Volume 126, Issue 13 e2021JD034545
Research Article
Free Access

Quantification of Enhancement in Atmospheric CO2 Background Due to Indian Biospheric Fluxes and Fossil Fuel Emissions

Santanu Halder

Santanu Halder

Indian Institute of Tropical Meteorology, Ministry of Earth Sciences, Pune, India

Department of Atmospheric and Space Sciences, Savitribai Phule Pune University, Pune, India

Search for more papers by this author
Yogesh K. Tiwari

Corresponding Author

Yogesh K. Tiwari

Indian Institute of Tropical Meteorology, Ministry of Earth Sciences, Pune, India

Department of Atmospheric and Space Sciences, Savitribai Phule Pune University, Pune, India

Correspondence to:

Y. K. Tiwari,

[email protected]

Search for more papers by this author
Vinu Valsala

Vinu Valsala

Indian Institute of Tropical Meteorology, Ministry of Earth Sciences, Pune, India

Department of Atmospheric and Space Sciences, Savitribai Phule Pune University, Pune, India

Search for more papers by this author
M. G. Sreeush

M. G. Sreeush

Centre for Climate Physics, Institute for Basic Science, Busan, Republic of Korea

Pusan National University, Busan, Republic of Korea

Search for more papers by this author
S. Sijikumar

S. Sijikumar

Space Physics Laboratory, Vikram Sarabhai Space Centre, Thiruvananthapuram, India

Search for more papers by this author
Rajesh Janardanan

Rajesh Janardanan

Satellite Observation Center, Center for Global Environmental Research, National Institute for Environmental Studies, Tsukuba, Japan

Search for more papers by this author
Shamil Maksyutov

Shamil Maksyutov

Satellite Observation Center, Center for Global Environmental Research, National Institute for Environmental Studies, Tsukuba, Japan

Search for more papers by this author
First published: 08 June 2021
Citations: 5

Abstract

Regional carbon emissions impact global atmospheric carbon dioxide (CO2) background concentrations. This study quantified the enhancement in the atmospheric CO2 mole fractions due to biospheric and fossil fuel fluxes from India. Sensitivity experiments using model simulations were conducted, allowing CO2 enhancement due to biospheric and fossil fuel fluxes from India to diffuse into the global atmospheric background. The areal extent of column-averaged enhancement of 0.2 ppm and above due to Indian fluxes are spread over a larger area covering the Indian subcontinent, neighboring Asian regions, and the north Indian Ocean in all four seasons. The boundary layer CO2 enhancement due to biospheric fluxes (fossil fuel fluxes) have a maximum range of −2.6 to +1.4 ppm (1.8–2.0 ppm) most time of the year. At higher altitude, the amplitudes of enhancement are reduced from ±1.8 to ±0.6 ppm as we go up from 850 to 500 hPa due to diffusion by prevailing atmospheric dynamics and convection. With the information of the areal extent of >0.2 ppm CO2 enhancement due to Indian fluxes, we have evaluated the representativeness of satellite observations (GOSAT and OCO-2) in capturing those enhancements. Both the satellite coverage show a similar number of observations (0.1 per day) during all seasons except for June to August, wherein the cloud screening eliminates almost all the satellite data over the region. Within this areal extent, the satellite XCO2 shows average anomalies of nearly ±2.0 ppm; it is a valuable piece of information because it is well above the retrieval uncertainty, yet capturing the potential enhancement due to fluxes from India. The study implies that the regions of enhancement greater than 0.2 ppm can be considered locations for surface observations representing Indian fluxes. Similarly, the region with enhancement greater than one ppm could be covered by satellites/airborne observations to discern enhancement in the atmospheric CO2 mole fractions due to Indian fluxes.

Key Points

  • Indian carbon fluxes significantly contribute to the CO2 concentrations as large as ± 2.0 ppm over India and neighboring regions throughout the year at surface level (i.e., 975 hPa)

  • Satellites are capable of representing the XCO2 anomalies up to ± 2.0 ppm over corresponding regions; however, they have no data coverage during rainy seasons

  • The study recommends both ground-based and satellite/aircraft measurements of CO2 be employed to observe the larger area of Indian CO2 enhancements

Plain Language Summary

This study calculates the seasonal enhancement of the atmospheric CO2 concentration (in ppm) due to biospheric fluxes and fossil fuel emissions from India. The regional fluxes of fossil fuel and biospheric activities contribute to the global atmospheric CO2 within a few ppm. Using a global atmospheric tracer transport model and regional fluxes, we show that the enhancement of the atmospheric CO2 due to India's biospheric fluxes and fossil fuel emission spread over a large extent over India and surrounding regions. A tangible value of ±1.8 ppm and above can be found in most parts of India and its surrounding regions during significant seasons. The areal extent grows from the boundary layer to the upper atmosphere due to corresponding atmospheric mixing and dynamics. On the surface, the enhancement due to biospheric fluxes is about −2.6 to +1.4 ppm in all four seasons, whereas the enhancement due to fossil fuel shows about 1.8–2 ppm. The extent of discernible enhancement in the global atmospheric background due to Indian fluxes helps us understand the optimum location where all observational efforts should be made to sample Indian emission scenarios from global backgrounds. This study also points out the capability in observing the enhancement of XCO2 by the satellite over India and its surroundings. Satellite observations are crucial where the observations of atmospheric CO2 are limited to constrain the surface fluxes in the inverse modeling study. Satellite measurements can capture the enhancement except June to August when the observations are screened out for cloud contamination. Therefore, surface observations are necessary to capture the enhancement of Indian emissions in the global atmospheric CO2 background in all seasons.

