Spatial biases of information influence global estimates of soil respiration: How can we improve global predictions?

Abstract

Soil respiration (Rs), the efflux of CO2 from soils to the atmosphere, is a major component of the terrestrial carbon cycle, but is poorly constrained from regional to global scales. The global soil respiration database (SRDB) is a compilation of in situ Rs observations from around the globe that has been consistently updated with new measurements over the past decade. It is unclear whether the addition of data to new versions has produced better‐constrained global Rs estimates. We compared two versions of the SRDB (v3.0 n = 5173 and v5.0 n = 10,366) to determine how additional data influenced global Rs annual sum, spatial patterns and associated uncertainty (1 km spatial resolution) using a machine learning approach. A quantile regression forest model parameterized using SRDBv3 yielded a global Rs sum of 88.6 Pg C year−1, and associated uncertainty of 29.9 (mean absolute error) and 57.9 (standard deviation) Pg C year−1, whereas parameterization using SRDBv5 yielded 96.5 Pg C year−1 and associated uncertainty of 30.2 (mean average error) and 73.4 (standard deviation) Pg C year−1. Empirically estimated global heterotrophic respiration (Rh) from v3 and v5 were 49.9–50.2 (mean 50.1) and 53.3–53.5 (mean 53.4) Pg C year−1, respectively. SRDBv5’s inclusion of new data from underrepresented regions (e.g., Asia, Africa, South America) resulted in overall higher model uncertainty. The largest differences between models parameterized with different SRDVB versions were in arid/semi‐arid regions. The SRDBv5 is still biased toward northern latitudes and temperate zones, so we tested an optimized global distribution of Rs measurements, which resulted in a global sum of 96.4 ± 21.4 Pg C year−1 with an overall lower model uncertainty. These results support current global estimates of Rs but highlight spatial biases that influence model parameterization and interpretation and provide insights for design of environmental networks to improve global‐scale Rs estimates.

A Data‐Driven Global Soil Heterotrophic Respiration Dataset and the Drivers of Its Inter‐Annual Variability

“…This uncertainty measure has also been used in J. Zeng et al (2020) for net ecosystem productivity and Warner et al (2019) as well as Stell et al (2021) for soil respiration. Through the comparison of two sources of uncertainties (alternative gridded explanatory data and uncertainty from extrapolation to under-sampled domain by spread of trees), we find that uncertainty from individual trees in Random Forest model is far larger than that from different explanatory variables datasets.…”

The cost‐effectiveness of reflectance spectroscopy for estimating soil organic carbon

“…They have wanted to understand the role of carbon (C) in the soil's behaviour, especially as the result of changes under cultivation. In recent years emphasis has switched towards the carbon cycle in ecosystem processes (Davidson & Janssens, 2006;Fontaine et al, 2007;Schmidt et al, 2011;Stell et al, 2021) and for carbon accounting (Atwood et al, 2017;Mishra et al, 2021;Viscarra Rossel et al, 2014). Scientists have taken soil material from the field into the laboratory, dried it, crushed and sieved it, and then analysed sub-samples of it chemically.…”

Ideas and perspectives: Enhancing research and monitoring of carbon pools and land-to-atmosphere greenhouse gases exchange in developing countries

“…2). In terms of a continental scale, measurements in Europe, North America, and Asia cover around 90 % of the global observations, while Africa and South America remain critically underrepresented (Stell et al, 2021;Jian et al, 2021;Gatica et al, 2020;Épule, 2015;Kim et al, 2013) compared to their importance in global GHG budgets (Fig. 3).…”

“…Research on C and land-to-atmosphere GHG exchange is thus critical to understand the consequences of rapidly increasing atmospheric GHG concentrations. This research should be carried out globally, in both developed and developing countries, since both have different sources and sinks of GHGs, different climate change vulnerabilities, and different capacities for mitigation and adaptation (Stell et al, 2021;López-Ballesteros et al, 2018;Ogle et al, 2014). Traditionally, this has required high-quality long-term or vastspatial-scale (e.g., regional or continental) data collected using advanced instruments, significant computing power with complex and/or proprietary software, and skilled technicians -all expensive to develop, implement, and maintain.…”