Coal fly ash is a major carbon flux in the Chang Jiang (Yangtze River) basin

Significance

Coal combustion releases CO₂ but also leaves behind solid waste, or fly ash, which contains considerable amounts of carbon. The organic carbon sourced from fly ash resists chemical breakdown, and we find that it now contributes nearly half of the fossil organic carbon exported by the Chang Jiang—the largest river in Asia. The fly ash flux in this basin is similar to the natural sediment flux to the oceans because dam building has reduced sediment transport, while increased coal consumption generates abundant fly ash. Our results show that fly ash is an important component of the present-day carbon load in rivers and illustrates that human-driven carbon cycling can match the pace of the geological carbon cycle at decadal timescales.

Abstract

Fly ash—the residuum of coal burning—contains a considerable amount of fossilized particulate organic carbon (FOC_ash) that remains after high-temperature combustion. Fly ash leaks into natural environments and participates in the contemporary carbon cycle, but its reactivity and flux remained poorly understood. We characterized FOC_ash in the Chang Jiang (Yangtze River) basin, China, and quantified the riverine FOC_ash fluxes. Using Raman spectral analysis, ramped pyrolysis oxidation, and chemical oxidation, we found that FOC_ash is highly recalcitrant and unreactive, whereas shale-derived FOC (FOC_rock) was much more labile and easily oxidized. By combining mass balance calculations and other estimates of fly ash input to rivers, we estimated that the flux of FOC_ash carried by the Chang Jiang was 0.21 to 0.42 Mt C⋅y⁻¹ in 2007 to 2008—an amount equivalent to 37 to 72% of the total riverine FOC export. We attributed such high flux to the combination of increasing coal combustion that enhances FOC_ash production and the massive construction of dams in the basin that reduces the flux of FOC_rock eroded from upstream mountainous areas. Using global ash data, a first-order estimate suggests that FOC_ash makes up to 16% of the present-day global riverine FOC flux to the oceans. This reflects a substantial impact of anthropogenic activities on the fluxes and burial of fossil organic carbon that has been made less reactive than the rocks from which it was derived.

Fossil particulate organic carbon (FOC) is a geologically stable form of carbon that was produced by the ancient biosphere and then buried and stored in the lithosphere; it is a key player in the geological carbon cycle (1–7). Uplift and erosion liberate FOC from bedrock, delivering it to the surficial carbon cycle. Some is oxidized in sediment routing systems, but a portion escapes and can be transported by rivers to the oceans (5, 8–10). Oxidation of FOC represents a long-term atmospheric carbon source and O₂ sink, whereas the reburial of FOC in sedimentary basins has no long-term net effect on atmospheric CO₂ and O₂ (1, 9, 11). Exhumation and erosion of bedrock provide a natural source of FOC (2, 8), which we refer to as FOC_rock. Human activities have introduced another form of FOC from the mining and combustion of coal. Burning coal emits CO₂ to the atmosphere but also leaves behind solid waste that contains substantial amounts of organic carbon (OC) that survives high-temperature combustion (12–14). This fossil-fuel-sourced carbon represents a poorly understood anthropogenic flux in the global carbon cycle; it also provides a major source of black carbon, which is a severe pollutant and climate-forcing agent (12–15).

Previous studies sought to quantify black carbon in different terrestrial and marine environments and to distinguish fossil fuel versus forest fire sources (14–18). In this study, we focused on fly ash—the material left from incomplete coal combustion. As a major fossil fuel, coal supplies around 30% of global primary energy consumption (19, 20). Despite efforts to capture and utilize fly ash, a fraction enters soils and rivers; the resulting fossil OC from fly ash (FOC_ash) has become a measurable part of the contemporary carbon cycle (14). FOC_ash is also referred to as “unburned carbon” in fly ash (21–25); it provides a useful measure of combustion efficiency and the quality of fly ash as a building material (e.g., in concrete) (23–26). Industrial standards of FOC_ash content in fly ash have been established for material quality assurance (23, 24, 26, 27). However, the characteristics and fluxes of FOC_ash released to the environment, and how these compare to FOC_rock from bedrock erosion, remain less well understood.

