The NASA Carbon Monitoring System Phase 2 synthesis: scope, findings, gaps and recommended next steps

Tracking plant biomass in terrestrial ecosystems is an essential component of CMSs. Biomass has been a focus of the NASA CMS since the program's inception, and many CMS projects have advanced our understanding of methods to map, track changes over time, characterize uncertainties in biomass estimates and project future storage potential. In addition, from these results CMS projects have been able to improve our understanding of the controls on biomass accumulation and loss in natural and managed systems.

Methodologically, CMS projects have shown that biomass mapping is achievable at multiple scales, and that such mapping is defensible in the context of MRV and international carbon programs such as REDD+. Mapping is possible through a combination of lidar, optical data, radar data, forest inventory data, and emerging statistical and machine learning methods (Junttila et al 2013, Tyukavina et al 2013, Montesano et al 2013, Zhang et al 2014, Potapov et al 2014, Lagomasino et al 2015, Babcock et al 2015, Duncanson et al 2015, Huang et al 2015, Gu et al 2016, Babcock et al 2016, Swatantran et al 2016, Deo et al 2016, 2017, Huang et al 2017, Treuhaft et al 2017, Fekety et al 2018, Babcock et al 2018, Thomas et al 2018, Rappaport et al 2018, Saarela et al 2018, Xu et al 2018, Kennedy et al 2018, Alonzo et al 2018, Huang et al 2019, Simard et al 2019, Patterson et al 2019, Bullock et al 2020, Arevelo et al 2020). In particular, many CMS projects have advanced the statistical tools needed to make such mapping defensible (Montesano et al 2013, Finley et al 2013, 2014, Cooke et al 2016, Datta et al 2016, Reimer et al 2016, Finley et al 2017, Babcock et al 2018, Saarela et al 2018, Patterson et al 2019, Olofsson et al 2020). Importantly, these new methods are also highly generalizable and can also provide quantitative assessments of product uncertainties. In addition, the work has been conducted across a range of diverse ecosystems and are applicable to new space-borne lidar missions such as GEDI (Healey et al 2012, Duncanson et al 2019, Patterson et al 2019).

CMS studies have also contributed significantly to our understanding of the patterns of biomass, drivers of biomass change, and storage potential in vegetation and the soil. For example, human-caused forest disturbance (deforestation and degradation) remains the key driver of biomass and carbon loss in temperate and tropical forest systems, with fire, climate, and longer-duration stressors (drought, insect pests) playing an important role in some systems (e.g. boreal, savanna, peatland) (French et al 2014, Morton 2016, Longo et al 2016, Cohen et al 2016, Baccini et al 2017, Noojipady et al 2017, Rappaport et al 2018, Rangel Pinage et al 2019, Li et al 2020, Tang et al 2020). There is evidence of recent diminishment of fire in regions of low- to intermediate-forest cover, perhaps because of transitions in human management associated with economic development and increase in closed canopy forests (Andela et al 2017). In global drylands systems, there is a strong link between precipitation and the rate and timing of carbon uptake and storage, but which is also mediated by plant adaptation strategies, including higher than expected investment in leaf-level photosynthetic machinery in global drylands ecosystems, with possibly important implications for response to climate change (Biederman et al 2016, 2017, Hinojo-Hinojo et al 2018). At the same time, increasing prevalence of prolonged drought periods may significantly outweigh any mediation by plant adaptation strategies leading to a chronic reduction in vegetation biomass in dryland systems, including the western U.S. Studies utilizing mechanistic ecosystem modeling have leveraged the high resolution lidar and optical data to move beyond traditional MRV and map both current stocks and future storage potentials for use in planning scenarios (Hurtt et al 2019, Ma et al 2021a; figure 9). The most recent results from these products are in active use in state-level carbon inventories and climate mitigation planning.