1 Introduction

The global carbon cycle plays an essential role in maintaining the earth's environment warm and hospitable (Reichstein et al., 2013). The atmospheric carbon dioxide (CO2) has rapidly increased from a pre-industrial value of 278–405 ppm in the recent past (Le Quéré et al., 2018; Friedlingstein et al., 2019). It is unequivocally agreed upon that atmospheric greenhouse gases (GHGs) such as CO2, which have a global warming potential, accelerate the contemporary warming of the earth's atmosphere (Etminan et al., 2016; IPCC, 2014). The increased resource demands escalated the anthropogenic emissions of CO2 are contributed by fossil fuel burning, deforestation, cement production, and modern agricultural practices (Leung et al., 2014). While the global fossil fuel burning and cement production emit 9.4 ± 0.5 Pg Cyr−1, the terrestrial biosphere and oceans sequestrate 3 ± 0.8 Pg Cyr−1 and 2.4 ± 0.5 Pg Cyr−1, respectively, thereby offsetting the net build-up of contemporary CO2 concentration in the atmosphere (Friedlingstein et al., 2019). Several studies have attempted to quantify the effect of natural source/sink and anthropogenic emission of atmospheric CO2 and its natural variability (Keenan et al., 2016; Tagesson et al., 2020).

As per the Global Carbon Project (GCP), India is the fourth highest CO2 emitting nation (7% of total global CO2) in the last decade (Friedlingstein et al., 2019; Le Quéré et al., 2018). India's energy demand has increased, resulting in an increased coal usage (Dwivedi, 2017) and large-scale land-use change. In addition to these anthropogenic causes, the monsoonal processes have a significant impact on the natural carbon cycle variability over the South Asian region (Fadnavis et al., 2018; Kumar et al., 2014; Revadekar et al., 20122013; Tiwari, Revadekar, & Kumar, 20132014; Tiwari, Vellore et al., 2014; Valsala et al., 2013). This study investigates the enhancement of atmospheric CO2 over India and surrounding regions due to India's emissions mentioned above and biospheric fluxes.

The distribution of CO2 concentration over the Indian and surrounding oceanic regions during 2009–2012 was studied by Nalini et al. (2018). They have examined CO2 observations from satellite, surface measurements, and concentrations derived from CarbonTracker (Peters et al., 2007) to understand the spatio-temporal variability of atmospheric CO2 over India. They found an increase of atmospheric CO2 during pre-monsoon (March-May) due to temperature-related biospheric respiration and a decrease during post-monsoon (June-September) due to vegetation growth. On the other hand, the satellite-based study about enhancing atmospheric CO2 mole fraction over the large point source regions over India showed an enhancement as significant as 0.6–2 ppm over and above the background value as detectable by satellite measurements (Janardanan et al., 2016). Although their study highlighted the enhancement of atmospheric mole fractions due to Indian fluxes over a limited region, a comprehensive understanding of the enhancement and its areal extent in and around the Indian subcontinent is still lacking.

There is a rationale behind examining the enhancement of atmospheric mole fractions of CO2 due to regional emissions, especially in the inverse estimations of sources and sinks (Cervarich et al., 2016; Sreeush et al., 2020). Inverse modeling constrains the surface sources and sinks using atmospheric tracer transport functions and observations of atmospheric CO2 dry-air mole fractions. Therefore, the spread of discernible enhancement due to fluxes from a particular region in the atmospheric background is crucial. Observations (either ground-based or instruments mounted onboard satellite) should capture this enhancement so that it can offer a better constrain in the inverse estimations. The satellites’ efficiency in observing enhancement of the Indian sources and sinks mixed in the global atmospheric CO2 is crucial too, given geospatial limitations such as cloud cover hindering the direct satellite observations over India and surroundings (Janardanan et al., 2016; Jiang & Yung, 2019). In this context, the present study examines the maximum areal extent over which the Indian CO2 enhancement is discernible in a particular measurement.

A complementary analysis of the above subject is available in Nalini et al. (2019). In that study, the authors address the question—what are the optimal locations over India where a maximum probable enhancement of Indian CO2 sources and sinks can be observed? Due to the seasonal heterogeneity (both fluxes and atmospheric dynamics) offered by the South Asian monsoon meteorology, the authors identified the best suitable sites over the Indian region to measure atmospheric CO2 mole fractions for four seasons such as summer (June–September), winter (December–February), pre-monsoon (March–May) and post-monsoon (October–November). On the contrary, the regions in and around the Indian subcontinent where the enhancement of Indian emissions are mixed to the global atmospheric background are not readily available in previous studies.

Satellite observations such as greenhouse gases observing satellite (GOSAT: Kuze et al., 2009) and orbiting carbon observatory-2 (OCO-2: Crisp et al., 2004; Jiang & Yung, 2019; Miller et al., 2007) provide global measurements of column-averaged CO2 dry-air mole fraction data sets for a better scientific understanding of regional carbon cycle processes and carbon budgets. However, satellite observations are known to have substantial data gaps due to geophysical limitations like the presence of clouds and aerosols in the atmosphere (Alkhaled et al., 2008; Barkley et al., 2006; Bösch et al., 2006; Buchwitz et al., 2005; Byrne et al., 2017; Engelen & McNally, 2005; Miller et al., 2007; Tiwari et al., 2006). Basu et al. (2013) have used GOSAT XCO2 retrievals to constrain regional flux. A recent study by Park et al. (2020) has shown XCO2 enhancement over megacity, that is, Seoul. Hakkarainen et al. (2019) have identified polluted locations by calculating XCO2 anomaly using OCO-2 satellite observations. Satellite XCO2 can provide knowledge about surface fluxes through inverse study where the surface atmospheric observations are limited, such as in India. The present study also verifies, “How well satellites observe a potential enhancement in the atmospheric CO2 due to Indian sources and sinks for all four seasons”. This information is vital because only those observations will have substantial implications in constraining the Indian CO2 sources and sinks in an inverse modeling framework.