To fill this knowledge gap, we examined the Chang Jiang (Yangtze River) basin in China—a system that allowed us to evaluate the influence of FOC_ash on the carbon cycle at continental scales. In the 2000s, China became the largest coal-consuming country in the world, with an annual coal consumption of over 2,500 Mt, equating to ∼50% of worldwide consumption (19, 20, 28). Coal contributed over 60% of China’s national primary energy consumption through the 2000s. A significant portion of this coal (approximately one-third) was consumed in the Chang Jiang (CJ) basin, where China’s most populated and economically developed areas are located (29). Significant amounts of fly ash and FOC_ash continue to be produced and consumed in the CJ basin. To determine the human-induced FOC_ash flux, we investigated the FOC_ash cycle in the CJ basin. We characterized OC in a series of samples including fly ash, bedrock sedimentary shale, and river sediment through multiple geochemical analyses. We then estimated the CJ-exported FOC_ash flux and evaluated how human activities modulated FOC transfer at basin scales. We found that in the CJ basin, coal combustion and dam construction have conspired to boost the FOC_ash flux and reduce the FOC_rock flux carried by the CJ; as a result, these two fluxes converged over an interval of 60 y.

The CJ Basin

The CJ basin is located in central China and has a drainage area of around 1,800,000 km²—nearly 20% of the total terrestrial area of China (30, 31). Originating from the Tibetan Plateau, the CJ drains mountainous areas in its upper reach and alluvial plains in its lower reach, before emptying into the Eastern China Sea (Fig. 1). The regional climate is controlled by the East Asian monsoon with peak precipitation and associated floods occurring from May to September (30, 31). The bedrock geology of the CJ basin is mainly composed of sedimentary rocks, with minor igneous and metamorphic exposures (Fig. 1A) (32, 33).

Fig. 1.

Maps of the CJ (Yangtze River) basin including (A) bedrock geology and (Inset) regional context (33, 38), (B) the distribution of reservoirs (capacity >0.1 km³) with the major tributaries of the CJ denoted (details in SI Appendix), and (C) the distribution of major coal-fired power plants (capacity >100 MW) (details in SI Appendix).

Over the past 60 y, the CJ basin has undergone significant anthropogenic change including the massive construction of industrial and hydrological infrastructure (e.g., coal-fired power plants and dams) and increased coal combustion (Fig. 1 B and C) (34–36). More than 50,000 dams, including the world’s largest—the Three Gorges Dam—have been built in the CJ basin; these dams trap sediment in the resulting reservoirs and have significantly reduced the CJ sediment export to the ocean (Fig. 2) (34, 37). In the process, they profoundly altered the natural transfer of carbon (38, 39). The growth in China’s coal consumption over the past decades has also fundamentally altered the CJ basin, and the annual production of fly ash is, remarkably, now comparable in total mass to the entire CJ sediment flux (Fig. 2). It is not a surprise, then, that fly ash particles and associated black carbon have been observed in recent river and delta sediment in the CJ basin (16, 40–43).

Fig. 2.

Time series of annual sediment export fluxes by the CJ plotted alongside fly-ash production in the basin during 1950 to 2010 (details in SI Appendix).

Materials and Approaches

We sampled fly ash, shale, river sediment, and plant debris (leaves of typical plants) in the middle-lower CJ basin and collected samples from a core drilled into the CJ estuary delta recovering Holocene sediment (Fig. 1). To characterize OC in the samples, we measured OC content and conducted chemical oxidation experiments, ramped pyrolysis oxidation (RPO) analysis, and Raman spectral analysis (details in SI Appendix).

Characteristics of Fly-Ash-Sourced OC.

The CJ fly-ash samples we have studied have an FOC_ash content ([FOC]_ash) of 2.25^+1.63/_−1.18% (Fig. 3A, reported as the median and 16th and 84th percentiles)—a value 2 to 100 times higher than the FOC content in global river sediments (38, 44). The measured FOC_ash content is comparable to the Chinese industrial standard (GB/T 477-2008) of 5% for use in concrete and cement and to the standards of FOC_ash content in other major coal-consuming countries (SI Appendix, section S6) (24, 26, 27).

Fig. 3.

Geochemical characteristics of CJ sediment and fly-ash samples. (A) Histograms of OC content in CJ fly-ash samples with a box-and-whisker plot showing the 5th, 16th, 50th, 84th, and 95th percentiles and mean of OC content for fly-ash samples. (B) Selected Raman spectra of CJ fly-ash samples showing a mix of organic and inorganic mineral phases with mineral and graphitic carbon peaks (D and G bands) denoted (for more data see SI Appendix, Figs. S2 and S3). (C) Probability density of activation energy (E_a) of a fly-ash sample and a shale sample derived from RPO results, indicating a higher modal E_a of the fly ash sample than the shale sample. (D) Oxidation fraction f_ox versus 1/[OC] (reciprocal of OC content) for studied samples, with two dashed lines indicating mixing trends between the fossil OC and biospheric OC end members. (E) Mixing relationships for the CJ sediment collected in the dry season (yellow) versus in the flood season (blue), defined by least-squares linear regression.