Zoom In Zoom Out Reset image size

Despite these advances, gaps remain. Major areas of the world, including a diversity of different ecosystems, disturbance regimes, land-use activities, etc have yet to be mapped or modeled or validated at high spatial resolution. Research has identified that a key bottleneck and source of uncertainty in gridded products is accurate quantification and application of the allometric equations needed to create and scale-up reliable biomass reference data (Xu et al 2018). Additionally, approaches to scale-up established methods to continental and global domains are needed. While the separate mapping of biomass and biomass change from disturbance has been successful, a more coordinated effort is needed to help reconcile issues related to biomass and disturbance loss and recovery in space and/or time. The spatial resolution and temporal mismatch in biomass and disturbance mapping efforts complicates the fusion of these datasets to inform C management, including adequately capturing uncertainties related to these isolated mapping efforts. In addition, a major challenge for the mapping of biomass in some high-latitude or tropical regions, is the general lack of suitable measurements needed to train upscaling approaches. This includes both the training data (e.g. measurements of plant height, structure, biomass) themselves and also the remote sensing observations needed for scaling and creating the maps. Finally, despite better inclusion of methodological uncertainties in estimation, it is unclear how to advise practitioners when uncertainties from different sources disagree.

From these findings, challenges, and knowledge gaps, several important next steps emerge. Efforts must continue to expand biomass mapping over broad (continental to global) domains at a fine spatial resolution (appropriate to capture variability in both natural patterning and in human interventions that drive change) and linked to models for attribution of changes and future projections needed for planning. Validation frameworks are needed to assess accuracy at a variety of scales against high-quality reference data (e.g. Menlove and Healey 2020). In addition, clear protocols for the collection of measurements to use with biomass mapping (including clearly defining how to incorporate measurement uncertainties into scaling efforts) need to be provided to influence how large-scale measurements efforts (e.g. FIA) develop the essential datasets used for biomass accounting. While the goals for MRV are defensible, broad-scale biomass estimates and novel approaches should continue to be explored to help understand the influence of fine-scale variation on scaling efforts and how fine-scale patterns influence larger scale, but more coarse resolution, mapping efforts. These include the continued use of detailed in-situ measurements and validation of allometry using lidar tools (e.g. Xu et al 2018), as well as the incorporation of other novel platforms such as unoccupied aerial systems (Alonzo et al 2020) to provide both fine-scale, targeted (e.g. using UAS to fill in critical observation gaps) and spatially-extensive information on plant structure and biomass to inform scaling approaches or validate other products. Similarly, the temporal cadence of biomass mapping should increase to improve our understanding of biomass change, thus allowing better attribution of changes to drivers. At the same time, a tighter coupling of efforts between biomass and disturbance mapping should be explored as this will be essential for increasing the temporal frequency of biomass estimates and also ensure landscape changes are correctly accounted for. The incorporation of more model-driven biomass upscaling should also be considered, which would allow for the fusion of observation data and process models to infer biomass state based on measurements and mechanistic modeling of plant growth (Fer et al 2018). In general, these improved maps of change must continue to be better integrated into process-based models and with tower-based flux observations. Throughout all of these steps, clearer definitions of product uncertainties are needed, particularly relative to the types of factors included within uncertainty estimates for different products.

Flux activities in CMS Phase 2 focused on developing new methods for estimating carbon fluxes, characterizing key sources of uncertainty, and deploying new measurements to support flux evaluation. CMS flux products represent both bottom-up (e.g. process models, inventories combined with land surface remote sensing data) and top-down (atmospheric inversion) approaches. CMS investigators have developed multiple observationally constrained bottom-up estimates of the major flows of carbon including fossil fuel (Asefi et al 2014, Gately and Hutyra 2017, Oda et al 2018) and fire emissions (Andela et al 2019) and land (Hardiman et al 2017, Weir et al 2021a) and ocean (Brix et al 2015, Gregg et al 2017, Carroll et al 2020) carbon flux. They have also improved inverse methods that use atmospheric observations to infer surface sources and sinks with a focus on attributing net flux to underlying processes at both regional (Fischer et al 2017, Graven et al 2018, Hu et al 2019) and global (Liu et al 2017, Wang et al 2018) scales (figure 10). Phase 2 projects also improved characterization of major sources of uncertainty including atmospheric transport model errors (Brophy et al 2019, Diaz-Isaac et al 2019, Butler et al 2020) and deployed both ground-based (Fischer et al 2017, Graven et al 2018) and aircraft measurements (Wolfe et al 2018, Hannun et al 2020) to validate local to regional scale flux estimates. CMS Phase 2 also included the first project designed to evaluate the consistency and completeness of CMS products for estimating global and regional carbon budgets.