Identifying these potential gap areas, we have addressed the following key questions in this study. (a) What is the spatial extent of enhancement of atmospheric CO2 due to the Indian fluxes is found and its seasonal variability? (b) Whether the existing ground-based stations and the satellite coverage adequate to sample this potential zone. Answering these questions provides an evaluation of how well the existing satellites observe the CO2 enhancement over India and gives insight into how the future satellite missions will represent better the CO2 enhancement based upon their CO2 retrieval capabilities. The rest of the paper is arranged as follows. Section 2 describes the model, data, and methodology. Section 3 presents the significant results, followed by a discussion and conclusion in Section 4.

2 Model, Data and Methodology

This study utilized a suite of atmospheric tracer transport models, ground-based measurements, and satellite-based observations of atmospheric CO2 as follows.

2.1 Model

We have employed a global atmospheric tracer transport model, the National Institute for Environmental Studies-Transport Model (NIES-TM) (Belikov et al., 2011; Belikov, Maksyutov, Krol, et al., 2013; Belikov, Maksyutov, Sherlock, et al., 2013; Maksyutov et al., 2008) for atmospheric CO2 simulations. NIES-TM is an off-line transport model, which solves the tracer transport equation in polar coordinates, namely,
urn:x-wiley:2169897X:media:jgrd57124:jgrd57124-math-0001(1)
where urn:x-wiley:2169897X:media:jgrd57124:jgrd57124-math-0002 denotes dry air mole fraction (volume) of k-th tracer, urn:x-wiley:2169897X:media:jgrd57124:jgrd57124-math-0003 is the vector combination of zonal and meridional wind component (u,v) taken from global reanalysis winds JRA-25-JCDAS (Onogi et al., 2007) prepared by the Japan Meteorological Agency. urn:x-wiley:2169897X:media:jgrd57124:jgrd57124-math-0004 stands for the velocity vector of the vertical wind component in urn:x-wiley:2169897X:media:jgrd57124:jgrd57124-math-0005coordinate with positive value indicating the downward component of the vertical wind. The model horizontal resolution is 2.5° × 2.5° latitude by longitude and 32 vertical terrain-following hybrid theta-sigma coordinate levels. The model calculates the vertical distribution of tracers with the help of the Kuo-type scheme (Grell et al., 1994). The boundary layer height is prescribed by 3-hourly varying PBL height from ECMWF (Simmons, 2006; Simmons et al., 2007).

NIES-TM uses monthly surface CO2 flux inputs as the biosphere, fossil fuel, biomass burning, and ocean (gC m−2 day−1) at 1° × 1° (latitude by longitude), and they were obtained from CarbonTracker (i.e., CT2019B; Jacobson et al., 2020). The model simulations were carried out from 2000 to 2018 for the validation purpose and from 2014–2017 for various experiments, as details below.

2.2 Data

2.2.1 Surface Observations

Atmospheric CO2 mixing ratio data sets used in this study were from four observation locations, (1) Mauna Loa (MLO: 19.47°N, 155.59°W; Zhao & Tans, 2006), (2) Seychelles (SEY: 4.69°S, 55.17°E; Ballantyne et al., 2012), (3) Cape Rama (CRI: 15.08°N, 73.92°E; Tiwari et al., 2011; Tiwari, Vellore et al., 2014), and (4) Sinhagad (SNG: 18.35°N, 73.75°E; Tiwari et al., 2014). CRI and SEY are marine observational sites with an elevation of 60 and 3 m above mean sea level, respectively. MLO and SNG represent high altitude stations with an elevation of 3397 and 1300 m above mean sea level, respectively. CRI and SNG are the Indian atmospheric CO2 observation sites located at the western boundary of peninsular India. CRI is a coastal site where prevailing wind blows from the Arabian Sea during summer monsoon months (i.e., June–September) and from the continental region during winter months (i.e., December–February).

Weekly flask samples collected at CRI were analyzed at CSIRO Australia (Francey et al., 2003; Tiwari et al., 2011), and SNG flask samples were analyzed at IITM Pune, India (Tiwari et al., 2014). Details of SNG measurements and their validation have been reported in Tiwari et al. (20112014) and Kumar et al. (2014). The monthly continuous atmospheric CO2 observations were selected from MLO (2001–2013), CRI (1993–2012), SNG (2011–2018), and SEY (2001–2015) to validate the model used in this study. Atmospheric CO2 observation at MLO, SEY, and CRI are available from the World Data Centre for Greenhouse Gasses. The data sources are provided in the data availability section.

2.2.2 Satellite CO2 Measurements

CO2 retrievals from satellites such as GOSAT (Yokota et al., 2009) and OCO-2 (Crisp et al., 2004; O’Dell et al., 2012) were utilized in this study to examine how well the satellite captures potential enhancements of Indian emissions mixed in the global atmospheric background CO2.

GOSAT provides the column-averaged dry-air mole fraction of atmospheric CO2 (XCO2) via Thermal and Near Infrared Sensor for Carbon Observation (TANSO) (1.56–2.08urn:x-wiley:2169897X:media:jgrd57124:jgrd57124-math-0006m) with a repeat cycle of three days in a sun-synchronous orbit. GOSAT uses Fourier-Transform Spectrometer (FTS) sensor with a 2-axis scanner to measure XCO2, which is sensitive to the lower troposphere and provides more realistic CO2 observations (Kuze et al., 2009). The ACOS-GOSAT v9 data available for the period from January 2016 to December 2016 were used as a case study. GOSAT products also provide the simulated 3-dimensional distribution of atmospheric CO2 concentration. The horizontal resolutions are 2.5x2.5, with 17 vertical levels (GOSAT L4B v2.05), and are utilized for the model validations.

