Spatio-temporal variability of surface water pCO₂ and nutrients in the tropical Pacific from 1981 to 2015

This study presents a synthesis of surface water partial pressure of CO₂ (pCO₂) and nutrient observations in the tropical Pacific (20°S–20°N) from 1981 to 2015 and characterizes the spatio-temporal variability. We used data from the Surface Ocean CO₂ Atlas version 5, which include about 2 million pCO₂ measurements in the tropical Pacific. Unlike many previous studies that estimated pCO₂ fields from relationships with other variables like sea surface temperature, we developed gridded products of monthly means using a technique to interpolate measured pCO₂ values. Large seasonal variation of pCO₂ appears in the cold tongue region (EQ–20°S, east of 120°W) corresponding to the seasonal variation of coastal upwelling, and in the off-equatorial region where the thermodynamic effect on pCO₂ dominates. Consistent with previous findings, pCO₂ along the equator declines during El Niño due to weakening of the easterly winds and therefore reduced upwelling of CO₂ rich subsurface water. We also quantified the spatial distribution of the long-term pCO₂ trend beyond the area-averaged trend presented previously. The long-term trend of pCO₂ is positive in all regions with an area average of 1.8 ± 0.1 μatm yr⁻¹. However, along the equator the trend is > 2 μatm yr⁻¹ linked to the Pacific Decadal Oscillation forcing. Using the same methodology, we also analyzed about 20,000 surface nutrient measurements from the Global Ocean Data Analysis Project version 2, World Ocean Database 2013, and ship-of-opportunity sampling by the National Institute for Environmental Studies, Japan. We present the spatial patterns of reduced surface nutrient concentration in the central to eastern tropics along the equator during El Niño, but there are not enough data to characterize the trends of nutrients in the tropical Pacific. Decorrelation analyses are also applied using covariance between pCO₂ anomalies separated by a lag increment for zonal, meridional, and temporal directions. The e-folding scale of pCO₂ is estimated to be 6° in latitude, 13° in longitude, and 2 months with a signal-to-noise ratio of 4, which are used as input for the interpolation as well as an assessment of ideal observation density and frequency. Decorrelation analyses such as this are critical for evaluating existing observing system design and informing future sampling strategies.