In general, FOC_ash comprises a spectrum of carbon species with different forms and origins (21–24). Substantial effort in prior work has been spent on imaging, characterizing, and separating FOC_ash of different carbon species (e.g., refs. 23, 24, 45–48 and SI Appendix, section S6). Notably, studies using electron microscopy found that FOC_ash includes nanometer-scale soot particles, micrometer-sized char particles, and carbon associated with inorganic minerals (23, 24, 48–50). The relative abundances of the different carbon forms in fly ash vary due a range of factors including coal rank and combustion conditions (22, 23, 51, 52), as well as separation method (see more discussion in SI Appendix, section S7). A study using liquid-suspension gravity separation found that soot represents a nonnegligible component in fly ash, contributing ∼35% carbon mass to the total FOC_ash (53).

Raman spectral analyses of our CJ samples provided insight into the chemical moieties of FOC_ash, specifically the presence of graphitic structure and its association with minerals (Fig. 3B and SI Appendix, Figs. S2 and S3). Graphitic carbon was indicated by characteristic G and D bands, at ∼1,350 cm⁻¹ and ∼1,600 cm⁻¹, respectively (Fig. 3B and SI Appendix, Figs. S2 and S3), with the G band corresponding to graphite and the D band induced by defects (5, 54). Nongraphitic carbon, or carbon bonded to heteroatoms (e.g., nitrogen and oxygen), likely caused the high background florescence observed in some samples (SI Appendix, Fig. S2) (55). Mineral-associated carbon featured spectral peaks of minerals and graphitic carbon, and often high background fluorescence (Fig. 3B and SI Appendix, Fig. S2). The Raman analyses also resolved FOC_ash as individual carbon particles, displaying no close association with minerals (Fig. 3B and SI Appendix, Figs. S2 and S3)—textures consistent with other observations of char and soot particles in coal ash (23, 24). We collected a total of 30 Raman spectra in seven of the CJ ash samples: 20 of these displayed mineral-associated carbon containing both graphitic and nongraphitic carbon, 5 revealed graphitic carbon not associated with minerals, and the remaining 5 displayed mixtures of graphitic and nongraphitic carbon not in association with minerals (SI Appendix, Figs. S2 and S3).

Complementing the Raman data, we collected new RPO and chemical oxidation data, which showed that FOC_ash is much more recalcitrant than shale-derived FOC_rock (Fig. 3 C and D). The RPO results were reported as thermograms and translated to spectra of activation energies (E_a) (56). In the thermograms, FOC_ash has a higher fraction of carbon decomposed and more CO₂ released at high temperatures (e.g., >700 °C) than FOC_rock (SI Appendix, Fig. S6). Converting the thermograms to the spectra of activation energies, we found that fly ash featured much higher E_a than the shale (Fig. 3C) and other FOC_rock-dominated river sediment samples (9, 56), revealing a high thermal stability and an extremely refractory phase of FOC_ash (E_a >220 kJ/mol). The chemical oxidation experiments employed sodium persulfate (Na₂S₂O₈)—a chemical with a very high oxidation potential (standard oxidation–reduction potential is E⁰ ∼2.01V) comparable to O₃ (E⁰ ∼2.07V) and much higher than O₂ (E⁰ ∼1.23V) (57). This agent has been used in soil studies to simulate oxidation in natural environments (58). The results from oxidation experiments are reported as f_ox, the mass fraction of OC that gets oxidized in the experiments. A low f_ox value means the sample was difficult to oxidize and contains a high proportion of OC that is recalcitrant. For the fly ash samples, f_ox is low (0 to 0.25), as expected since FOC_ash has undergone pedogenesis, petrogenesis, and high-temperature incineration, leaving behind the most refractory OC class (14). For shale samples, f_ox values were much higher (0.4 to 0.85), implying that a significant portion of FOC_rock is reactive and labile, which is consistent with field observations of substantial loss of FOC_rock in soils and sediment routing systems (5, 59, 60). Expectedly, the plant samples were highly labile, with a high f_ox of 0.9 to 1.