OCO-2 provides a column-averaged dry-air mole fraction of atmospheric CO2 with a swath of 10km in a sun-synchronous orbit with a sampling frequency of 0.333s (Crisp et al., 2004; Frankenberg et al., 2015). OCO-2 has a 3-channel imaging grating spectrometer, sensitive to high resolution reflected radiation in oxygen and CO2 bands (Byrne et al., 2017). Nadir sounding gives column-averaged atmospheric CO2 over land, even in partially cloudy conditions. On the other hand, glint-sounding observations provide a more promising profile over the ocean. The OCO-2 v9 data available for the period from January 2016 to December 2016 are used as a case study. Only those observations with quality flag zero implying high-quality data (Patra et al., 2020) were used. All are data sources are listed in the Data Availability section.

2.3 Methodology

2.3.1 Model Experiments

The following model experiments were conducted to understand the CO2 enhancement in the global atmospheric background due to the biospheric fluxes and fossil fuel emissions from India (from now on referred to as “enhancement”). The NIES-TM is run for three years between 2014 and 2017 by prescribing fluxes only over India in all the years. Further, the fluxes were prescribed in the model not only for the month-of-interest but also for all simulation years. For example, in the first experiment, fluxes were kept zero for months except January, and in the second experiment, except for February, and so on, untill December. Therefore, considering the two major fluxes (i.e., biosphere and fossil fuel) and 12 months (from January to December), there were a total of 24 experiments. The simulations were extended untill 2017 to complete three full years in each of the experiments, appropriately, that is, for the month-of-interest June; the simulations were from June, 2014 to May, 2017. The contribution from a forest fire and biomass burning is not considered in the present study. The choice of three years is made because it is assumed as a time scale good enough to sufficiently mix the emissions from a particular region into the entire global atmosphere. Three years is also a choice used in the TransCom experiments to generate the model enhancement in atmospheric CO2 for a given regional emission (Baker et al., 2006; Gurney et al., 2004).

The “enhancement” is represented as the average of 6-hourly model concentrations derived out of three months stretching from (month-of-interest – 1) to (month-of-interest + 1) and from the third year of the simulation. We considered only those concentrations that stand above ±1 standard deviation (σ) from the mean (of the above three months) corresponding to the respective model grids. Finally, we calculated the seasonal mean representing the “enhancement” for seasons such as December to February (DJF), March to May (MAM), June to August (JJA), and September to November (SON). In figures, however, we restricted the lower bound of the “enhancement” to 0.1 ppm because the measurable atmospheric CO2 concentration's accuracy is 0.1 ppm for the flask measurements (Chen et al., 2010; Winderlich et al., 2010; Yver Kwok et al., 2015). However, the biases in the satellite XCO2 can grow well above one ppm (Araki et al., 2010; Inoue et al., 2013; O’Dell et al., 2012), and in the model, it could be even more (Rödenbeck et al., 2003).

2.3.2 Satellite Data Processing

The quality flagging by cloud cover and aerosols in the satellite products pose challenges in retrieving XCO2 over regions such as India in light of the persistent seasonal monsoons and aerosol loadings (Tiwari et al., 2014; Sanap & Pandithurai, 2015). To understand how well the present satellites can capture the “enhancement” in and around India, we calculated the number of observations per day available by GOSAT and OCO-2 over this region from their Level-2 retrieval for one full year (i.e., 2016) as a case study. At first, we counted the number of observations within every 2.5° × 2.5° grid cell (to match with our model grid cells) at a daily time step. Likewise, we counted it in each season and represented it as the number of observations per day (i.e., DJF, MAM, JJA, and SON).

From the information of the region of “enhancement” over India and the surrounding regions obtained from our model experiments, we have investigated how well the satellite XCO2 captures this enhancement in its retrieval. Recent studies by Park et al. (2020) and Hakkarainen et al. (2019) have proposed two methodologies for retrieving regional emission enhancements in the satellite XCO2. We followed a similar methodology as detailed below.
  • (a)

    For each satellite track of XCO2 (by GOSAT and OCO-2), which crosses the zone of “enhancement” of 0.2 ppm as identified from our model experiment, we calculated the mean XCO2 of each track and found a deviation of satellite XCO2 from the mean. Due to the unavailability of any background XCO2 (Park et al., 2020), we took the XCO2 track-mean as our background and subtracted it from the individual XCO2 data in the track to find the satellite enhancement (Equation 2). Further, we averaged them into a 2.5o×2.5o and over three months, that is, DJF, MAM, JJA, and SON. With this calculation, the positive anomalies closely resembled the “enhancement” due to Indian fossil fuel emission as measured by the satellites, and negative anomalies could be related to the biospheric activity.

    urn:x-wiley:2169897X:media:jgrd57124:jgrd57124-math-0007(2)

  • (b)

    The study by Hakkarainen et al. (2019) examined XCO2 anomalies over a region by subtracting the daily median from corresponding daily individual XCO2 observation. Similarly, we first calculated the median for each 10° latitude band. Further, these medians were linearly interpolated at each satellite observation. Then we subtracted that median from the individual XCO2 observation (Equation 3). We averaged them into a 2.5° × 2.5°, and over three months, that is, DJF, MAM, JJA, and SON to find the satellite XCO2 anomalies due to anthropogenic and biosphere activity.

