2018 ESA Annual Meeting (August 5 -- 10)

COS 130-8 - What drives the rhythm of tropical ecosystems? Carbon uptake seasonality in ecosystem models

Friday, August 10, 2018: 10:30 AM
338, New Orleans Ernest N. Morial Convention Center
Maria del Rosario Uribe, Ecological Sciences and Engineering and Forestry and Natural Resources, Purdue University, West Lafayette, IN, Carlos A. Sierra, Max-Planck-Institute for Biogeochemistry, Jena, Germany and Jeffrey S. Dukes, Purdue Climate Change Research Center, Purdue University, West Lafayette, IN
Background/Question/Methods

Tropical ecosystems play a disproportionate role in the global carbon cycle. However, seasonality of carbon uptake in the tropics is poorly captured by ecosystem models. Field- and satellite-based studies have identified variation in the strength of this seasonality and the different climate variables that drive it, both at the site and regional level. Yet, carbon uptake seasonality representation by models has only been analyzed at the site level, providing limited information on model performance at a larger scale. Here, we set out to determine how models represent carbon uptake seasonality in the tropics.

Cross-correlation function (CCF) was used to identify relationships between climate variables (i.e., precipitation and radiation) and gross primary productivity (GPP) simulations. Precipitation and radiation data were obtained from the TRMM (TMPA/3B43) and from the CERES-EBAF products, correspondingly. The models studied were the Community Land Model (CLM4.5), the Joint UK Land Environment Simulator (JULES) and the Lund-Potsdam-Jena General Ecosystem Simulator (LPJ-GUESS). We used time series of monthly data between 2000 and 2015 for each variable and for each pixel in the tropics (20°N - 20°S). Using the CCF, direct and lagged correlations of each climate variable with each model GPP time series were calculated for all pixels.

Results/Conclusions

Significant direct or lagged correlations between modeled GPP and a climate variable represent between 42 to 78% of all the land pixels in the tropics. The strength of the correlation differs between the two variables, as precipitation and GPP show greater and positive correlation coefficients. Radiation has both positive and negative correlations with GPP, but high correlation coefficients are less common than for precipitation. In terms of the relationship timing, precipitation and radiation precede the modeled GPP by up to three months in a large part of the tropics. These patterns in strength and timing in both climate variables are consistent across the three models.

Our results indicate that, in a large part of the tropics, modeled GPP has a stronger relationship with precipitation than radiation. Previous research has shown a large spatial variation in the response of carbon uptake seasonality to climate, with radiation being an important driver of productivity seasonality. Although models capture some of the regional variation, they poorly represent the important relationship between radiation and productivity. In order to improve the modeled GPP seasonality, future research should identify where there is disagreement between models and observations regarding the seasonal response to climate.