Zoom In Zoom Out Reset image size

CMS projects yielded a diverse set of results that represent an increased understanding of both anthropogenic and natural flux processes. These findings show that global inverse models that incorporate satellite CO₂ observations are able to reduce flux uncertainty and quantify the relative distribution of regional net fluxes (Liu et al 2014, Wang et al 2020). When combined with other types of observations (e.g. shorter-lived trace gasses, indicators of vegetation productivity), such models can provide additional information about specific flux processes including respiration, gross primary production, and fire emissions (e.g. Liu et al 2017, Konings et al 2019, Magney et al 2019). Multiple studies focused on quantifying the impact of El Nino events on atmospheric CO₂ and on understanding the regional mechanisms that control these changes in the tropics (Bowman et al 2017, Chen et al 2017b, Liu et al 2017) and midlatitudes (Hu et al 2019). CMS Phase 2 projects also demonstrated the ability to constrain and validate anthropogenic emissions estimates on regional scales using a combination of in situ measurements and remote sensing data (Chen et al 2016, Fischer et al 2017, Graven et al 2018, Sargent et al 2018). During the COVID-19 pandemic, global data assimilation systems were able to detect the small decrease in global fossil fuel emissions and to evaluate independent country-level estimates of emissions changes (Weir et al 2021b). Bottom-up datasets also yielded new insights into the key processes and trends driving flux changes across land (Chen et al 2017a, Andela et al 2019, Fu et al 2019, Bloom et al 2020) and ocean ecosystems (Gregg et al 2017, Carroll et al 2020). This progress, particularly at global scales, supports the capacity building potential of CMS. Many global flux products provide detailed information for non-Annex I countries that may improve national level reporting mandated by the Paris climate accord.

In addition to demonstrating understanding of processes, many CMS projects focused on highlighting the observational and modeling priorities that are needed to further refine flux estimates. The availability of high-resolution, low-latency global flux and emissions datasets remains an unmet need in the atmospheric carbon monitoring community (Oda et al 2018, Weir et al 2021a). Though uncertainty quantification (UQ) methods have improved throughout CMS (e.g. Bousserez et al 2015, Oda et al 2019), atmospheric transport remains a sizable source of uncertainty for inverse estimates at both regional and global scales (Brophy et al 2019, Diaz-Isaac et al 2019, Butler et al 2020). Many critical carbon cycle processes are poorly captured by climate models (Andela et al 2019, Fu et al 2019) despite research advances in recent years.

While Phase 2 CMS flux projects reflected progress in quantifying both human emissions and natural fluxes, many challenges remain. Large differences in bottom-up flux estimates persist despite incorporation of satellite data in land and ocean models (Ott et al 2015). There is a lack of independent atmospheric data at appropriate spatio–temporal scales for evaluating fluxes (Wolfe et al 2018), especially in the tropics. Though substantial improvements have been made over the past decade, satellite bias and coverage gaps limit both the accuracy (Basu et al 2018) and spatial scale of top-down global fluxes (Wang et al 2018). Regional top-down estimates, which can help improve the spatial scale of flux estimates in key regions like the United States where denser observations are available, are complicated by a lack of boundary condition information. Uncertainty estimates, a required element for all CMS products, are often difficult to interpret because of differences in the quantification method used by various projects. In addition to challenges that influence product quality, flux projects have also had more difficulty connecting with stakeholders than some other areas within CMS. Relevant factors include technical roadblocks like a lack of familiarity with scientific data formats among stakeholders, incompatibility between geographical boundaries and spatial resolution of models, and differences in scientific and policy-relevant carbon accounting definitions. In addition, most CMS research products have long latencies and irregular update schedules, which also limits relevance to stakeholder communities.

