Constraining Uncertainty in Projected Gross Primary Production With Machine Learning

Abstract

The terrestrial biosphere is currently slowing down global warming by absorbing about 30% of human emissions of carbon dioxide (CO 2). The largest flux of the terrestrial carbon uptake is gross primary production (GPP) defined as the production of carbohydrates by photosynthesis. Elevated atmospheric CO 2 concentration is expected to increase GPP ("CO 2 fertilization effect"). However, Earth system models (ESMs) exhibit a large range in simulated GPP projections. In this study, we combine an existing emergent constraint on CO 2 fertilization with a machine learning approach to constrain the spatial variations of multimodel GPP projections. In a first step, we use observed changes in the CO 2 seasonal cycle at Cape Kumukahi to constrain the global mean GPP at the end of the 21st century (2091-2100) in Representative Concentration Pathway 8.5 simulations with ESMs participating in the Coupled Model Intercomparison Project Phase 5 (CMIP5) to 171 ± 12 Gt C yr −1 , compared to the unconstrained model range of 156-247 Gt C yr −1. In a second step, we use a machine learning model to constrain gridded future absolute GPP and gridded fractional GPP change in two independent approaches. For this, observational data are fed into the machine learning algorithm that has been trained on CMIP5 data to learn relationships between present-day physically relevant diagnostics and the target variable. In a leave-one-model-out cross-validation approach, the machine learning model shows superior performance to the CMIP5 ensemble mean. Our approach predicts an increased GPP change in northern high latitudes compared to regions closer to the equator. Plain Language Summary About a quarter of human emissions of carbon dioxide (CO 2) is absorbed by vegetation and soil on the Earth's surface and hence does not contribute to global warming caused by CO 2 in the atmosphere. Thus, in order to better define the remaining carbon budgets left to meet particular warming targets like the 1.5°C of the Paris Agreement, it is important to accurately quantify the carbon uptake by plants in the future. Currently, this is modeled by Earth system models yet with great uncertainties. In this work, we present an alternative machine learning approach to reduce spatial uncertainties of vegetation carbon uptake in future climate projections using observations of today's conditions.

Machine Learning‐Based Modeling of Vegetation Leaf Area Index and Gross Primary Productivity Across North America and Comparison With a Process‐Based Model

“…For example, Yang et al (2007) trained SVM to predict vegetation GPP using explanatory remote sensing variables, such as land surface temperature, enhanced vegetation index (EVI), land cover, and ground-measured climate variables. Schlund et al (2020) proposed a new two-step approach that apply an existing emergent constraint on CO 2 fertilization in combination with a supervised machine learning model to constrain uncertainties in multimodel predictions of GPP. The machine learning models once developed and trained have fast computing capabilities, and can be conveniently applied to studies on the continental or global scale.…”

A Review of Machine Learning Applications in Land Surface Modeling

“…An RT model utilizes a decision tree as a predictive model, and target variables are real values. RTs can be further advanced by gradient boosting (GBRT) [58,59]. 'Bagging' is a machine learning ensemble technique.…”

“…Schlund et al [58] developed a two-step approach to constrain the projected GPP at the end of the 21st century in the Representative Concentration Pathway 8.5 scenario with ML. They fed observational data into the ML algorithms that have been trained on Coupled Model Intercomparison Project (CMIP5) data to learn the relationships between present-day carbon estimates and the future scenario GPP (target variables).…”

How the Updated Earth System Models Project Terrestrial Gross Primary Productivity in China under 1.5 and 2 °C Global Warming

“…They found that total NPP in China is projected to increase continuously under different scenarios, with CO 2 concentration playing the dominant role. Using a machine learning model to constrain the spatial uncertainty in GPP projections, Schlund et al [5] predicted a higher increase in GPP in northern high latitudes over the 21st century under the Representative Concentration Pathway [6] 8.5 W m −2 (RCP8.5) in comparison with regions closer to the equator. Under 1.5 • C of global warming, the GPP in China is expected to increase by 15.5% ± 5.4% on a stabilized pathway and by 11.9% ± 4.4% on a transient pathway [7].…”

“…In summary, GPP/NPP in China under different scenarios is expected to show a trend of increase in the 21st century. However, large uncertainties exist in the various ESMs [5,9]. Under the global warming targets of 1.5 and 2 • C above preindustrial levels set by the Paris Agreement, many regional impacts wait to be assessed.…”