Contrasting Decarbonation Patterns in Ultra-Cold Tonga and Ultra-Hot Colombia Subduction Zones: Insights from Thermal-Petrological Modeling, Gas Emissions, and Geochemical Observations

The decarbonation process of subducting slab plays a key role in the carbon cycle of subduction zones. This process is influenced by several factors, including thermal structure, bulk-rock composition and thickness of lithologic layers. However, the mechanisms, flux and efficiency of subduction-related decarbonation remain widely debated. To better quantify the carbon release within subducting slab, two end-member subduction zones are selected in this study: Tonga and Colombia, which represent ultra-cold and ultra-hot oceanic subduction zone, respectively. Based on the numerical simulations using a newly developed coupled thermomechanical-metamorphism-dissolution decarbonation model in open system, combined with observational data on gas emissions and geochemical signatures from volcanic arcs, we systematically investigate the slab decarbonation processes in Tonga and Colombia subduction zones. This study reveals contrasting decarbonation patterns between Tonga and Colombia, highlighting the influence of thermal structure and bulk-rock composition on slab decarbonation.

Model results indicate that, thermal structure is first-order control on decarbonation. The cumulative decarbonation flux of hot Colombia is three times higher than that of cold Tonga. The overall decarbonation efficiency shows a greater contrast, with values of 1.3 % for Tonga and 9.7 % for Colombia. The dominant decarbonation mechanism also differs: carbonate dissolution dominates in Tonga, while metamorphic decarbonation prevails in Colombia. Additionally, the bulk-rock composition of lithologic layers also plays roles. In Tonga, CO₂-poor sediments and adequate surface slab temperature facilitate efficient carbon release, whereas in Colombia, sediments are highly enriched in carbon and calcium, hindering metamorphic decarbonation, and only minor carbon is released via carbonate dissolution.

We further collected annual gas emission data to constrain the CO₂ outflux from arc volcanic degassing. Volcanic CO₂ outflux from Tonga arc is significantly lower than that of Colombia arc, consistent with the trend predicted by model. However, in Colombia, the predicted total carbon outflux released from slab is lower than the observed volcanic CO₂ outflux. Besides, compiled geochemical data from overlying arc rocks reveals obvious signals of sediment melt contributions to mantle sources in Colombia. Based on results of experimental petrological studies, we propose that a portion of sediments in Colombia forms diapirs that ascend into the mantle wedge, generating melting-induced decarbonation. This could account for the discrepancy between the model predictions and observed volcanic CO₂ outflux in Colombia.