Although current methods cannot delineate the microscale (nanometer to micrometer) characteristics of FOC_ash and directly quantify all the different carbon forms in fly ash (SI Appendix, section S2), our Raman and RPO results together provided constraints on the carbon species composing FOC_ash in our samples—and this compositional information helped explain results from the oxidation experiments. First, the Raman data indicated the presence of multiple different carbon species, including the dominance of mineral-associated carbon. Second, considering that graphitic carbon is thermally decomposed at temperatures of 700 to 800 °C, the RPO thermograms designated that 20 to 50% of FOC_ash is composed of graphitic carbon (SI Appendix, Fig. S4) (61). Altogether, these analyses of our fly ash samples reveal that graphitic carbon is the major constituent in FOC_ash and nongraphitic carbon occurs mostly in association with minerals. These two forms of carbon—graphitic and mineral-associated—hinted at two plausible mechanisms for preserving FOC_ash during high-temperature combustion: selective preservation as stable graphitic carbon and mineral protection (62, 63). Nongraphitic carbon is less recalcitrant than graphitic carbon (5, 47). Thus, the relative proportions of graphitic versus nongraphitic carbon may also explain the natural variation in f_ox observed in our fly ash samples (Fig. 3 and SI Appendix, Figs. S2 and S3). In any case, the low f_ox of FOC_ash underscored its recalcitrance; simply put, it is a less reactive component in the surficial carbon cycle than rock-derived FOC.

OC in the CJ River Sediments.

Chemical oxidation procedures, in conjunction with radiocarbon analyses, also helped resolve the bulk composition of riverine OC. OC in river sediment represents a mixture of radiocarbon-enriched biospheric OC and radiocarbon-dead FOC (11, 64). In f_ox-1/OC space, CJ suspended sediment sits on a mixing trend between a biospheric OC end member and a radiocarbon-dead FOC end member (Fig. 3D and SI Appendix). Coupling this mixing relationship with prior chemical oxidation results reveals that CJ river FOC could not be oxidized via the chemical treatment method and made up 85% of the residual OC after oxidation (38).

The f_ox of the core sediment samples from the CJ estuary had a range similar to the CJ suspended sediment samples (0.5 to 0.8). Those samples came from floodplain and fluvial sedimentary facies (65), representing ash-uncontaminated, preindustrial CJ sediment. If we assumed a behavior of FOC in those preindustrial samples similar to that found during the oxidation experiment (i.e., riverine FOC cannot be oxidized and makes up 85% of the residual OC after oxidation), we estimated a FOC_{preindustrial} content of 0.15 ± 0.02%—a value 6 to 30 times lower than [FOC]_ash (2.25^+1.63/_−1.18%). We treated this estimate as an approximation of rock-derived FOC in the modern CJ sediment (FOC_CJ0) without FOC_ash contamination. Note that FOC_CJ0 represents the final product of FOC_rock after alteration during erosion and transfer before entering the sedimentary OC pool, thus it is not necessarily equivalent to the fresh FOC_rock derived from rocks within the CJ basin. To validate our estimate of [FOC]_{preindustrial}, we defined a similar OC mixing trend for preindustrial CJ sediment using our inferred FOC end member (f_ox = 0 and [FOC] = 0.15 ± 0.02%) and the biospheric OC end member (Fig. 3D). The Holocene sediment samples fit this mixing trend (Fig. 3D), suggesting our estimate of [FOC]_{preindustrial} is robust.

Examining the geochemical data, we also identified hydrologic and sediment transport controls on the degree of mixing observed between FOC_ash and FOC_CJ0. When separating the CJ sediment samples by river discharge (flood versus dry season) (38), distinct trends emerged (Fig. 3E and SI Appendix, Fig. S2): The dry season featured an FOC-enriched end member, whereas the flood season marked an FOC-diluted end member. The difference could be explained by sediment transport conditions varying with the hydrograph. Fly ash is mainly composed of fine cenosphere particles (∼1s to 10s µm) with a density close to or lower than typical river sediment (25, 28) and can be easily entrained by rivers during low flow. The proportion of fly ash in sediment, therefore, is lower in the flood season than in the dry season, because flooding can entrain coarser and denser sediment. The high fraction of coarse sediment and its low [FOC]_CJ0 (assuming it is equivalent to the preindustrial [FOC]) would lead to a mixed [FOC] during flooding that is lower than the dry season. Thus, our observation indicated some hydraulic control on the composition of riverine FOC as a mixture of high-[FOC] fly ash and low-[FOC] natural sediment, corroborating the low [FOC]_CJ0 seen in Holocene sediment.

