A physiology‐based Earth observation model indicates stagnation in the global gross primary production during recent decades

Abstract

Earth observation‐based estimates of global gross primary production (GPP) are essential for understanding the response of the terrestrial biosphere to climatic change and other anthropogenic forcing. In this study, we attempt an ecosystem‐level physiological approach of estimating GPP using an asymptotic light response function (LRF) between GPP and incoming photosynthetically active radiation (PAR) that better represents the response observed at high spatiotemporal resolutions than the conventional light use efficiency approach. Modelled GPP is thereafter constrained with meteorological and hydrological variables. The variability in field‐observed GPP, net primary productivity and solar‐induced fluorescence was better or equally well captured by our LRF‐based GPP when compared with six state‐of‐the‐art Earth observation‐based GPP products. Over the period 1982–2015, the LRF‐based average annual global terrestrial GPP budget was 121.8 ± 3.5 Pg C, with a detrended inter‐annual variability of 0.74 ± 0.13 Pg C. The strongest inter‐annual variability was observed in semi‐arid regions, but croplands in China and India also showed strong inter‐annual variations. The trend in global terrestrial GPP during 1982–2015 was 0.27 ± 0.02 Pg C year−1, and was generally larger in the northern than the southern hemisphere. Most positive GPP trends were seen in areas with croplands whereas negative trends were observed for large non‐cropped parts of the tropics. Trends were strong during the eighties and nineties but levelled off around year 2000. Other GPP products either showed no trends or continuous increase throughout the study period. This study benchmarks a first global Earth observation‐based model using an asymptotic light response function, improving simulations of GPP, and reveals a stagnation in the global GPP after the year 2000.

Different Satellite Products Revealing Variable Trends in Global Gross Primary Production

“…The LRF GPP uses an ecosystem‐level physiological approach based on an asymptotic LRF between GPP and PAR to determine upper layer GPP values based on phenology and temperature/moisture factors that can reduce GPP from its maximum potential value (Tagesson et al., 2021). Although LRF‐GPP model does not take into account the effects of land cover change, CO 2 and other factors bring some errors, LRF‐GPP is better able to characterize the observed response at high spatial and temporal resolution, and the asymptotic model with daily temporal resolution plays a very important role in improving the temporal and spatial resolution of GPP product (Tagesson et al., 2021). The LRF GPP has been shown to better capture changes in GPP (Tagesson et al., 2021; Zhang et al., 2008).…”

“…With the development and application of quantitative remote sensing technology, satellite remote sensing provides continuous observation of vegetation dynamics on large regional scales, offering valuable opportunities for monitoring spatial and temporal changes in GPP on a global scale (Liang et al., 2021). Over recent decades, numerous models have been developed to simulate GPP (e.g., light energy use efficiency (LUE) models and process‐based models), and several global GPP products have been derived (Bai et al., 2021; Jiang & Ryu, 2016; Jung et al., 2011; Running et al., 2015; Tagesson et al., 2021; Tramontana et al., 2016; Wang et al., 2021; Yuan et al., 2010; Zhang et al., 2017). For example, the Global LAnd Surface Satellite (GLASS) GPP is a long‐time series, high accuracy, global surface GPP product derived from multi‐source data and ground‐based observations (Yuan et al., 2007, 2010).…”

Contrasting ecosystem vegetation response in global drylands under drying and wetting conditions

“…Increasing VPD and radiation (less cloud cover) increase transpiration, which causes plants to close their stomata to minimize water loss, and ultimately lead to a reduction in photosynthesis (Tagesson et al, 2021). The probability of plants to frequently experience water‐related stress (called stress hereafter) further increases with water deficit in the soil from reduced precipitation (Barkhordarian et al, 2019).…”

Predicting resilience and stability of early second‐growth forests

“…Thus, we measured resilience as the increase of the primary productivity in the time series trend (deterministic change) (Legendre & Legendre, 2012). The trend in the interannual time series of the NDVI is an effective proxy for long‐term vegetation growth, recovery and succession (Caughlin et al., 2020; Fensholt & Proud, 2012; Tagesson et al., 2020; Verbesselt et al., 2010). Thus, the magnitude of the deterministic trend of the NDVI time series can indicate the recovery rate of a forest after deforestation (Caughlin et al., 2020).…”