Abstract. This paper presents an updated estimation of the bottom-up global surface seawater dimethyl sulfide (DMS) climatology. This update, called DMS-Rev3, is the third of its kind and includes five significant changes from the last climatology, L11 (Lana et al., 2011), that was released about a decade ago. The first change is the inclusion of new observations that have become available over the last decade, creating a database of 873 539 observations leading to an ∼ 18-fold increase in raw data as compared to the last estimation. The second is significant improvements in data handling, processing, and filtering, to avoid biases due to different observation frequencies which result from different measurement techniques. Thirdly, we incorporate the dynamic seasonal changes observed in the geographic boundaries of the ocean biogeochemical provinces. The fourth change involves the refinement of the interpolation algorithm used to fill in the missing data. Lastly, an upgraded smoothing algorithm based on observed DMS variability length scales (VLS) helps to reproduce a more realistic distribution of the DMS concentration data. The results show that DMS-Rev3 estimates the global annual mean DMS concentration to be ∼ 2.26 nM (2.39 nM without a sea-ice mask), i.e., about 4 % lower than the previous bottom-up L11 climatology. However, significant regional differences of more than 100 % as compared to L11 are observed. The global sea-to-air flux of DMS is estimated at ∼ 27.1 TgS yr−1, which is about 4 % lower than L11, although, like the DMS distribution, large regional differences were observed. The largest changes are observed in high concentration regions such as the polar oceans, although oceanic regions that were under-sampled in the past also show large differences between revisions of the climatology. Finally, DMS-Rev3 reduces the previously observed patchiness in high productivity regions. The new climatology, along with the algorithm, can be found in the online repository: https://doi.org/10.17632/hyn62spny2.1 (Mahajan, 2021).

Abstract. This paper presents an updated estimation of the bottom-up global surface seawater dimethyl sulfide (DMS) climatology. This update, called DMS-Rev3, is the third of its kind and includes five significant changes from the last climatology, ‘L11’ (Lana et al., 2011) that was released about a decade ago. The first change is the inclusion of new observations that have become available over the last decade, creating a database of 872,427 observations leading to a ~18-fold increase in raw data as compared to the last estimation The second is significant improvements in data handling, processing, and filtering, to avoid biases due to different observation frequencies which results from different measurement techniques. Thirdly, we incorporate the dynamic seasonal changes observed in the geographic boundaries of the ocean biogeochemical provinces. The fourth change involves the refinement of the interpolation algorithm used to fill in the missing data. And finally, an upgraded smoothing algorithm based on observed DMS variability length scales (VLS) helps to reproduce a more realistic distribution of the DMS concentration data. The results show that DMS-Rev3 estimates the global annual mean DMS concentration to be ~1.87 nM (2.35 nM without a sea-ice mask), i.e., about 4 % lower than the previous bottom-up ‘L11’ climatology. However, significant regional differences of more than 100 % as compared to L11 are observed. The global sea to air flux of DMS is estimated at ~27 TgS yr−1 which is about 4 % lower than L11, although, like the DMS distribution, large regional differences were observed. The largest changes are observed in high concentration regions such as the polar oceans, although oceanic regions that were under-sampled in the past also show large differences between revisions of the climatology. Finally, DMS-Rev3 reduces the previously observed patchiness in high productivity regions.

Abstract. Dimethylsulfide (DMS) emitted from the ocean makes a significant global contribution to natural marine aerosol and cloud condensation nuclei, and therefore our planet’s climate. Oceanic DMS concentrations show large spatiotemporal variability, but observations are sparse, so products describing global DMS distribution rely on interpolation or modelling. Understanding the mechanisms driving DMS variability, especially at local scales, is required to reduce uncertainty in large scale DMS estimates. We present a study of mesoscale and sub-mesoscale (<100 km) seawater DMS variability that takes advantage of the recent expansion in high frequency seawater DMS observations and uses all available data to investigate the typical distances over which DMS varies in all major ocean basins. These DMS spatial variability lengthscales (VLS) are uncorrelated with DMS concentrations. DMS concentrations and VLS can therefore be used separately to help identify mechanisms underpinning DMS variability. When data are grouped by sampling campaigns, almost 80 % of the DMS VLS can be explained using the VLS of sea surface height anomalies, density, and chlorophyll-a. Our global analysis suggests that both physical and biogeochemical processes play an equally important role in controlling DMS variability, in contrast with previous results based on data from the low–mid latitudes. The explanatory power of sea surface height anomalies indicates the importance of mesoscale eddies in driving DMS variability, previously unrecognised at a global scale and in agreement with recent regional studies. DMS VLS differs regionally, including surprisingly high frequency variability in low latitude waters. Our results independently confirm that relationships used in the literature to parameterise DMS at large scales appear to be considering the right variables. However, contrasts in regional DMS VLS highlight that important driving mechanisms remain elusive. The role of sub-mesoscale features should be resolved or accounted for in DMS process models and parameterisations. Future attempts to map DMS distributions should consider the length scale of variability.

No abstract

Abstract. Dimethylsulfide (DMS) emitted from the ocean makes a significant global contribution to natural marine aerosol and cloud condensation nuclei and, therefore, our planet's climate. Oceanic DMS concentrations show large spatiotemporal variability, but observations are sparse, so products describing global DMS distribution rely on interpolation or modelling. Understanding the mechanisms driving DMS variability, especially at local scales, is required to reduce uncertainty in large-scale DMS estimates. We present a study of mesoscale and submesoscale (< 100 km) seawater DMS variability that takes advantage of the recent expansion in high-frequency seawater DMS observations and uses all available data to investigate the typical distances over which DMS varies in all major ocean basins. These DMS spatial variability length scales (VLSs) are uncorrelated with DMS concentrations. The DMS concentrations and VLSs can therefore be used separately to help identify mechanisms underpinning DMS variability. When data are grouped by sampling campaigns, almost 80 % of the DMS VLS can be explained using the VLSs of sea surface height anomalies, density, and chlorophyll a. Our global analysis suggests that both physical and biogeochemical processes play an equally important role in controlling DMS variability, which is in contrast with previous results based on data from the low to mid-latitudes. The explanatory power of sea surface height anomalies indicates the importance of mesoscale eddies in driving DMS variability, previously unrecognised at a global scale and in agreement with recent regional studies. DMS VLS differs regionally, including surprisingly high-frequency variability in low-latitude waters. Our results independently confirm that relationships used in the literature to parameterise DMS at large scales appear to be considering the right variables. However, regional DMS VLS contrasts highlight that important driving mechanisms remain elusive. The role of submesoscale features should be resolved or accounted for in DMS process models and parameterisations. Future attempts to map DMS distributions should consider the length scale of variability.