Fly-Ash-Sourced OC Flux.

During 2007 to 2008, around 130 Mt of fly ash was produced in the CJ basin (36% of the total fly-ash production in China; SI Appendix). Taking an ash utilization rate of 67% estimated for China (28), ∼40 Mt of fly ash was released to the environment in the CJ basin; this is a remarkable flux given that the total CJ sediment flux is around 130 Mt⋅y⁻¹. There is currently no systematic storage or treatment of the unutilized ash, suggesting much of this may be released into the environment. From the budget of production alone, it is unclear how much of the wasted fly ash and FOC_ash enter rivers, but we can use three lines of evidence to constrain the riverine-transported FOC_ash flux.

First, we conducted mass balance calculations using the preindustrial [FOC] (0.15 ± 0.02%) as an approximation of the [FOC] of ash-free sediment. Assuming the modern-day CJ sediment is a mixture of fly ash and ash-uncontaminated sediment, we calculated the flux of FOC_ash:

$${\left[\text{FOC}\right]}_{\text{ash}}\times \left(f_{\text{sed}-\text{ash}}\right)+{\left[\text{FOC}\right]}_{\text{CJ0}}\times \left(1-f_{\text{sed}-\text{ash}}\right)={\left[\text{FOC}\right]}_{\text{CJ}},$$ [1]

where f_sed-ash is the mass fraction of fly ash in the ash-contaminated CJ sediment and [FOC]_CJ represents the modern-day CJ FOC. We took 2.25^+1.63/_−1.18% for [FOC]_ash, 0.15 ± 0.02% for [FOC]_CJ0, and 0.45 ± 0.10% for [FOC]_CJ (38), to resolve f_sed-ash and the fraction of FOC_ash in the CJ-exported FOC, f_FOC-ash. We then employed Monte Carlo random sampling techniques to propagate errors and estimate uncertainties (SI Appendix), reported as medians and the 16th to 84th percentiles of the sampling results. We obtained a f_FOC-ash of 72⁺⁸/₋₁₂%, with a FOC_ash riverine flux of 0.42^+0.14/_−0.15 Mt C, whereas the total CJ-exported FOC flux is 0.58 ± 0.13 Mt C⋅y⁻¹ (38). We also estimated a f_sed-ash of 13⁺¹⁸/₋₇%, which is 13⁺¹⁸/₋₇% of the total produced fly ash and 39⁺⁵⁴/₋₂₁% of the ash wasted in the basin (Fig. 4A).

Fig. 4.

Fluxes of FOC, fly ash, and sediment in the CJ basin. (A) f_FOC-ash (fraction of fly ash-sourced fossil OC in the total flux of fossil OC exported by CJ) as a function of f_sed-ash (mass fraction of fly ash in the total sediment flux) (curves with uncertainty bands determined from Monte Carlo simulations; SI Appendix), with the mixing trend between FOC_ash and FOC_CJ0 (FOC in ash-free sediment), fly-ash release (dashed line, difference between produced and utilized fly ash), and estimates of f_sed-ash (red line from changes in MS of the CJ sediment, and black line and gray range from mass balance calculations). (B) Different types of carbon flux in the CJ basin where “1” and “2” denote riverine FOC_ash flux estimated from mass balance calculations and from changes in MS of CJ sediment, respectively. The released FOC_ash flux is estimated from this study as the product of the released fly ash flux and the FOC_ash content. The 2007 to 2008 riverine FOC flux is estimated in ref. 35, and the silicate weathering flux (2006) is from ref. 25. The riverine FOC_rock flux (2007 to 2008) and the riverine FOC flux (1950s) are quantified from this study. (C) Changes of riverine FOC_rock flux versus released (produced − utilized) FOC_ash flux in 1950 to 2010. (D) Sediment flux and fly-ash flux in the CJ basin in 1950 to 1970 (predamming) and 2004 to 2010 (postdamming).