    urn:x-wiley:2169897X:media:jgrd57124:jgrd57124-math-0008(3)

3 Results and Discussions

3.1 Comparing NIES-TM Simulated CO2 and Observations

Figure 1a shows the area integrated climatological CarbonTracker (CT2019B) biosphere flux (NEE) over India calculated from 2010–2018. Being a climate-specific region with strong monsoonal regimes, the biosphere-atmosphere interactions over the Indian continent is highly complex across various spatio-temporal (e.g., sub-seasonal to seasonal) time scales (Valsala et al., 2013). The biosphere-atmosphere interaction over the Indian region possesses a strong seasonal cycle (Cervarich et al., 2016; Chan et al., 2008; Chevallier et al., 2010). NEE during summer months (JJA) acts as a source of atmospheric CO2, and when vegetation grows after summer rains (i.e., after August), it acts as a sink of atmospheric CO2 (Kumar et al., 2014; Revadekar et al., 2012).

Details are in the caption following the image

(a) Upper panel shows the climatological (constructed from 2000–2018) seasonal cycle of net ecosystem exchange (NEE) integrated over the entire India. The shade shows the uncertainty for each month. The uncertainty was calculated by standard deviation using all available grids over India using three-hourly CT2019B posterior flux. The unit is in PgC yr−1. (b)The lower panel shows the annual mean fossil fuel flux over India using CT2019B for 2016. The unit is in gC m−2 day−1.

In the case of fossil fuel emission, we noticed a relatively high emission over the Indo-Gangetic plain in the annual mean maps of CT2019B for 2016 (Figure 1b). Delhi (the capital region, 28.7°N, 78.10°E) is one of the major pollutant territory (Dhaka et al., 2020; Mukherjee et al., 2018), with emission gauging as large as 6 gC m−2 day−1 albeit with moderately high values in other major cities such as Mumbai (19.01°N, 72.87°E; Bharadwaj et al., 2017). Kort et al. (2012) has calculated XCO2 enhancement about 2.4 ± 1.2 ppm for Mumbai using GOSAT.

Figure 2 shows the fidelity of NIES-TM in simulating the atmospheric CO2 realistically. The long-term simulations compared between model and observations, the station MLO, SEY, and CRI (during monsoon season), gave a good NIES-TM performance in simulating atmospheric CO2. The seasonal amplitude (5–6 ppm) and growth rate (∼2.5 ppm) of MLO is simulated remarkably well by the model. The NIES-TM simulations were further compared with GOSAT L4B v2.05 data. The accuracy of CO2 simulations depicts the quality of surface fluxes utilized and the quality of the transport simulated. It is interesting to note that the seasonal cycle and growth rate of CO2 at SEY are reasonably comparable. Unlike MLO, the SEY seasonal cycle does not depict a clean biospheric signal; instead, due to the proximity of the equator, inter-hemispheric transport, and monsoonal reversal in circulation, all add up to the complexity in seasonal variability (Figure2b).

Details are in the caption following the image

NIES-TM comparison with various station observations. (a) Mauna Loa (MLO) (b) Seychelles (SEY) (c) Cape Rama (CRI) and (d) Sinhagad (SNG). Green, blue, and red stand for monthly CO2 concentration for NIES-TM, GOSAT, and corresponding stations. Units in ppm.

The CO2 simulations for continental stations are rather challenging (Figures 2c and 2d). Both NIES-TM and GOSAT L4B show a weaker seasonal cycle compared to the observations. Moreover, the sharp variability in the atmospheric CO2 mostly arises due to the local biosphere and fossil fuel variability, which will not capture by the model unless the fluxes resolve it. In the case of CRI, the growth rate of CO2 between the model and observations agrees reasonably well (Figure 2c). However, the representation of the seasonality in the NIES-TM and GOSAT-L4B is rather poor due to imperfect biospheric fluxes implied or a problem of site representativeness of large scale. The difficulty at CRI is common with other transport and inverse models (Maksyutov et al., 2021).

The weekly flask measurements at SNG suggest that during winter, a considerable CO2 variability (about 15 ppm) was observed at SNG and reduced by half when the clean wind blows from the ocean to the site during the summer (Tiwari et al., 2014). Tiwari et al. (2014) also pointed out that SNG is quite free from vegetation, and CO2 signal could be transported to SNG from the Indo-Gangetic plain during winter. Both NIES-TM and GOSAT-L4B could not represent the SNG observations well (Figure 2d), could also be due to problems with fluxes (both biosphere and fossil fuel) used and difficult-to-model the site meteorology.

3.2 Enhancement of Atmospheric CO2 Concentration by Indian Fluxes

Figures 3 and 7 show the “enhancement” of atmospheric CO2 concentration at 975 hPa due to the biospheric and fossil fuel fluxes from India, respectively, as imprinted in the boundary layer of the global background and detectable at a concentration of 0.1 ppm or above. Climatological wind vectors constructed from the meteorology data (Hersbach et al., 2020) are overlaid in the figures. At 975 hPa, the maximum negative enhancement of CO2 concentration due to the Indian biosphere is noticed over the southern part of India with a magnitude greater than 2 ppm during SON (i.e., growing season) due to strong uptake after the Indian summer season, that is, JJA (Figure 3d). During JJA (biosphere mainly respire during this season), we noticed positive enhancement due to the Indian biosphere with a maximum magnitude of 1.4 ppm over India's central part.