Progress made across individual CMS projects provides the opportunity for substantial advances in flux estimation in coming years. Quasi-operational flux modeling systems are reducing latency in flux estimates to support research and stakeholder communities. Better integration of regional and global modeling activities could help support more robust and reliable flux estimates across scales with improved characterization of uncertainties. Net flux estimates may also be improved through incorporation of observations from multiple satellites, allowing either a longer period of record for better understanding of interannual variability and trends (e.g. GOSAT and OCO-2) and/or denser observations in key regions (e.g. OCO-2 and GeoCarb). Integration of multiple types of observations that simultaneously constrain flux, stocks, and disturbance into dynamical data assimilation systems can improve consistency of flux products and improve their ability to yield policy relevant information. New observations that support evaluation of regional flux evaluation can improve confidence in flux estimates. Cross-project coordination is also critical in addressing several known gaps. Improving consistency in uncertainty methods of bottom-up CMS flux datasets would allow them to better inform top-down estimates and to improve quantification of top-down uncertainties. CMS also needs to work toward an integrated approach for engaging potential flux stakeholders, particularly at national and global scales where it is important for NASA and other government agencies to provide consistent, regularly updated, high-quality flux information.

CMS methane activities in Phase 2 focused on enabling the use of satellite observations of atmospheric methane to quantify and attribute emissions on local, regional, and global scales. The work involved strong collaborations with climate policy stakeholders including the California Air Resources Board (CARB), the U.S. Environmental Protection Agency (EPA), Environment and Climate Change Canada, the Mexican National Institute of Ecology and Climate Change, the Integrated Global Greenhouse Gas Information System, and the Rocky Mountain Institute.

A major component of CMS methane activities has been the exploitation of GOSAT satellite observations (2011–present) to quantify methane emissions, the methane sink (mainly tropospheric OH), and their trends on the global scale (Maasakkers et al 2019, Zhang et al 2021, Lu et al 2021). Other GOSAT analyses have focused on North America (Turner et al 2015, 2016, Sheng et al 2018a, Maasakkers et al 2021), and on tropical wetlands (Parker et al 2018). CMS investigators have conducted atmospheric measurement campaigns to quantify methane emissions on urban to regional scales and support satellite observations. This has involved evaluation of commercial solar-viewing shortwave infrared (SWIR) spectrometers with the TCCON satellite validation standard (Hedelius et al 2016, 2017), and application of these spectrometers to study emissions in Boston (McKain et al 2015) and from dairies (Viatte et al 2017). It has also involved inversion of ground-based network observations to quantify emissions in the Los Angeles Basin (Yadav et al 2019), inversion of NASA SEAC⁴RS aircraft observations to quantify emissions in the Southeast U.S. (Sheng et al 2018b), and aircraft campaigns using the AVIRIS-NG imaging spectrometer to map methane plumes from point sources and infer emissions (Duren et al 2019, Thorpe et al 2020, Cusworth et al 2020a). Combined analysis of satellite, aircraft, and surface observations over the Los Angeles Basin demonstrated the power of the integrated observing system (Cusworth et al 2020b), as did the combined analysis of multiple satellite data streams to quantify emissions from a gas well blowout (Cusworth et al 2021a).

Inversions of satellite data to infer methane emissions require high-quality, spatially resolved emission inventories to serve as prior estimates. Developing such inventories has been a priority for CMS and has focused globally on wetlands (Tian et al 2015a, Bloom et al 2017, Poulter et al 2017, Zhang et al 2017, Treat et al 2018), rice (Zhang et al 2016), livestock (Wolf et al 2017), and fossil fuels (Scarpelli et al 2020a). It has also involved spatial disaggregation of national inventories for the U.S. (Maasakkers et al 2016) and Mexico (Scarpelli et al 2020b). A detailed map of methane-emitting infrastructure has been developed for the Los Angeles Basin to guide atmospheric measurements (Carranza et al 2018).

