Yilong Wang

Variability of fire carbon emissions in equatorial Asia and its nonlinear sensitivity to El Niño

The large peatland carbon stocks in the land use change-affected areas of equatorial Asia are vulnerable to fire. Combining satellite observations of active fire, burned area, and atmospheric concentrations of combustion tracers with a Bayesian inversion, we estimated the amount and variability of fire carbon emissions in equatorial Asia over the period 1997–2015. Emissions in 2015 were of 0.51u2009±u20090.17u2009Pg carbon—less than half of the emissions from the previous 1997 extreme El Nino, explained by a less acute water deficit. Fire severity could be empirically hindcasted from the cumulative water deficit with a lead time of 1 to 2u2009months. Based on CMIP5 climate projections and an exponential empirical relationship found between fire carbon emissions and water deficit, we infer a total fire carbon loss ranging from 12 to 25u2009Pg by 2100 which is a significant positive feedback to climate warming.

Reducing uncertainties in decadal variability of the global carbon budget with multiple datasets

Significance The conventional approach of calculating the global carbon budget makes the land sink the most uncertain of all budget terms. This is because, rather than being constrained by observations, it is inferred as a residual in the budget equation. Here, we overcome this limitation by performing a Bayesian fusion of different available observation-based estimates of decadal carbon fluxes. This approach reduces the uncertainty in the land sink by 41% and in the ocean sink by 46%. These results are significant because they give unprecedented confidence in the role of the increasing land sink in regulating atmospheric CO2, and shed light on the past decadal trend. Conventional calculations of the global carbon budget infer the land sink as a residual between emissions, atmospheric accumulation, and the ocean sink. Thus, the land sink accumulates the errors from the other flux terms and bears the largest uncertainty. Here, we present a Bayesian fusion approach that combines multiple observations in different carbon reservoirs to optimize the land (B) and ocean (O) carbon sinks, land use change emissions (L), and indirectly fossil fuel emissions (F) from 1980 to 2014. Compared with the conventional approach, Bayesian optimization decreases the uncertainties in B by 41% and in O by 46%. The L uncertainty decreases by 47%, whereas F uncertainty is marginally improved through the knowledge of natural fluxes. Both ocean and net land uptake (B + L) rates have positive trends of 29 ± 8 and 37 ± 17 Tg C⋅y−2 since 1980, respectively. Our Bayesian fusion of multiple observations reduces uncertainties, thereby allowing us to isolate important variability in global carbon cycle processes.

On the causes of trends in the seasonal amplitude of atmospheric CO2

No consensus has yet been reached on the major factors driving the observed increase in the seasonal amplitude of atmospheric CO2 in the northern latitudes. In this study, we used atmospheric CO2 records from 26 northern hemisphere stations with a temporal coverage longer than 15xa0years, and an atmospheric transport model prescribed with net biome productivity (NBP) from an ensemble of nine terrestrial ecosystem models, to attribute change in the seasonal amplitude of atmospheric CO2 . We found significant (pxa0<xa0.05) increases in seasonal peak-to-trough CO2 amplitude (AMPP-T ) at nine stations, and in trough-to-peak amplitude (AMPT-P ) at eight stations over the last three decades. Most of the stations that recorded increasing amplitudes are in Arctic and boreal regions (>50°N), consistent with previous observations that the amplitude increased faster at Barrow (Arctic) than at Mauna Loa (subtropics). The multi-model ensemble mean (MMEM) shows that the response of ecosystem carbon cycling to rising CO2 concentration (eCO2 ) and climate change are dominant drivers of the increase in AMPP-T and AMPT-P in the high latitudes. At the Barrow station, the observed increase of AMPP-T and AMPT-P over the last 33xa0years is explained by eCO2 (39% and 42%) almost equally than by climate change (32% and 35%). The increased carbon losses during the months with a net carbon release in response to eCO2 are associated with higher ecosystem respiration due to the increase in carbon storage caused by eCO2 during carbon uptake period. Air-sea CO2 fluxes (10% for AMPP-T and 11% for AMPT-P ) and the impacts of land-use change (marginally significant 3% for AMPP-T and 4% for AMPT-P ) also contributed to the CO2 measured at Barrow, highlighting the role of these factors in regulating seasonal changes in the global carbon cycle.