Details are in the caption following the image

Spatial spread of enhancement due to Indian biospheric flux (in ppm) in global atmospheric CO2 concentrations at boundary layer (975 hPa) and calculated for four different seasons (December–February; DJF, March-May; MAM, June-August; JJA and September–November; SON). Concentrations above ±0.1 ppm are shown in colors. Boundary layer wind vectors correspondingly averaged over four seasons are overlaid (in ms−1).

Details are in the caption following the image

Same as Figure 3 but for 850 hPa.

Details are in the caption following the image

Same as Figure 3 but for 700 hPa.

Details are in the caption following the image

Same as Figure 3 but for 500 hPa.

Details are in the caption following the image

Spatial spread of enhancement due to Indian fossil fuel flux (in ppm) in global atmospheric CO2 concentrations at boundary layer (975 hPa) and calculated for four different seasons (December–February; DJF, March–May; MAM, June–August; JJA and September–November; SON). Concentrations above 0.1 ppm are shown in colors. Boundary layer wind vectors correspondingly averaged over four seasons are overlaid (in ms−1).

Tiwari et al. (2013) investigated the variability of dry-air mole fraction of atmospheric CO2 with vegetation and rainfall. The study also pointed out a strong biospheric uptake reflected as a low concentration of atmospheric CO2 during September-November. We noticed an “enhancement” of about 1.2–1.4 ppm over central India during JJA (Figure 3c). Strong surface wind from the ocean confined the atmospheric CO2 “enhancement” mostly over the Indian continent. A negative enhancement of atmospheric CO2 of −0.7 to 0.9 ppm was noticed over central India during DJF in the model (Figure 3a) when the biosphere acts as a sink to the atmospheric CO2, and a positive enhancement of about 0.3–0.4 during MAM (Figure 3b).

Throughout all seasons, the model CO2 enhancement due to fossil fuel was about 1.8–2 ppm over India (Figures 7a–7d); with the north-eastern part of India exhibiting lower values (Figures 7a–7d). During JJA the enhancement was noticed stretching over central India to the eastern parts, while during SON, we noticed a prominent pattern over the western part of peninsular India.

The biospheric “enhancements” spread toward the northeast, consistent with the cross-equatorial wind patterns during JJA. During DJF, the northeasterly winds direct the biospheric enhancement toward the Indian Ocean (Figure 3a) with the same pattern in the fossil fuel enhancement (Figure 7a). When the biosphere is active during SON, the maximum biosphere fluxes induced CO2 signal is noticed over India's southern part with an amplitude of 2.6 ppm (Figure 3d) while the fossil fuel enhancement at 975 hPa is about 2.0 ppm (Figure 7d). The atmospheric CO2 “enhancement” over India's north-eastern parts remained small for the biospheric fluxes and fossil fuel emission during all seasons at 975 hPa. In JJA and SON, the enhancement due to both the fluxes shows similar features with different amplitudes. The “enhancement” due to fossil fuel is higher throughout the season with an average value of 1.5 ppm than biosphere except for SON when biospheric uptake (2.6 ppm) is more than fossil fuel emission (2.0 ppm). It points out the importance of regional variability associated with the CO2 emission in response to vegetative growth, as India witnesses the seasonally reversing monsoon winds.

The CO2 “enhancement” due to the vertical transport of both the biospheric fluxes and fossil fuel emissions has a significant impact on the global carbon cycle (Niwa et al., 2012) as they are a manifestation of the localized atmospheric convection and mixing processes over a regional scale. Figures 4 and 8 represent the atmospheric CO2 ‘enhancement’ due to Indian biospheric fluxes and fossil fuel emission at 850 hPa level. During SON, the maximum negative “enhancement” of the Indian biospheric flux is observed over India's southern and western parts with an amplitude of about −1.6 to 2.0 ppm (represented as mean; Figure 4d). Strong monsoon winds carry the biospheric signal from central India to the northeast, as seen at 850 hPa (Figure 4c). During MAM, the “enhancement” due to biospheric fluxes is evident over a larger area with a magnitude up to −0.2 to −0.4 (Figure 4b).

Details are in the caption following the image

Same as Figure 7 but for 850 hPa.

An enhancement due to fossil fuel over the central to India's western part is found with a value of 1.2–1.4 and 1.2–1.6 ppm during MAM and SON, respectively (Figures 8b and 8d). Expectedly, the spatial distribution of the signature of both biospheric fluxes and fossil fuel emission follows the prevailing wind circulation patterns (Figures 4 and 8). The atmospheric CO2 negative enhancement due to biospheric fluxes was observed with a value of −0.2 to −0.8 ppm over central India during DJF (Figure 4a). However, the enhancement due to fossil fuel emission is about 1.4 to 1.6 ppm over the western, central, and eastern parts of India during DJF (Figure 8a). Over western parts of India, we noticed a negative enhancement due to biosphere is about −0.4 to −0.6 ppm, while the negative enhancement with a magnitude of −0.6 to −0.8 ppm was noticed over eastern parts of India during MAM (Figure 4b). MAM is the transition period when the biosphere started to be less active.

Similarly, the biospheric and fossil fuel “enhancement” over the Indian region at 700 and 500 hPa is also investigated (Figures 569, and 10). A larger spread of enhancement is observed in biosphere and fossil fuel fluxes over 700 and 500 hPa (Figures 5, 6 and 9, 10). The biosphere shows about −0.8 ppm negative “enhancement” over India's southern part at 700 hPa during SON (Figure 5d). Fossil fuel shows about 0.8–1.2 ppm “enhancement” over the central, southern, and western parts of India during MAM (Figure 9b). During JJA, a positive enhancement of magnitude 0.2–0.4 ppm was noticed over India due to the biosphere (Figure 5c). Due to the presence of anti-cyclonic circulation at 500 hPa, the signal of the biosphere and fossil fuel “enhancement” spreads toward a larger area (Figures 6c and 10c). Air-borne and satellite-based retrievals are equally important and useful for covering such a large extent and capturing CO2 enhancement to constrain sources and sinks in the inverse modeling framework. The intercontinental spread of enhancement can be captured only by satellite and airborne observations (Figures 6 and 10).