Another priority for CMS has been to enable the next generation of satellite observations of methane. The new high-density TROPOMI satellite observations have been used to infer methane emissions on the ∼10 km scale (Varon et al 2019, Zhang et al 2020, Shen et al 2021). Studies have determined the combined value of thermal and shortwave IR (TIR + SWIR) measurements to resolve lower-tropospheric methane (Worden et al 2015) and infer trends in tropospheric OH (Y Zhang et al 2018). CMS has contributed to specifications for future geostationary instruments such as GeoCARB and GeoFTS to constrain methane emissions from the regional scale down to the scale of individual facilities (Cusworth et al 2018, Turner et al 2018, Sheng et al 2018b). CMS has been a mechanism for developing the power of GHGSat and imaging spectrometers such as PRISMA and Sentinel-2 to observe point sources of methane (Cusworth et al 2019, 2021a, Varon et al 2018, 2019, 2020, 2021).

CMS methane activities have yielded a number of important results. They have established that inundation and ecosystem respiration are major drivers of variability and trends in methane emissions from wetlands, as shown by GOSAT observations (Parker et al 2018) and biogeochemical models (Bloom et al 2017, Poulter et al 2017, Zhang et al 2017). Wetland models with low methane to temperature sensitivity generally agree better with satellite observations (Ma et al 2021b). Through the development of spatially-resolved national inventories and their application to inversion of satellite observations, CMS has shown that emissions from the oil sector are underestimated by a factor of 2 in the national inventories for the U.S. (Maasakkers et al 2021) and Mexico (Shen et al 2021). On a point source level, CMS investigators have achieved better understanding of the role of 'super-emitters' in contributing disproportionately to state and national methane emissions (Duren et al 2019, Varon et al 2019, 2021, Cusworth et al 2021b) and have identified the underlying processes for landfills (Cusworth et al 2020a) and gas storage facilities (Thorpe et al 2020). Results from these point-source CMS studies were shared directly with industry and agency stakeholders who conducted follow-up site visits with surface sensors to verify and further refine source attribution. In roughly 50% of those cases, facility operators indicated that the confirmed super-emitters were fixable and a subset of those were repaired. The net magnitude of mitigated methane emissions from those studies is currently being reviewed.

Better understanding has also been achieved of the drivers of global methane trends over the past decade, as illustrated by figure 11. Inversion of GOSAT data suggests that tropical livestock could be a major driver for the decadal trend, while wetlands may contribute to the acceleration of the trend since 2016 (Zhang et al 2021). Changes in OH concentrations could greatly contribute to the interannual variability of methane (Turner et al 2017) though probably not to the decadal trend (Zhang et al 2021).

Zoom In Zoom Out Reset image size

Major gaps remain in our understanding of methane emissions. There is considerable uncertainty in quantifying wetland emissions and how these emissions contribute to the global methane trend. This includes uncertainties in the carbon respiration rate (Bloom et al 2017), inundation dynamics (Parker et al 2018), and contributions from the non-growing season (Treat et al 2018) Improved biogeochemical models for methane emission from wetlands are needed (Xu et al 2016). The factors contributing to the global trend in atmospheric methane over the past decade are also still open to debate.

The next several years offer considerable opportunity for improving our ability to use satellite data for quantifying methane emissions on all scales (Jacob et al 2016). Planet and Landsat products can enable better understanding of land surface characteristics to improve our ability to quantify wetland methane emissions. TROPOMI observations should be transformative for improving the capability of inversions of atmospheric methane to constrain the methane budget on global to regional scales. GeoCARB geostationary observations over the Americas should enable better quantification of emissions from South American wetlands. The emerging constellation of satellite instruments able to quantify point sources (GHGSat, PRISMA, S-2, EnMAP, EMIT, MethaneSat, Carbon Mapper) opens up the possibility for operational leak detection and sustained emissions quantification at the scale of individual facilities. The ability to constrain global OH trends from satellite observations of methane should improve through exploitation of TROPOMI and CrIS data.

CMS projects have contributed to various efforts to better characterize the role of oceans and coastal interfaces in global carbon cycling. Some of this work has been summarized in the Second State of the Carbon Cycle Report (USGCRP 2018) as well as published papers on the North American coastal carbon cycle (Fennel et al 2019) and the carbon budget of Eastern North America (Najjar et al 2018). Additionally, Benway et al (2016) produced a synthesis of current information about coastal carbon budgets and provided a series of recommendations for future research.

