Finding local anthropogenic CO2 hotspots in OCO-2 data

(1)

Finding local anthropogenic CO

2

hotspots

Linde van de Ven - 10559051

29 June 2017

Netherlands Institute for Space Research - SRON Universiteit van Amsterdam

Report Bachelor Project Physics, 12 EC Conducted between 10-04-2017 and 22-06-2017

Supervisors: Dr. R.G. Detmers Prof. Dr. E.A.A. Aben

Second Assessor: Dr. W. Vassen

(2)

Abstract

CO2 is a greenhouse gas that causes global warming. To stop the increase of

CO2in the atmosphere, sources and sinks of CO2have to be monitored. OCO-2

is a satellite that measures the total vertical column of CO2in the atmosphere

above an area. This satellite can be used to locate sources and sinks of CO2on

a regional scale. The OCO-2 CO2fields were corrected for seasonal and yearly

trends to be able to localize potential CO2hotspots. Several potential hotspots

were subsequently further examined. Time series for those locations were then inspected to assess the robustness of the positive identification of the hotspots. The so localized hotspots in CO2 were then compared to the EDGAR and ODIAC databases. Most of the observed hotspots in the OCO-2 data were also visible in the EDGAR database, but some were not seen in either the EDGAR or ODIAC database.

(3)

Populair wetenschappelijke samenvatting

CO2is een broeikasgas dat zorgt voor opwarming van de aarde. De concentratie

CO2is sinds de industri¨ele revolutie van 280 parts per million gestegen naar 400

ppm, door menselijke uitstoot veroorzaakt door onder andere het gebruik van fossiele brandstoffen. Dit heeft overstromingen, droogtes en stijging van de zeespiegel tot gevolg. Om de stijging van CO2tegen te gaan is wetgeving nodig,

zoals de afspraken die gemaakt zijn op de klimaattop in Parijs in 2015. Deze wetten moeten ook nageleefd worden, en om dit te controleren in zijn objectieve metingen nodig.

Er zijn verschillende databases met informatie over CO2 uitstoot, zoals de

EDGAR en de ODIAC database. Deze databases baseren hun informatie on-der anon-dere op data van overheden en bedrijven. Het probleem is dat er veel onnauwkeurigheden bij komen kijken. Bedrijven kunnen hun emissies verkeerd inschatten of te laag opgeven om aan de wetgeving te voldoen. Een relatief nieuwe manier om CO2 objectief te meten is met een satelliet, zoals het

Orbit-ing Carbon Observatory-2 (OCO-2). Deze satelliet kan vanuit een baan om de aarde het gereflecteerde zonlicht meten, en met behulp van spectrometers de concentratie CO2 in een rechte kolom boven een klein gebied op aarde meten.

Hiermee kan dus gemeten worden of een bepaald gebied een verhoogde uitstoot heeft.

In dit onderzoek wordt er gezocht naar zogenaamde hotspots in de OCO-2 data. Dit zijn bronnen van CO2 ter grootte van een stad of kleiner. De

concentratie van CO2 in de atmosfeer is 400 ppm, maar een hotspot verschilt

daar maar 2 tot 3 ppm van. Dit is dus maar een verhoging van ongeveer 0.5 %, en dus een uitdaging om te detecteren.

Figure 1: OCO-2 data van de westkust van de Verenigde Staten. De hotspot links is Los Angeles, rechts is Rapid City Elke hotspot is vergeleken met

de informatie uit de EDGAR en de ODIAC databases.

Een van de hotspots die was gevonden was Los Angeles, aan de westkust van de Verenigde Staten. Deze bron was verwacht, en was ook heel duidelijk te zien in de EDGAR en ODIAC data.

Een meer opmerkelijke vondst was de verhoging boven Rapid City, een kleine stad in South Dakota. Voor deze locatie geven de emissie databases EDGAR en ODIAC geen uitstoot. In dit onderzoek zijn een aantal interessante CO2 hotspots

gevonden die niet (prominent) aan-wezig zijn in de EDGAR of ODIAC

databases. Voor een volgend onderzoek zouden deze bronnen verder onderzocht kunnen worden.

(4)

1 Introduction

CO2is a greenhouse gas, and its presence in the atmosphere causes the so called

greenhouse effect. Human activity causes increased CO2 levels, which causes

global warming. (Bovensmann et al., 2010, Schneising et al., 2013) Emitted CO2 is partially absorbed by the oceans, on which it has an acidifying effect,

and by the land biosphere. (Eldering et al., 2017) CO2 has a decay half-life

in the atmosphere of 100 years, so even small amounts of emission can have a long-lasting effect. (Cherubini et al., 2011)