Estimation of observation errors for large-scale atmospheric inversion of CO2 emissions from fossil fuel combustion

Abstract National annual inventories of CO2 emitted during fossil fuel consumption (FFCO2) bear 5–10% uncertainties for developed countries, and are likely higher at intra annual scales or for developing countries. Given the current international efforts of mitigating actions, there is a need for independent verifications of these inventories. Atmospheric inversion assimilating atmospheric gradients of CO2 and radiocarbon measurements could provide an independent way of monitoring FFCO2 emissions. A strategy would be to deploy such measurements over continental scale networks and to conduct continental to global scale atmospheric inversions targeting the national and one-month scale budgets of the emissions. Uncertainties in the high-resolution distribution of the emissions could limit the skill for such a large-scale inversion framework. This study assesses the impact of such uncertainties on the potential for monitoring the emissions at large scale. In practice, it is more specifically dedicated to the derivation, typical quantification and analysis of critical sources of errors that affect the inversion of FFCO2 emissions when solving for them at a relatively coarse resolution with a coarse grid transport model. These errors include those due to the mismatch between the resolution of the transport model and the spatial variability of the actual fluxes and concentrations (i.e. the representation errors) and those due to the uncertainties in the spatial and temporal distribution of emissions at the transport model resolution when solving for the emissions at large scale (i.e. the aggregation errors). We show that the aggregation errors characterize the impact of the corresponding uncertainties on the potential for monitoring the emissions at large scale, even if solving for them at the transport model resolution. We propose a practical method to quantify these sources of errors, and compare them with the precision of FFCO2 measurements (i.e. the measurement errors) and the errors in the modelling of atmospheric transport (i.e. the transport errors). The results show that both the representation and measurement errors can be much larger than the aggregation errors. The magnitude of representation and aggregation errors is sensitive to sampling heights and temporal sampling integration time. The combination of these errors can reach up to about 50% of the typical signals, i.e. the atmospheric large-scale mean afternoon FFCO2 gradients between sites being assimilated by the inversion system. These errors have large temporal auto-correlation scales, but short spatial correlation scales. This indicates the need for accounting for these temporal auto-correlations in the atmospheric inversions and the need for dense networks to limit the impact of these errors on the inversion of FFCO2 emissions at large scale. More generally, comparisons of the representation and aggregation errors to the errors in simulated FFCO2 gradients due to uncertainties in current inventories suggest that the potential of inversions using global coarse-resolution models (with typical horizontal resolution of a couple of degrees) to retrieve FFCO2 emissions at sub-continental scale could be limited, and that meso-scale models with smaller representation errors would effectively increase the potential of inversions to constrain FFCO2 emission estimates.

Lower land-use emissions responsible for increased net land carbon sink during the slow warming period

The terrestrial carbon sink accelerated during 1998–2012, concurrently with the slow warming period, but the mechanisms behind this acceleration are unclear. Here we analyse recent changes in the net land carbon sink (NLS) and its driving factors, using atmospheric inversions and terrestrial carbon models. We show that the linear trend of NLS during 1998–2012 is about 0.17u2009±u20090.05 Pg C yr−2 , which is three times larger than during 1980–1998 (0.05u2009±u20090.05 Pg C yr−2). According to terrestrial carbon model simulations, the intensification of the NLS cannot be explained by CO2 fertilization or climate change alone. We therefore use a bookkeeping model to explore the contribution of changes in land-use emissions and find that decreasing land-use emissions are the dominant cause of the intensification of the NLS during the slow warming period. This reduction of land-use emissions is due to both decreased tropical forest area loss and increased afforestation in northern temperate regions. The estimate based on atmospheric inversions shows consistently reduced land-use emissions, whereas another bookkeeping model did not reproduce such changes, probably owing to missing the signal of reduced tropical deforestation. These results highlight the importance of better constraining emissions from land-use change to understand recent trends in land carbon sinks.Accelerated storage of terrestrial carbon during the slow warming period (1998–2012) can be predominantly attributed to lower land-use emissions due to decreased tropical forest loss and increased afforestation in the northern temperate regions.

Temporal response of soil organic carbon after grassland-related land-use change

The net flux of CO2 exchanged with the atmosphere following grassland-related land-use change (LUC) depends on the subsequent temporal dynamics of soil organic carbon (SOC). Yet, the magnitude and timing of these dynamics are still unclear. We compiled a global data set of 836 paired-sites to quantify temporal SOC changes after grassland-related LUC. In order to discriminate between SOC losses from the initial ecosystem and gains from the secondary one, the post-LUC time series of SOC data was combined with satellite-based net primary production observations as a proxy of carbon input to the soil. Globally, land conversion from either cropland or forest into grassland leads to SOC accumulation; the reverse shows net SOC loss. The SOC response curves vary between different regions. Conversion of cropland to managed grassland results in more SOC accumulation than natural grassland recovery from abandoned cropland. We did not consider the biophysical variables (e.g., climate conditions and soil properties) when fitting the SOC turnover rate into the observation data but analyzed the relationships between the fitted turnover rate and these variables. The SOC turnover rate is significantly correlated with temperature and precipitation (pxa0<xa00.05), but not with the clay fraction of soils (pxa0>xa00.05). Comparing our results with predictions from bookkeeping models, we found that bookkeeping models overestimate by 56% of the long-term (100xa0years horizon) cumulative SOC emissions for grassland-related LUC types in tropical and temperate regions since 2000. We also tested the spatial representativeness of our data set and calculated SOC response curves using the representative subset of sites in each region. Our study provides new insight into the impact grassland-related LUC on the global carbon budget and sheds light on the potential of grassland conservation for climate mitigation.

