Abstract
Abstract. The commonly adopted biogeochemistry spin-up process in an Earth system model (ESM) is to run the model for hundreds to thousands of years subject to periodic atmospheric forcing to reach dynamic steady state of the carbon–nitrogen (CN) models. A variety of approaches have been proposed to reduce the computation time of the spin-up process. Significant improvement in computational efficiency has been made recently. However, a long simulation time is still required to reach the common convergence criteria of the coupled carbon–nitrogen model. A gradient projection method was proposed and used to further reduce the computation time after examining the trend of the dominant carbon pools. The Community Land Model version 4 (CLM4) with a carbon and nitrogen component was used in this study. From point-scale simulations, we found that the method can reduce the computation time by 20–69% compared to one of the fastest approaches in the literature. We also found that the cyclic stability of total carbon for some cases differs from that of the periodic atmospheric forcing, and some cases even showed instability. Close examination showed that one case has a carbon periodicity much longer than that of the atmospheric forcing due to the annual fire disturbance that is longer than half a year. The rest was caused by the instability of water table calculation in the hydrology model of CLM4. The instability issue is resolved after we replaced the hydrology scheme in CLM4 with a flow model for variably saturated porous media.
Highlights
The initial starting values of carbon–nitrogen (CN) models are not commonly available, especially for large-scale applications, but they have an important influence on the subsequent C–N states simulated by the models
The meteorological forcing, site information such as soil texture, vegetation cover, and satellite-derived phenology at each site are provided by the North American Carbon Program (NACP) site synthesis team for the sites located in North America and by the Large Scale BiosphereAtmosphere Experiment in Amazônia Model Intercomparison Project (LBA-MIP) for the sites located in South America
We described a gradient projection method to further speed up the spin-up process based on the slow nature of soil organic C decomposition
Summary
The initial starting values of carbon–nitrogen (CN) models are not commonly available, especially for large-scale applications, but they have an important influence on the subsequent C–N states simulated by the models. Initialization of the CN model is usually achieved by a spin-up run of the CN model given an arbitrary initial condition until an approximate C equilibrium is reached This time marching of the model requires several hundreds to thousands of years of model simulations before a dynamic steady state is reached. A number of approaches have been proposed in the past to improve upon the explicit forward time integration of ordinary differential equations in their native form and rate parameters for CN models These approaches include the initialization of soil organic matter carbon pools with observations (Zhang et al, 2002), the accelerated decomposition method using a higher decomposition rate for litter and soil carbon pools (Thornton and Rosenbloom, 2005), decelerated bulk denitrification and leaching method (Shi et al, 2013), and a semianalytical steady-state solution for soil organic C and N pools
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