Some CMS projects have considered the nature of coastal margins as boundaries to the continental carbon cycle. Related to this is improved observation and modeling of lateral transport of terrestrial carbon and nutrients into the watershed and ultimately to coasts (Liu et al 2013, Lohrenz et al 2013, Xue et al 2013, Tao et al 2014, Ren et al 2015, 2016, Tian et al 2015b, 2015c, 2020; figure 12). A key to understanding this important term is how land use and land cover along with other drivers such as human activity and climate-related changes affect lateral transport processes.

Zoom In Zoom Out Reset image size

Considerable focus was given to the potentially large reservoirs of carbon biomass undergoing substantial change in sensitive coastal ecosystems. Both mangrove and tidal wetland ecosystems are critical coastal buffer zones and are undergoing rapid changes. CMS efforts include studies of mangroves (Lee and Fatoyinbo 2015, Lagomasino et al 2015, 2016, Fatoyinbo et al 2018, Simard et al 2019) and tidal wetlands (Hopkinson et al 2012, Morris et al 2016, Hinson et al 2017, Byrd et al 2018, Holmquist et al 2018a, 2018b). Mangroves account for a substantial amount of carbon biomass and various CMS efforts examined Lidar-based algorithms for refining aboveground biomass estimates (Lagomasino et al 2015, Fatoyinbo et al 2018, Simard et al 2019). Tidal wetlands also represent an important coastal carbon reservoir (Hopkinson et al 2012, Byrd et al 2018, Windham-Myers et al 2018, Holmquist et al 2018a, 2018b). CMS projects have also evaluated sources of methane and nitrous oxide emissions. This included examination of tidal wetlands and aquaculture sites (Poulter et al 2017, Zhang et al 2017, EPA 2019), as well as more comprehensive assessments over North America and globally (Xu et al 2012, Tian et al 2012b, 2015a, 2016).

CMS efforts have greatly expanded the information about oceans and coastal interfaces and their key role in global carbon cycling. Efforts such as the Second State of the Carbon Cycle Report (USGCRP 2018, Fennel et al 2019) have provided an initial assessment of air-sea and land-ocean fluxes for North America. CMS investigators contributed to chapters covering Inland Waters (Ch 14), Tidal Wetlands and Estuaries (Ch 15), Coastal Ocean and Continental Shelves (Ch 16), and Biogeochemical Effects of Rising Atmospheric Carbon Dioxide (Ch 17). Coastal margins in North America act as a net sink of carbon. The North American Exclusive Economic Zone is estimated to be a net sink for carbon on the order of 160 ± 80 Tg C y⁻¹ (Fennel et al 2019). The estimated carbon input from land is 106 ± 30 Tg C y⁻¹. A global inventory of carbon dioxide fluxes in coastal margins influenced by large rivers (Cai et al 2013) found that estuaries were a net source of CO₂ due to metabolism of terrestrial carbon entering these systems. In contrast, a global assessment of coastal areas influenced by large river plumes found that they were a net sink for CO₂. Considerable focus was given to the northern Gulf of Mexico and the Mississippi River plume, including extensive ship-based mapping and buoy measurements of pCO₂ (Guo et al 2012, Cai et al 2013, Huang et al 2013, 2015a, 2015b) as well as model simulations using a coupled physical-biogeochemical model (Xue et al 2014) and satellite-derived estimation of pCO₂ and air-sea flux of CO₂ (Lohrenz et al 2018). The biological dynamics influencing carbon dynamics were considered by Chakraborty et al (Chakraborty and Lohrenz 2015, Chakraborty et al 2017), who explored the relationships between phytoplankton community composition and physiological properties of populations and relationships to environmental conditions. Other efforts have examined carbon properties in both the Gulf of Mexico and the Atlantic coast (Wang et al 2013, Najjar et al 2018).

