Subsurface biogeochemical response to Hurricane Idalia within a cyclonic eddy and river plume–stratified environment

See manuscript for more details. Below are sections that are relevant for the data from the manuscript. 

2. Data and Methods 

2.1 Hurricane Idalia data

Observations of Hurricane Idalia’s track, intensity, and structure were obtained from the International Best Track Archive for Climate Stewardship (IBTrACS) (Gahtan et al., 2024; Knapp et al., 2010), based on the National Hurricane Center’s best track data (Cangialosi & Alaka, 2024). In this archive, the best-known center coordinates (within 0.1˚ latitude/longitude), maximum sustained wind speed (2.6 m/s increments), minimum sea level pressure (within 1 hPa), and 17 m/s (34 kt), 26 m/s (50 kt), and 33 m/s (64 kt) wind radii (within 10 nautical miles) for each quadrant (NE, SE, SW, NW) were input for 6-hour intervals. 

To determine the location of the autonomous platforms (i.e., BGC-Argo float and saildrone) within Hurricane Idalia’s wind field, we interpolated the IBTrACS original 6-hourly data to a 1-minute resolution using Modified Akima piecewise cubic Hermite interpolation. To determine the horizontal extent of the wind field, at the time the interpolated intensity exceeded a wind speed radius threshold (17 m/s, 26 m/s, and 33 m/s), the radius for each quadrant (NE, SE, SW and NW) was interpolated to 1-minute resolution. We then determined the region within each radius using a time-interpolated 1-degree radial resolution. 

Using the new spatiotemporally interpolated extent of Hurricane Idalia’s wind field, we determine when each autonomous platform was before, during, and after the passage of the 17 m/s wind field. These labels correspond to the blue (before Idalia), green (within Idalia), and magenta (after Idalia) colors in Figure 1. 

2.2 Satellite data and analytical methods

2.2.1 Satellite Data

Daily fields of the National Oceanic and Atmospheric Administration (NOAA) CoastWatch gap-filled ocean color chlorophyll data were obtained at 9 km daily resolution. This product uses a Data INterpolating Empirical Orthogonal Function (DINEOF) algorithm to combine Multi-Sensor Level 1 and Level 2 (MSL12) ocean color data from multiple Visible Infrared Imaging Radiometer Suite (VIIRS) sensors (Suomi National Polar-Orbiting Partnership (SNPP) and NOAA-20), to create a gap-filled analysis, which allows for a cloud-free view of ocean color derived chlorophyll (X. Liu & Wang, 2019). Sea level anomalies (SLA) and geostrophic currents from the NOAA Radar Altimeter Database System (RADS) Level 4 merged near-real time (NRT) product were used and are available on a 0.25° daily grid from 2017 (2019 for currents) through the present from NOAA CoastWatch (Scharroo et al., 2013). Blended 6-hourly wind stress and surface wind measurements from NOAA National Centers for Environmental Information (NCEI) Blended Seawinds version 2 product (NBSv2) were obtained from NOAA CoastWatch on a 0.25° grid. NBSv2 combines satellite observations from multiple scatterometers (up to 7 since 2002) with L-band and the AMSR2 all-weather channel observations (Saha & Zhang, 2022). NOAA’s Geo-polar night SSTs were obtained on a daily 9 km grid from NOAA CoastWatch (Maturi et al., 2017). Additional satellite-derived SSS from the Soil Moisture Active Passive (SMAP) mission processed by NASA’s Jet Propulsion Lab as the Combined Active Passive (CAP) version 5.0 product were obtained from PO.DAAC on 0.25° daily grids as an 8-day interpolated product (Fore et al., 2016). 

BGC-Argo float positions were collocated with mesoscale eddies identified using the MUltiparameter NRT System for Tracking Eddies Retroactively (MUNSTER) product suite from NOAA CoastWatch (McWhorter et al., 2024; Roman‐Stork et al., 2023). MUNSTER is a threshold-free, closed-contour eddy tracking method adapted from algorithms originally developed by Chaigneau et al., (2008, 2009) and Pegliasco et al., (2015). An asymmetric Gaussian high-pass spatial filter with a 5°/10° latitude/longitude half-width was applied to the daily NOAA RADS SLA field to remove planetary wave contamination. Eddies in MUNSTER are identified based on closed contours surrounding local maxima (anticyclonic eddies) and minima (cyclonic eddies) of filtered daily NOAA RADS SLA at a 0.1 cm contour interval. For the purposes of this study, eddy contours from MUNSTER were used to collocate the BGC-Argo float profiles and identify when the float surfaced within or outside of an eddy. 