Details are in the caption following the image

Same as Figure 7 but for 700 hPa.

Details are in the caption following the image

Same as Figure 7 but for 500 hpa.

In order to evaluate the spatial and temporal coverage of GOSAT and OCO-2 satellite observations over India and adjacent regions for different seasons and its potential to capture the biospheric and fossil fuel signals, we have calculated the number of observations per day available from these satellites (Figures 11 and 12) within a region enveloped by a minimum column CO2 “enhancement” of 0.2 ppm. Here the choice of 0.2 ppm is due to two reasons. (a) The global average enhancement turns out to be 0.2 ppm from our above model experiments. The Indian total biospheric and fossil fuel fluxes per season tend to contribute a global enchantment of 0.2 ppm, assuming they are uniformly mixed. (b) The observational uncertainty of any ground-based measurements of atmospheric CO2 is also within 0.1–0.2 ppm.

Details are in the caption following the image

Total number of observations per day in each season from GOSAT satellite observations. The contours are column averaged of enhancement due to biospheric fluxes and fossil fuel emission together. The blue and green contour stand for 0.2 and 0.5 ppm, respectively.

Details are in the caption following the image

Same as Figure 11 but for OCO-2 satellite observations.

Inside 0.2 ppm contour (i.e., an average of both biospheric flux and fossil fuel enhancement), the number of observation available per day during monsoon season (i.e., JJA) is very limited as compared to the other seasons due to filtering for cloud contamination in the satellite data (Figures 11c and 12c). During DJF, we found both the satellites have the nearly same number of observations per day over India (Figures 11a and 12a). During SON, GOSAT has many observations per day over India compared to OCO-2 but has less coverage (Figures 11d and 12d). Overall, in all seasons, OCO-2 has a broader coverage of observations per day than GOSAT, yet the summer monsoon season has fewer usable observations in OCO-2. It suggests that satellite data coverage is insufficient to constrain India's carbon cycle variability during JJA seasons. Therefore, ground-based observations play a crucial role in filling this data gap over regions with persistent seasonal cloud cover (Tiwari et al., 2014). It implies that surface observation and satellite retrievals are equally important for constraining CO2 fluxes over India, particularly during monsoon and winter months.

Table 1 offers the percentage coverage by satellite data capturing Indian biospheric and fossil fuel enhancement within two ranges of column-averaged CO2 enhancement such as 0.2 and 0.5 ppm (blue and green contour, respectively, in Figures 11 and 12). Inside the 0.2 ppm boundary, OCO-2 shows about 64% coverage during JJ (Figure 12c). However, a careful examination pointed out that OCO-2 has more observations, especially over India during JJ (Figure 12c), whereas; GOSAT shows nearly no observation over India during this season (Figures 11c).

Table 1. Number of Observations (in a 2.5° x 2.5° configuration) Inside the Enhancement Region Marked by 0.2 and 0.5 ppm Contributed by Indian Biospheric and Fossil Fuel Fluxes (Contours Shown in Figure 11) Measurable by GOSAT and OCO-2 Satellites for Each Season. The Percentage of Available Observations Concerning Total Grids Within the Contours is Indicated in Parenthesis
Enhancement GOSAT OCO2
DJF MAM JJA SON DJF MAM JJA SON
0.2 ppm 103(54%) 51(39%) 7(6%) 121(44%) 187(98%) 118 (91%) 72 (64%) 244 (89%)
0.5 ppm 20(87%) 4 (80%) 0 (0%) 29 (51%) 23 (100%) 5(100%) 2 (50%) 56 (98%)

3.3 Anomalies of Atmospheric CO2 Observed by Satellites

We have investigated the XCO2 anomalies using two methods (see Section 2.3.2) from both the satellites inside a 0.2 ppm column “enhancement” zone as identified from the model experiments (Figures 13-16). Our purpose here is to see whether the current satellites can detect the above CO2 anomalies so that they may add value in an inversion-based modeling framework. It is particularly important as the GCP reports put India in the fourth place in fossil fuel emissions (Friedlingstein et al., 2019). However, it is hard to discern from satellite observations the enhancement due to emissions originating from different regions. We could do the best to identify the potential zones of CO2 enhancement above the global mean background, which we achieved from our model experiments. We adopted two independent methods in a search for this information and for maintaining fidelity as given in Section 2.3.2. From both methods, we get information about fossil fuel emission (anthropogenic activity) and biospheric activity. The positive anomalies may imply anthropogenic activity with less biospheric sink. Whereas the negative anomalies may indicate the biospheric sink with less anthropogenic activity.

Details are in the caption following the image

Seasonal mean XCO2 anomalies after subtracting daily track mean from each XCO2 observation for GOSAT in each 2.5° X 2.5° grid. The unit is in ppm. The blue and green contour stand for 0.2 and 0.5 ppm, respectively.

Details are in the caption following the image

Same as Figure 13 but for OCO-2.

Details are in the caption following the image

Seasonal mean XCO2 anomalies calculated as an average after subtracting daily median from individual daily XCO2 observation for GOSAT (Section 2.3.2). The unit is in ppm.

Details are in the caption following the image

Same as Figure 15 but for OCO-2.