Global warming has consequences like floodings and droughts. The amount of CO2 in the atmosphere has risen from 280 ppm pre-industrial revolution to

about 400 ppm in present day. (Hakkarainen et al., 2016, Bovensmann et al., 2010, Keppel-Aleks et al., 2013) The extra emitted CO2in this period originates

from anthropogenic sources, mainly the use of fossil fuels, cement production and land use change. (Le Qu´er´e et al., 2015) Most of the emission comes from highly populated regions, especially cities. (Duren & Miller, 2012, Kort et al., 2012)

To stop the rising CO2 levels, the anthropogenic emission of CO2 has to

be reduced. (IPCC, 2014) Legislation, based on international agreements like the ones made at the Paris convention of 2016, can limit CO2 emission. The

effects of this legislation need to be checked in order to keep emitters to these arrangements. (Janardanan et al., 2016, Keppel-Aleks et al., 2013)

Inventories of CO2emissions exist, like the EDGAR database. These databases

estimate emissions based on data from governments and companies. (Bovens-mann et al., 2012) The uncertainties in these databases are usually high, com-panies can publish wrong numbers, governments can adjust data for political reasons and sometimes accurate data is just not available. (Andres et al., 2012, Gregg et al., 2008, Duren & Miller, 2012) Especially in China the uncertainty of emissions can be higher than 50%. This is partially due to rapid developments and resulting changes in emissions. (Guan et al., 2012, Zhao et al., 2011)

In order to get a clear view of global CO2emissions an objective and neutral

method is needed, and satellites measuring CO2is an interesting new

develop-ment, that could achieve this. (Bovensmann et al., 2012, Duren & Miller, 2012, Schneising et al., 2013) A few of these satellites exist, like the Orbiting Carbon Observatory-2 (OCO-2), SCIAMACHY and GOSAT. These satellites can de-tect trends of CO2, which can be compared to the already existing databases.

(Schneising et al., 2013)

The background level of CO2is high because of the long residence time of the

gas in the atmosphere. A hotspot of CO2usually has an elevated concentration

of about +2 ppm, so this is only about 0.5% of the background level of CO2.

(Kort et al., 2012, Mandrake et al., 2015) This means that a high level of precision is required in satellite measurements to be able to find elevations in small regions. In the data of GOSAT, SCIAMACHY and OCO-2 some hotspots have been found, like the cities of Mumbai and Los Angeles. (Kort et al., 2012, Hakkarainen et al., 2016)

(6)

vertical column of CO2in the atmosphere, while orbiting the globe. The

accu-racy of OCO-2 is high enough to detect anthropogenic emissions. (Keppel-Aleks et al., 2013) OCO-2 data have been mostly used to recognize regional or global trends, and Hakkarainen et al. (2016) have succesfully located a few hotspots using this satellite.

The objective of this study is to determine whether OCO-2 can be used to de-tect anthropogenic point sources of CO2 and subsequently localize these sources. These hotspots are then compared to the databases EDGAR and ODIAC. This could eventually lead to an update of existing databases of CO2 emission, which could be used in legislation and the enforcement of these laws.

2 Data & Method

2.1 OCO-2

OCO-2 was launched on 2 July 2014, and runs in the A-train. It was designed to measure the amount of CO2 in a vertical column above an area of 1.25 by

2.4 km. The intensity of sunlight reflected by the earth is measured by three different Near Infrared wavelength bands. (Eldering et al., 2017) OCO-2 can measure in nadir, glint or target mode. In nadir mode the instrument measures the vertical column directly below the instrument. In glint mode, mostly used over sea, OCO-2 points to the glint spot to make measurements. For expected hotspots or reference stations on earth target mode can be used, OCO-2 then points to a specific area for as long as possible. OCO-2 has a 16-day repeat cycle, in that time it will make about 233 orbits.(Wunch et al., 2017)

About 7-12 % of the data is sufficiently cloud free to be used. Using reference ground based measurements it was found that the OCO-2 data is biased by less than 1.5 ppm (standard deviation 1.2 ppm). (Eldering et al., 2017, Wunch et al., 2017) The cloud-filtered data is saved in the LITE-files, which have one file of data per day.

OCO-2 data comes with a quality flag, which is either zero (”good”) or one (”bad”). It is recommended to use this quality flag to filter the LITE data and we therefore only use data with a quality flag of 0. (Mandrake et al., 2015) Data from September 2014 until March 2017 was used.

2.2 Emission databases

The EDGAR database is an inventory of anthropogenic CO2-emissions, based

mainly on governmental data. The EDGAR data gives a clear image of the different categories of anthropogenic sources of emission. Among the categories are sources like biomass burning, aviation and oil production (for a full list, see appendix). (IPCC, 1996) This database is used to identify hotspots found in the OCO-2 data, and to have a starting point looking for hotspots in the OCO-2 data. In this way the OCO-2 data can be linked to a certain type of emission. (European Commission, 2016) The global mean emission according to EDGAR

(7)

is 0.3954 gram/m2/day. The most recent available data is from 2010, this data was used. (European Commission, 2016)

ODIAC is another inventory that is based on power plant data from database CARMA and satellite nightlight data. This data is not subdivided in different categories. From the year 1970 onwards data is available. (Oda & Maksuytov, 2011) In this study, data from 2015 on a 1 x 1 degree resolution was used. (Oda & Maksuytov, 2015) This resolution might make spotting hotspots difficult, a hotspot that is visible on a 0.25 x 0.25 degree grid is 16 times ’diluted’ on a grid of 1 x 1 degree. The global mean emission according to ODIAC is 0.2212 gram/m2_/day.

