Strong Dependence of Simulated Carbon-Nitrogen Dynamics on Numerical Coupling Between Nitrogen Uptake and Mineralization Processes
Simulating land biogeochemistry using earth system models (ESMs) relies on the robust numerical representation of a wide range of ecosystem processes. Here we use the Energy Exascale Earth System Model (E3SM) land model (ELMv1) to show that the numerical approach used to couple nitrogen uptake and mineralization processes strongly influences simulated land carbon dynamics and responses to environmental change (e.g., atmospheric CO2). In particular, we show that when nitrogen uptake and mineralization processes are concurrently coupled (rather than using the common asynchronous approach), land carbon dynamics become less sensitive to increasing atmospheric CO2. This study is the first of its kind demonstrating that more effort is needed in ESMs to address the numerical coupling of different land biogeochemical processes.
We demonstrated that: (1) the existing numerical treatment of nutrient limitations overestimates the modeled sensitivity of land carbon uptake to increasing atmospheric CO2; (2) uncertainties resulting from numerical coupling of nutrient uptake and mineralization are as large as those from climate forcing and model formulations; and (3) our proposed approach, the multi-substrate co-limiting algorithm, resolves this numerical uncertainty.
Land biogeochemistry includes many interconnected processes, which makes robust representation in ESMs challenging. However, very few analyses have quantified the influence on simulated land carbon dynamics from different numerical implementations. To explore this uncertainty, we took advantage of the modular biogeochemical transport and reaction module (BeTR) coupled with ELMv1 (ELMv1-BeTR) to show how three common numerical coupling schemes of nitrogen uptake and mineralization processes affect simulated land carbon dynamics under historical climate and increasing atmospheric CO2 over the 21st century. We found that the popular method used in many existing land models significantly overestimates responses to increasing atmospheric CO2. In contrast, our multi-substrate co-limiting algorithm is able to substantially reduce the numerical uncertainty. This study highlights the importance of numerical implementation on land model predictions and demonstrates how to improve land biogeochemical representations in ELM.