The data from GOSAT and OCO-2 may discern about ±1.8 ppm XCO2 on an average within the zone of enhancement marked by 0.2 ppm in Figures 13 and 14, respectively. GOSAT hardly captures any anomaly over India during JJA (Figure 13c), which is due to intense cloud cover, whereas OCO-2 captures negative anomalies about a maximum of 1.0 ppm over Indo-Gangetic plain and positive anomalies 1 ppm over central parts(Figure 14c). Compared to GOSAT, OCO-2 captured XCO2 positive anomalies over central to the east region during MAM (∼1.8–2.0 ppm; Figures 14b and 16b), which may be related to the anthropogenic activity with less biospheric sink. During the growing season, i.e., SON, strong negative anomalies were noticed with a magnitude of −1.0 to 1.6 ppm from both the methods (Figures 13-16). During JJA a few grids are showing positive anomalies (∼1.0 ppm) over the north-eastern part of India from OCO-2 (Figures 14c and 16c), but we have not noticed considerable impact by anthropogenic activity over that region (Figure 1b). During DJF, GOSAT captures positive anomalies of 1.2–1.4 ppm over eastern parts, whereas OCO-2 shows 0.6–0.8 ppm positive anomalies over central to eastern parts of India (Figures 13a and 14a). However, we noticed negative anomalies during DJF (Figure 13a) related to the biospheric sink during that season. A strong positive XCO2 is noticed over India (∼2.0 ppm) from OCO-2 (Figure 16b). However, GOSAT captures XCO2 anomalies mostly over central India with a magnitude of ∼2.0 ppm (Figure 15b). During DJF strong negative XCO2 anomalies with a magnitude of ∼−2.0 ppm are noticed over North-East of North India from OCO2, whereas a few grids with comparable magnitude are highlighted from GOSAT (Figures 15a and 16a). In summary, both the methods consistently show that OCO-2 captures the XCO2 anomalies better than GOSAT with more data coverage.

Both the methods yielded more or less similar information. The amplitude of XCO2 positive (negative) anomalies due to the median-based method (Hakkarainen et al., 2019) was higher (smaller) than the mean-based method (similar to Park et al., 2020).

4 Conclusions

This study evaluated the enhancement of atmospheric CO2 concentration brought by the Indian biospheric fluxes and fossil fuel emission using NIES-TM as a transport model. The NIES-TM is validated with surface observations from MLO, SEY, SNG, and CRI and using GOSAT L4B data and proved that the transport derived are suitable for this study. NIES-TM replicates seasonal amplitude (5–6 ppm) and growth rate (∼2.5 ppm year−1) of MLO, a high altitude site, while producing significant amplitude mismatches in the Indian sites such as CRI and SNG implying imperfectness in the fluxes provided.

Using the model, we have conducted a suite of simulations with controlled fluxes from India. The study finds significant regions over India and surrounding where CO2 enhancement due to biospheric fluxes and fossil fuel emissions from India are mixed into the global atmospheric background. The zone of enhancement with notable amplitude (0.2 ppm) is found well spread from the Indian region to surrounding areas, including over the oceanic regions according to the prevailing atmospheric dynamics, and the results have been discussed. The study finds significant regions over India and around where CO2 enhancement reaches up to ±2.6 ppm due to biospheric fluxes during SON at 975 hPa. In comparison, the enhancement due to fossil fuel can reach up to 1.2–1.6 ppm during DJF, MAM, JJA, and SON seasons. At the surface, collectively, the CO2 enhancement by Indian biospheric fluxes and fossil fuel emissions contribute to the global background by ±1.5 ppm in four major seasons.

We investigated satellite coverage inside the regions marked by signals above 0.2 ppm as suggested by our model experiments by calculating the number of observations per day. We have found that GOSAT hardly observes these regions compared to the OCO-2 during JJA. Both the satellite showed a nearly similar number of observations per day during all other seasons (i.e., DJF, MAM, SON). The percentage of the number of observations inside the zone of the enhancement detected by the GOSAT satellite is 54%, 39%, 6%, and 44% for DJF, MAM, JJA, and SON, respectively, and that for OCO-2 are 98%, 91%, 64%, and 89% for DJF, MAM, JJA, and SON respectively. Despite 64% coverage by OCO-2, we noticed a relatively large data gap over India during JJA.

We have also found the satellite XCO2 anomalies inside the zones (marked by 0.2 ppm as guided by our model experiments) as available from GOSAT and OCO-2. The data from both the satellites may discern about ±1.8 ppm anomalies in XCO2 on an average within this region. The study highlights the need for more ground-based CO2 observations over India, especially during the summer months (June to September) when cloud cover masks the satellite observations.

Acknowledgments

NIES-TM and JCDAS/JRA-55 meteorology to run the model is obtained from the National Institute of Environmental Studies, Tsukuba, Japan. Model simulations are carried out at the High-Performance Computing (HPC) facility at IITM, Ministry of Earth Sciences (MoES), Govt of India. MoES support this work through its various programs operating at IITM. SMG acknowledges the funding support by the Institute for Basic Science (IBS), Republic of Korea, under IBS-Ro28-D1.

    Data Availability Statement

    ACOS-GOSAT version 9 L2 daily bias-corrected observations are taken from (https://disc.gsfc.nasa.gov/datasets/ACOS_L2_Lite_FP_9r/summary?keywords=gosat), OCO-2 observations are obtained from OCO-2 Level-2 daily Lite product of version 9 (https://co2.jpl.nasa.gov/#mission=OCO-2). Surface CO2 observations of MLO, SEY, and CRI are taken from WDCGG (https://gaw.kishou.go.jp/). The GOSAT Level-4 data can be found at https://data2.gosat.nies.go.jp/.