Divergent response of seasonally dry tropical vegetation to climatic variations in dry and wet seasons

Interannual variations of photosynthesis in tropical seasonally dry vegetation are one of the dominant drivers to interannual variations of atmospheric CO2 growth rate. Yet, the seasonal differences in the response of photosynthesis to climate variations in these ecosystems remain poorly understood. Here using Normalized Difference Vegetation Index (NDVI), we explored the response of photosynthesis of seasonally dry tropical vegetation to climatic variations in the dry and the wet seasons during the past three decades. We found significant (pxa0<xa00.01) differences between dry and wet seasons in the interannual response of photosynthesis to temperature (γint ) and to precipitation (δint ). γint is ~1%xa0°C-1 more negative and δint is ~8% 100xa0mm-1 more positive in the dry season than in the wet season. Further analyses show that the seasonal difference in γint can be explained by background moisture and temperature conditions. Positive γint occurred in wet season where mean temperature is lower than 27°C and precipitation is at least 60xa0mm larger than potential evapotranspiration. Two widely used Gross Primary Productivity (GPP) estimates (empirical modeling by machine-learning algorithm applied to flux tower measurements, and nine process-based carbon cycle models) were examined for the GPP-climate relationship over wet and dry seasons. The GPP derived from empirical modeling can partly reproduce the divergence of γint , while most process models cannot. The overestimate by process models on negative impacts by warmer temperature during the wet season highlights the shortcomings of current carbon cycle models in representing interactive impacts of temperature and moisture on photosynthesis. Improving representations on soil water uptake, leaf temperature, nitrogen cycling, and soil moisture may help improve modeling skills in reproducing seasonal differences of photosynthesis-climate relationship and thus the projection for impacts of climate change on tropical carbon cycle.

GOLUM-CNP v1.0: a data-driven modeling of carbon, nitrogen and phosphorus cycles in major terrestrial biomes

Global terrestrial nitrogen (N) and phosphorus (P) cycles are coupled to the global carbon (C) cycle for net primary production (NPP), plant C allocation, and decomposition of soil organic matter, but N and P have distinct pathways of inputs and losses. Current C-nutrient models exhibit large uncertainties in their estimates of pool sizes, fluxes, and turnover rates of nutrients, due to a lack of consistent global data for evaluating the models. In this study, we present a new model–data fusion framework called the Global Observation-based Land-ecosystems Utilization Model of Carbon, Nitrogen and Phosphorus (GOLUM-CNP) that combines the CARbon DAta MOdel fraMework (CARDAMOM) data-constrained C-cycle analysis with spatially explicit data-driven estimates of N and P inputs and losses and with observed stoichiometric ratios. We calculated the steady-state Nand P-pool sizes and fluxes globally for large biomes. Our study showed that new N inputs from biological fixation and deposition supplied > 20 % of total plant uptake in most forest ecosystems but accounted for smaller fractions in boreal forests and grasslands. New P inputs from atmospheric deposition and rock weathering supplied a much smaller fraction of total plant uptake than new N inputs, indicating the importance of internal P recycling within ecosystems to support plant growth. Nutrient-use efficiency, defined as the ratio of gross primary production (GPP) to plant nutrient uptake, were diagnosed from our model results and compared between biomes. Tropical forests had the lowest N-use efficiency and the highest P-use efficiency of the forest biomes. An analysis of sensitivity and uncertainty indicated that the NPP-allocation fractions to leaves, roots, and Published by Copernicus Publications on behalf of the European Geosciences Union. 3904 Y. Wang et al.: GOLUM-CNP v1.0 wood contributed the most to the uncertainties in the estimates of nutrient-use efficiencies. Correcting for biases in NPP-allocation fractions produced more plausible gradients of Nand P-use efficiencies from tropical to boreal ecosystems and highlighted the critical role of accurate measurements of C allocation for understanding the N and P cycles.

Recent Changes in Global Photosynthesis and Terrestrial Ecosystem Respiration Constrained From Multiple Observations

©2018. American Geophysical Union. All Rights Reserved. To assess global carbon cycle variability, we decompose the net land carbon sink into the sum of gross primary productivity (GPP), terrestrial ecosystem respiration (TER), and fire emissions and apply a Bayesian framework to constrain these fluxes between 1980 and 2014. The constrained GPP and TER fluxes show an increasing trend of only half of the prior trend simulated by models. From the optimization, we infer that TER increased in parallel with GPP from 1980 to 1990, but then stalled during the cooler periods, in 1990–1994 coincident with the Pinatubo eruption, and during the recent warming hiatus period. After each of these TER stalling periods, TER is found to increase faster than GPP, explaining a relative reduction of the net land sink. These results shed light on decadal variations of GPP and TER and suggest that they exhibit different responses to temperature anomalies over the last 35xa0years.