Impacts of human and climate-related forcing on terrestrial watersheds affect export of carbon and other materials to the coastal margins. Changes in land cover and land use along with climate-related factors were determined to impact lateral movement of carbon and other materials through the watershed. Work by Tian et al and others utilized the Dynamic Land Ecosystem Model along with extensive satellite and ground-based observations to examine historical trends and patterns as well as modeled simulations of future scenarios under differing climate and atmospheric carbon forcing (Tian et al 2012a, 2015b, 2015c, Liu et al 2013, Lohrenz et al 2013, Xue et al 2013, Tao et al 2014, Ren et al 2015, 2016, Yu et al 2018). Urban land conversion (Zhang et al 2012, 2014), livestock manure nutrient production (Yang et al 2016), crop production (Lu et al 2018), forest disturbance (Chen et al 2013), drought (Chen et al 2012), and large fires (Yang et al 2015) were considered as significant factors in terrestrial carbon dynamics. More recent work has addressed factors affecting nitrogen loading to the Mississippi River basin (Lu et al 2020, Tian et al 2020). An additional carbon-related concern for coastal ecosystems is ocean acidification and its implications for coastal ecology and carbon cycling. Increasing ocean acidification may also reduce the buffering capabilities of coastal waters diminishing their capacity as carbon sinks (Wang et al 2013, Salisbury et al 2015, Fennel et al 2019).

Consideration by CMS projects was given to productivity and influencing factors in the Upper Great Lakes (Yousef et al 2014, Fahnenstiel et al 2016). Information about primary productivity in the upper Great Lakes was significantly expanded through CMS efforts. Primary production in Lake Michigan was found to decrease over the 1998–2010 period, largely attributed to a decline in chlorophyll biomass as a result of the quagga mussel activity (Yousef et al 2014). A comprehensive assessment of primary production was done for the upper Great Lakes for the period 2010–2013 (Fahnenstiel et al 2016), thereby providing a baseline for future study.

Despite progress in understanding marine, freshwater and wet carbon dynamics, scientific uncertainties and priorities remain. These include continued refinement of terrestrial ecosystem models needed to reduce uncertainties in carbon flux quantification. Improvements are needed in parameterization in areas such as in-stream organic production and transformation, as well as hydrological processes including effects of dams and rivers require further study. Constraining estimates of contributions by coastal margins and inland waters to continental and global carbon budgets requires improved assessments of exchange fluxes and the associated seasonal and interannual variability. While the amount of information has grown substantially, uncertainties represent a challenge because carbon is exchanged across multiple interfaces (land-coastal ocean, coastal ocean-open ocean, ocean-benthic, ocean-atmosphere). There is a need for improved characterization of spatial patterns and relationships to forcing, as well as assessments of these exchange fluxes and the associated seasonal and interannual variability. How coastal margins will be affected by changing climate forcing, sea level rise, and human impacts remains an important question. Tidal wetlands and mangroves represent large reservoirs of carbon and are important as potential sources of carbon dioxide and other greenhouse gases. Contributions of coastal margins to methane budgets are still poorly understood. Also, contributions of tidal wetlands and marshes to nitrous oxide budgets need to be examined. Tidal wetlands and mangroves have been identified as important sinks and sources, but still associated with significant uncertainties. Continental to global assessments are needed, but current knowledge is very limited regarding coastal margins as sources of methane or nitrous oxide.

Although considerable progress has been made in understanding contributions of coastal margins to global carbon, these contributions are still not well constrained, particularly with regard to seasonal and spatial patterns. There is a need for coordinated modeling and observations across the land-ocean continuum, including continued and expanded time-series observations to better discern temporal variability. Approaches should integrate observations, modeling and stakeholder needs (Fennel et al 2019). There is also a need for better quantification of methane and nitrous oxide fluxes in different coastal types (e.g. river, estuary, tidal wetland, mangrove, etc). Continued advancement of novel approaches such as remote sensing techniques for characterizing changes in carbon biomass and fluxes in coastal environments will be important for reducing uncertainty in these variables. Uncertainties in feedbacks need also to be addressed—how will future changes including climate, human impacts, and increasing atmospheric CO₂ affect efficiency of ocean uptake? Improved understanding of how coastal margins and associated carbon dynamics will change and what factors, including human activity, will influence that change will be important considerations for policy and decision-making.