2.3 Saildrone data

Saildrones are equipped with oceanographic and meteorological sensors to measure near-surface wind velocity, air temperature, relative humidity, barometric pressure, solar radiation, SST, SSS, dissolved oxygen concentration, chlorophyll concentration, wave height and period, and profiles of ocean currents over depths from ~ 6 to ~80 m with 2 m resolution (Supplementary Information Table 1; Zhang et al., 2023). These sensors are validated before the mission to ensure data quality control. Wind is measured at a height of ~3.45 m and air temperature and humidity are measured at ~2.3 m. It should be noted that the exact height per measurement may vary based on the vehicle’s pitch and roll. However, we found that this was minimal with the mean 1-minute height of the wind measurements in Idalia being 3.36 ± 0.03 m. All measurements reported here have not been adjusted to the standard 10 m height, which would increase wind speed by about 10 - 15%, assuming neutral atmospheric stability. SST, SSS, dissolved oxygen, and chlorophyll are measured at depths of ~1.7 m. Saildrones are operated remotely through satellite communications, powered by solar radiation, and propelled by the wind. Through a partnership between NOAA and Saildrone Inc. from 2021 to 2024, saildrone uncrewed surface vehicles have been used to observe hurricanes and transmit their 1-minute measurements in near-real time, with higher resolution data (1 to 20 Hz) recorded and downloaded upon retrieval at the end of the mission. During the 2023 mission, saildrone 1083 (hereafter ‘SD-1083’) was directed to, and intercepted, Hurricane Idalia in the eastern Gulf. 

Prior to being directed for intercept on August 24, SD-1083 was positioned 25 - 50 km west of Hurricane Idalia’s eventual track. SD-1083 moved ~45 km to the SE from August 27 - 28 working best with the prevalent winds and currents to better position for intercept. On August 29, SD-1083 traveled to the NE for its final intercept positioning. SD-1083 started measuring tropical-storm-force 1-minute sustained winds on August 29 at 14:32 UTC, with sustained winds peaking at 36 m/s and maximum gusts of 43 m/s nearly 8 hours later. SD-1083 then entered the northeastern eye of Idalia around August 29 at 22:24 UTC and remained in the eye until 22:59 UTC, when it exited through the southeastern eyewall. The surface pressure reached a minimum of 964.4 hPa and the significant wave height peaked at 9.6 m. SD-1083 measured its final tropical-storm-force sustained wind from Idalia over 14 hours after tropical-storm-force winds began. The track of SD-1083 is shown in Figure 1.

Following the intercept of Hurricane Idalia, SD-1083 drifted ~60 km to the NE within one day and remained over the next four days. Then, SD-1083 traveled back and remained within 20 km of the intercept location for the following 8 days, reaching a minimum distance of less than 150 m on September 9. On September 11, 2023 12:00 UTC, more than 12.5 days after the intercept, SD-1083 began to travel offshore of its intercept location. 

Pre-storm measurements from SD-1083 were defined as the average over a set period of time (i.e., 6 hours, 12 hours, 24 hours, 3 days, 5 days, and 10 days) ending 6-hours prior to SD-1083 entering into the region of 17 m/s winds. Multiple periods were used to define the uncertainty. A similar analysis was done to define post-storm SST, with averaging periods of 6 hours, 12 hours, and 24 hours starting 6 hours following SD-1083 exiting the region of 17 m/s winds. 

Following Chiodi et al. (2024) and Brenner et al. (2023), in the absence of vertical measurements of temperature and salinity, which prevent the estimate of the mixed layer depth based on a density criterion, the mixed layer depth at the saildrone’s location (MLD_saildrone) was derived from the vertical shear of the horizontal currents measured by the saildrone’s downward-looking 300 kHz ADCP. ADCP observations where the total percentage of good pings per ensemble of less than 50% were removed, which on average, removed the lowest 10 m, where noise is high from the ADCP’s backscatter signal. Subsequent results are not sensitive to the threshold of good pings used. The magnitude of the vertical shear was calculated based on the vector differences of the horizontal components (i.e., u, v) of the current for each time step. The depths that met the following three criteria at each time step were identified: (1) the shear had a value within half a standard deviation of the maximum value of shear across depths, (2) the shear had a value greater than one standard deviation above the mean shear across depths, and (3) the shear was not the deepest nor shallowest depth.  Less than 4% of the total time steps failed to have depths that met these criteria. Following this, if the standard deviation of the identified depths at a given time step was 10 m or greater (occurring less than 9% of the time), then that time step was deemed invalid. For all remaining valid time steps, the identified depths were then averaged to determine the initial MLD_saildrone. Since each time step is treated independently, it is possible that some unrealistic rapid changes in the initial MLD_saildrone exist. In order to remove these, all times when each depth was greater than one standard deviation away from the mean based on a 6-hour moving window were deemed invalid (17% of the time). The MLD_saildrone used was then calculated from a 6-hourly moving average of the remaining times (over 70% of the total).