. Figure 3 from Giglio et al. (2025.)

Figure 3. (a) Fraction of DJF precipitation by atmospheric rivers (ARs), during 2005–2021, based on MERRA‐2 precipitation and the Rutz AR climatology. Black lines highlight every other contour. (b, c) March‐minus‐November near‐surface salinity difference (in color), based on (b) the Argo climatology by Roemmich and Gilson (2.5 dbar level), during 2005–2021 and(c) ECCOv4r4 (at 5 m), during 1993–2016. In panel (b), black contours are as in panel (a). In panel (c), the black box encloses the largest decrease in salinity between November and March. (d) ECCOv4r4 salinity budget within the box in panel (c):(black line) tendency, (red) forcing, (blue) advective convergence, (cyan) diffusive convergence, and (gray) residual, averaged during DJF and integrated from the shallowest model level to each of the depths shown (e.g., the deepest point represents the integral between 5 and ≈140 m). Figure 1 from Giglio et al. (2025.)

Citation:

Giglio, D., Sala, J., Gilson, J., Hoffman, L., Kawzenuk, B., Katie-Anne Mills, B., Purkey, S. G., Scanderbeg, M., Subramanian, A. C., Wilson, A. M., & Ralph, F. M. (2025). Adaptive sampling of the upper ocean by autonomous floats during atmospheric river precipitation. Geophysical Research Letters, 52(23), e2025GL117069. https://doi.org/10.1029/2025GL117069