During Phase 2, stakeholder engagement accelerated at the federal, state, and local level, and with international partners. The CMS program funds basic and applied research that is created while engaging with a user community in a way that bridges carbon science and user communities to better serve societal needs (Brown et al 2020; figure 13).

Zoom In Zoom Out Reset image size

By connecting carbon cycle science research to stakeholders who use the data in their decision making, NASA CMS contributes to understanding and meeting the needs of the climate data user community. For example, Phase 2 produced novel data products on forest carbon product designed to meet state needs beginning in Maryland (Huang et al 2015, Hurtt et al 2019, Lamb et al 2021a), and later expanding to the multi-state Regional Greenhouse Gas Initiative (RGGI) region (Tang et al 2021, Ma et al 2021a, Lamb et al 2021b). New efforts on mapping and monitoring methane across California and other regions (Carranza et al 2018) engaged the CARB to identify and remove methane 'hotspots' across urban areas. Other stakeholders across a wide variety of CMS projects include Northwest Management Inc. The Nature Conservancy, Worcester Tree Initiative, Maryland Valleys Planning Council, World Resources Institute, Blue Source, Chesapeake Conservancy, U.S. Forest Service, the Environmental Defense Fund, the Sonoma County Agricultural Preservation and Open Space District, among many others.

Ensuring that critical information on carbon is used in day-to-day decisions and policy making by stakeholders such as governments, businesses, and institutions requires early engagement and frequent communication between the user and the producer of the information (Cash et al 2006). The CMS program allows scientists to build mature relationships with stakeholders that result in greater success in moving a carbon cycle product from conceptualization to actual use within a decision-making context. It is only through funding a single individual who provides consistent engagement and builds relationships among and between the scientists and these institutions is CMS able to accelerate uptake of datasets in Phase 2 (Brown et al 2020).

Major gaps from Phase 2 include the need for carbon products to be repeated, updated regularly, and for CMS to provide information that describes change through time instead of a single point in time. Change products would require standardization and repeat acquisition of input data, ongoing support of computing resources, access services, and for the policy relevance of the product to be clearly acknowledged and articulated. Gaps also include a relative lack of products that describe the sensitive ecosystems of mangroves and urban forests, and products that link biomass to methane emissions. Economic studies are needed to demonstrate the value of these ecosystems and how information on them can be used in decision making in ways that may result in improved resilience and functioning (Horita et al 2017). Economic analysis of the value of information may include data documenting impact of carbon cycle uncertainty (Cooke et al 2016), and the costs and benefits of forest policy that incorporates carbon sequestration. For example, Lamb et al (2021b) describe how afforestation and reforestation in Maryland can provide both carbon sequestration potential and economic opportunity via forest carbon pricing verified through the use of geospatial products generated through the CMS program.

Recommendations include increased focus on: reconciling carbon stocks and fluxes, developing consistency across scales, quantifying movement and transport of carbon, attributing carbon emissions and sinks to respective sources, cross-sectoral accounting, and UQ (West et al 2013). More data products are needed in new regions and new ecosystems with new stakeholders supporting city, state, and national/international initiatives where demand for decision support is increasing. Capacity to use carbon data products for people who are developing policy is a critical need, as Lamb et al (2021a) demonstrated. With an improved understanding of the value of carbon and the alternative land uses, appropriate incentives and policies can be developed that increase the value of land use while sequestering more carbon.

CMS has developed a community of practice where scientists have learned how to do meaningful stakeholder engagement, the value of this engagement, and have learned through annual Science Team meetings and stakeholder workshops about applications of CMS products (Brown et al 2020). This engagement by each funded project is one of the most novel parts of the CMS program and results in rapid prototyping and discovery of new carbon cycle data and models. By emphasizing in future funding opportunities the need for proposers to identify and engage with stakeholders before proposing, more rapid uptake of new carbon cycle products can be achieved (Brown et al 2020). As Arnott et al (2020) points out, funders of science are receptive to new ways of revisiting the 'social contract' for science so that co-production of knowledge can be prioritized. Ensuring CMS scientists prioritize relationships as well as producing products and writing papers is essential.