Second, we referred to a prior study that estimated a f_sed-ash of 7% based on the changes in the magnetic susceptibility (MS) of river sediment and islet deposits. An abrupt increase in the MS of the CJ sediment has been observed in recent years and attributed to the input of fly ash, which has MS ∼30 times higher than ash-free sediment (40). Using a f_sed-ash of 7%, a CJ sediment flux of 130 Mt (66), and an FOC content of 2.25^+1.63/_−1.18% in our fly ash samples, we estimated a FOC_ash flux of 0.21^+0.15/_−0.11 and a f_FOC-ash of 37⁺⁴⁰/₋₁₉% of the CJ FOC export. In this case, 21% of the unutilized fly ash in the CJ basin enters the rivers (Fig. 4A).

Third, we noticed that the FOC content (0.45 ± 0.10%) in our sediment samples from the lower reach of CJ was typically higher than in the samples (0.10 to 0.20%) from regions upstream of the Three Gorges Dam and areas of intense coal consumption (39, 67). Attributing this downstream increase in FOC content to ash input resulted in estimates of f_FOC-ash of ∼60 to 80%—values similar to those achieved via mass balance. To formally demonstrate that FOC content increases downstream as a result of fly ash input, one would need more depth-profile sampling to capture variations in riverine OC across hydraulic gradients, and additional accounting for the bedrock and landscape heterogeneities within the CJ basin (e.g., variations in lithology and contributions from different tributaries) (11, 38, 39, 67). Nonetheless, this first-order estimate was similar to that achieved with our two other approaches for quantifying f_FOC-ash.

The estimate of f_FOC-ash from the change in MS (37⁺⁴⁰/₋₁₉%) is somewhat lower than the estimate from mass balance calculations (72⁺⁸/₋₁₂%) and the downstream trends in FOC content (∼60 to 80%). This discrepancy can be explained because the mass balance calculations assume FOC_ash is the only anthropogenic FOC input to the CJ FOC pool. As there may be other anthropogenic sources of fossil carbon, e.g., from petrochemicals (68–70), our estimate based from mass balance likely sets an upper bound on f_FOC-ash and FOC_ash flux.

Overall, the multiple approaches yield results of a similar magnitude, suggesting this estimate of FOC_ash flux is robust to a first order. Combining the results above, we concluded that in 2007 to 2008, fly ash contributed a FOC_ash flux of 0.21 to 0.42 Mt C⋅y⁻¹, making up 37 to 72% of the CJ-exported FOC (Fig. 4B). For comparison, the estimated atmospheric carbon drawdown via silicate weathering in the basin is 2.29 ± 0.53 Mt C⋅y⁻¹ (Fig. 4B) (32).

Fossil Carbon Flux Perturbed by Human Activities.

Globally, the total fly ash production during the 2000s was around 750 Mt⋅y⁻¹ (28, 71), with annual global river sediment export estimated at 17,800 Mt⋅y⁻¹ (72). Thus, while the mass flux of fly ash in the CJ now matches the scale of natural sediment transfer in this basin (Fig. 2), the same is not true globally. The globally averaged utilization rate of fly ash is not well determined, but the major coal consumers (China, United States, and India, accounting for 70% of the total consumption) reported an average utilization rate of ∼50% during 2007 to 2008 (28), leaving ∼375 Mt⋅y⁻¹ of fly ash that was not utilized. A global-average FOC_ash content is challenging to estimate accurately, because FOC_ash depends on a range of factors including coal types and combustion conditions (see expanded discussion in SI Appendix, section S6) (23, 24, 45, 51)—conditions that likely vary from region to region. For a first-order constraint, we compiled data on FOC_ash content for 247 samples from different regions and found an FOC_ash content of 4.70^+9.69/_−3.40% (median and 16th to 84th percentiles; SI Appendix, section S7 and Fig. S6), which was on the same order of magnitude as the FOC_ash content (2.25^+1.63/_−1.18%) measured in our CJ fly-ash samples. Note that most global ash FOC content data were estimated via the loss on ignition—a method that can overestimate the true FOC content and thus may partially explain why the CJ FOC_ash content is lower than estimated globally (see more discussion in SI Appendix, section S7). Considering the global data compilation, the CJ FOC_ash content likely represents a conservative estimate of the actual carbon content in fly ash. Notably, FOC_ash contents we measured in CJ ash samples and those from the compiled global dataset were similar to the industrial standards of FOC_ash content of 5 to 10% in different countries and regions (SI Appendix, section S6) (23, 24, 26, 27); this lent confidence to the overall estimates of the FOC_ash flux. If [FOC]_ash found in the CJ samples is typical, a global FOC_ash yield of 8.43^+6.11/_−4.43 Mt C⋅y⁻¹ can be expected. Assuming 20% of the unutilized fly ash is transported by rivers (we found that this number was 13 to 27% for the CJ basin), the global riverine FOC_ash flux to the oceans is then 1.69^+1.22/_−0.89 Mt C⋅y⁻¹, making up 3.9⁺¹²/_−3.2% of the modern-day riverine FOC flux (43⁺⁶¹/₋₂₅ Mt C⋅y⁻¹) (44). Note that this estimate of global FOC_ash flux carries large uncertainties, and further studies of FOC_ash contents and ash supply to rivers in different regions will be required to improve upon it. Nonetheless, our first-order estimate of the global-average f_FOC-ash of ∼4% is lower than the CJ case (37 to 72%), meaning that the CJ basin likely represents an upper end member in the distribution of FOC_ash production and export. So why does the CJ basin have such a high f_FOC-ash, and such a dominant overall flux of fly ash? We attributed this to two of the major anthropogenic modifications of the CJ basin: increasing coal consumption and dam construction.

