A sliding-window-threshold algorithm for identifying global mesoscale ocean fronts from satellite observations

Global mapping and evolution of persistent fronts in Large Marine Ecosystems over the past 40 years

“…Persistent fronts exhibit low intensity, reaching only 1.10 °C/100 km within tropical LMEs, potentially attributed to spatially uniform heating from strong shortwave radiation 37 . The highest occurrence and intensity of persistent fronts are observed in subtropical LMEs, exceeding 4% occurrence and 2 °C/100 km intensity, primarily associated with the presence of many famous boundary currents and upwelling systems 28 . We selected typical LMEs around boundary currents and upwelling systems and found that persistent fronts exhibit high occurrence and intensity, with an average occurrence of 4.43% and 4.16%, as well as an intensity of 2.20 °C/100 km and 1.92 °C/100 km, respectively (Fig.…”

“…The hit rate was utilized as a statistical metric to assess the precision of persistent fronts in both reanalysis and climate model datasets within LMEs. It is defined as the percentage of satellite-identified persistent frontal pixels that fall within a 3-pixel radius of those identified in the reanalysis or climate model datasets 28 , 29 .…”

“…Frontal occurrence is determined by counting the number of frontal pixels within a 6-pixel radius (~30 km at the equator) of persistent fronts during their respective occurrence seasons and dividing this by the total number of available SST pixels within the same distance during those seasons. Frontal intensity is derived from the SST gradient field in persistent fronts during their occurrence seasons, employing a modified Sobel operator tailored to filter out submesoscale information of <50 km 21 , 28 . The frontal intensity on the location in the input SST image is intended to eliminate information at spatial scales of , and is formulated as follows: where, represents the spatial resolution (~5 km in this study) of the input SST image, while denote the latitudinal and longitudinal gradient vector components on the location but at spatial scales on the order of .…”

“…Satellite observations represent the most effective tool for acquiring information on global frontal activities and monitoring their long-term variations. Previous research has provided a preliminary global perspective on fronts using an automatic front detection algorithm and satellite-observed sea surface temperature (SST) images 27 , 28 . What is most remarkable is that numerous persistent fronts have been found within Large Marine Ecosystems (LMEs), and these fronts have further been manually delineated based on global frontal occurrence probability and named according to their geography 13 .…”