First, coal consumption in China has substantially increased over the past 60 y, boosting fly ash production (Figs. 2 and 4C). Hosting China’s most economically developed and populated areas, the CJ basin has witnessed intense construction of coal-fired power and steel plants (Fig. 1C), providing a major source of FOC_ash, especially in its middle and lower reaches (35, 70, 73). In the 2000s, China’s consumption increased to more than 50% of the global coal consumption. Alone, all the provinces in the CJ basin comprise 36% of China’s coal consumption—a value equivalent to 18% of the global coal consumption. Thus, the CJ basin represents a major locus of fly ash production.

Second, the continued construction of dams and reservoirs in the CJ basin has decreased fluxes of sediment and FOC_rock. After the impoundment of the Three Gorges Reservoir in 2003, the CJ sediment export has reduced to ∼100 Mt⋅y⁻¹ from ∼500 Mt⋅y⁻¹ before the dams were emplaced (Fig. 4 C and D) (34), contributing <1% to the global riverine sediment flux to the oceans (72). Since FOC_rock is mainly eroded from upstream areas and scales with sediment flux (38, 44), a significant proportion of FOC_rock is being sequestered in the reservoirs, leading to a reduction in the riverine-carried FOC_rock flux in the lower CJ reach where many large-scale power and steel plants, the major sources of FOC_ash, are located (Fig. 1C).

Transport and Fate of Fossil OC.

Our analysis revealed that FOC_ash and FOC_rock have systematically distinct reactivities. When compared overall to biospheric carbon, the low f_ox and high E_a values we observed indicated that FOC_ash is expected to be significantly more recalcitrant and conserved during transport and storage compared to FOC_rock. These differences will affect their fates during fluvial transport and storage, leading to different impacts on the carbon cycle. For FOC_ash, the graphitic component is less reactive and is expected to be more conserved as it transits the landscape, whereas the nongraphitic carbon is more reactive and thus more likely to interact with active carbon-cycle processes in surface and subsurface environments (5, 47).

The fate of FOC_rock as it becomes exposed to surface processes depends on its chemical properties and the geomorphic setting. We saw that FOC_rock in the CJ basin contains a large fraction that is labile and prone to oxidation (Fig. 3), whereas FOC in river sediments is comparatively recalcitrant (38). This suggested the preferential loss of the labile component in FOC_rock during denudation, fluvial transport, and sediment storage in the CJ system—a pattern consistent with prior observations of significant oxidation of FOC_rock in large floodplain systems (e.g., Amazon) (59). Although laboratory experiments have intimated slow reaction kinetics of FOC_rock oxidation (e.g., first-order kinetic coefficients on the order of 10⁻³ to 10⁻⁴ y⁻¹) (74), the long transit time of OC and sediment (e.g., ∼10⁴ to 10⁵ y) in large floodplain systems (75, 76) matches or exceeds the characteristic reaction timescales of FOC_rock and thus would allow sufficient reaction time for FOC_rock oxidation. In contrast, FOC_rock oxidation might be kinetically limited in river systems with smaller catchment sizes and shorter transit timescales such as the rivers in mountainous islands (8). Overall, we hypothesized that the differences in the reactivity of FOC_rock and FOC_ash translate into the differences in their fates most profoundly in large alluvial systems (e.g., CJ and Amazon) with long transit times, and such differences are expected to be dampened in smaller catchments with shorter transit times.

Dam building in the CJ basin has probably influenced the fate of FOC_rock during fluvial transit as well. Previous studies suggested that the high sedimentation rate in the reservoirs would limit oxygen exposure time of carbon-bearing particles and promote their preservation (11, 38, 39, 64). Thus, the reservoirs in the CJ basin might be expected to help sequester and preserve FOC_rock from upstream CJ, buffering its oxidation in downstream floodplains and estuaries (38). In addition, although there has been an increased supply of FOC from coal ash in the middle-lower CJ basin; one prior study hypothesized that the emplacement of the Three Gorges Dam in 2003 would lead to younger and fresher OC exported by the CJ (39). We do not have upstream samples during our study time interval to delineate the downstream changes of riverine OC in 2007 to 2008. However, we noticed that the average proportion of FOC in our 2007 to 2008 sediment samples (∼25%) was higher than the FOC proportion in the middle-to-lower CJ sediment samples (∼10%) collected 1 to 2 y later after our sampling time. This difference suggested a temporal shift toward a lower proportion of FOC in the CJ-exported OC—a trend consistent with the proposed change toward a younger and fresher riverine OC flux after the impoundment of the Three Gorges Reservoir. Nonetheless, continued monitoring and systematic sampling of the whole CJ fluvial network are needed for a more detailed picture of how hydraulic engineering impacts carbon cycling in this system (38, 39).

FOC_ash and FOC_rock are also carried by particles of different sizes, which can affect their fate via transport processes. FOC_ash is mostly encapsulated in micrometer-sized fly-ash particles, whereas FOC_rock is bound to coarser (e.g., sand-sized) grains. Thus, FOC_ash could more easily bypass dams during flow release, whereas FOC_rock is likely to be sequestered in reservoirs. When delivered to the CJ estuary and the East China Sea margin where hydraulic conditions and sediment transport and storage processes are complex, the fine-grain-carried FOC_ash may have more dynamic behavior (e.g., flocculation settling, suspension, and dispersion) and may be spread over a larger depositional area than FOC_rock (77). The fine particle sizes that carry FOC_ash can also be more efficiently transported by aeolian processes, which can deliver FOC_ash to remote areas beyond riverine transport within a given catchment (78). The aeolian flux of FOC_ash both within and outside of the CJ basin requires further assessment, but we anticipated that these fluxes are minor compared to the riverine flux, considering the dominance of the wet, monsoonal climate in the basin that limits ash transport by aeolian processes (79, 80). In northern China where a drier climate dominates, aeolian processes may well play a more important role transporting FOC_ash (80, 81).

Conclusions and Implications

The CJ basin illustrates how human activities have significantly altered the carbon cycle at continental scales. In the CJ basin, fly ash contributes a remarkable 37 to 72% of the riverine fossil OC exported to the oceans. Driven by the human pursuit of energy, the riverine-carried FOC_ash flux has increased while the riverine FOC_rock flux decreased—and as a result, these two fluxes have converged over an interval of 60 y to amplify the concentration of FOC_ash on the landscape. This serves as an example of how the pace of the human-induced alteration of the carbon cycle can catch up with nature-sourced carbon at decadal timescales and demarcates another dimension of the human imprint on the short-term carbon cycle beyond that directly associated with CO₂ emission during fossil fuel combustion (19, 82).

Our results showed that not all fossil OC is made equal: FOC_rock has a significant fraction that is labile and can be oxidized during transport, whereas FOC_ash is highly recalcitrant (i.e., unreactive) and can be conserved during transport. While coal burning is a leaky process with respect to OC, the way that carbon is transformed by incomplete combustion means that the FOC_ash that escapes this process is much less likely to end up as CO₂ compared to the FOC_rock naturally derived from erosion. Furthermore, its fossil origin means FOC_ash is radiocarbon-dead (7, 11). With increasing coal consumption and ash production (19), FOC_ash flux is expected to increase and to contribute to a greater proportion of the total riverine FOC flux to the oceans. With this growing human-induced carbon flux, caution will need to be taken when interpreting radiocarbon-based material flux as well as records from recent offshore sediments. With this observation in mind, given that the magnitude of fly ash release can match natural sediment fluxes at regional scales (e.g., in the CJ basin), the unique properties of FOC_ash make it a useful tracer of anthropogenic impacts on the OC cycle. By reflecting the history of coal consumption, FOC_ash in sedimentary cores and other archives could provide a distinct marker of the Anthropocene (78, 83).

Data Availability

All study data are included in the article and/or supporting information.