2.3 Corrections for seasonal and yearly trends

OCO-2 has a small swath-width of about 10 km, so the data has no complete coverage of the world. For some regions only a small amount of data is available, like the regions above 48◦_{N, because of low sun and corresponding low signal}

levels. (Mandrake et al., 2015) Above cloudy areas and deserts, aerosols and clouds make retrieving accurate data impossible. The gaps in the data can be partially ’filled’ by choosing a lower resolution. A lower resolution however can result in the disappearance of highly localized hotspots in the data. A grid of 0.25 x 0.25 degrees was chosen as the optimal resolution for this study. Because of a lack of data, some regions have not been used to find hotspots.

(8)

The OCO-2 data were thus stored on a grid of 0.25 x 0.25 degrees, and then plotted on a world map. A clear difference in CO2 concentration was

seen between the different years, with a difference of about 3-4 ppm per year. This agrees with the global rising CO2 level. (Solomon et al., 2009) A clear

seasonal trend was also visible (see Figure 2). Growing and decaying vegetation causes a variation of atmospheric CO2. This seasonal trend is stronger on the

Northern Hemisphere because of the large amount of vegetation, and can cause a difference of 3 ppm in the concentration of CO2. (Schneising et al., 2013,

Hakkarainen et al., 2016) Hotspots of +2 ppm (Kort et al., 2012) can not be detected with this amount of seasonal and yearly variability.

To be able to detect hotspots, the data was therefore compared to back-ground CO2 levels. For this background a monthly mean was calculated and

the data was compared to this monthly mean. In this way the seasonal and yearly trends on a large scale are corrected for. The background monthly mean was subtracted for each 0.25 x 0.25 degrees gridcell.

deltaXCO2= XCO2(gridcell) − XCO2(monthlymean)

The delta CO2 is not affected by seasonal or yearly trends on a large scale,

because the background will be affected by the same external influences as the deviation. All trends that affect a region the same size or bigger than the background will be removed, so hotspots will still be intact. (Hakkarainen et al., 2016, Janardanan, 2016).

Optimally, the background is chosen such that it best represents the back-ground CO2 levels around a hotspot. This would favour to choose smaller areas to determine the background. On the other hand, the background area should also be chosen such that its average CO2 level is accurately determined. For this a large amount of measurements is needed and this favours the choice of a larger background area. As the optimal size of the background area is not easily determined we used two different sizes in this study.

Two different sized background areas (large and small) were used. The large background is defined as the region USwest, Middle East or South Africa (see Figures 3, 5, and 7). These areas were divided in four parts, which formed the small backgrounds. This division was made by simply dividing the region in four equal parts.

A weighted average over time per gridcell was then calculated by taking the error given with the OCO-2 data.

weighteddeltaXCO¯ 2=

PN

i=1XCO2error−2∗ deltaXCO2

PN

i=1XCO2error−2

N is the total amount of datapoints within a gridcell. This produced one average value for every gridcell in the hotspot.

(9)

the standard deviation over the CO2 values in all the gridcells in a hotspot. SD = v u u t MPN

i=1XCO2error−2i ∗ (XCO2i− weighteddeltaXCO¯ 2)

(M − 1)PN

i=1XCO2error−2i

M stands for the total number of weights that are nonzero, N is the number of observations within the entire hotspot. From now on whenever the delta CO2

is mentioned, this means the weighteddeltaXCO¯ 2.

3 Results

A first general survey was done to make a selection of interesting areas with enough data to look for hotspots. China has high emissions, especially on the east coast, but individual hotspots were difficult to point out. In Europe the amount of OCO-2 data was not high enough, just like in South America and the northern part of Africa. These areas were therefore not used. Areas that were further studied were the western part of the US, the Middle East and the southern part of Africa. Only hotspots over land were studied.

A hotspot was defined as two or more gridcells with a difference of at least 1.5 ppm compared to its surrounding.

For each hotspot, the delta CO2 time series were plotted. A couple of the

most clear hotspots are presented here. More information on these hotspots can be found in the appendix, as well as a number of other identified hotspots.

For all these hotspots the delta CO2 is higher than the spread in the

data-points (the standard deviation). The delta CO2 is comparable to the expected

elevated concentration for a hotspot of +2 ppm. (Kort et al., 2012) The time se-ries show that for most measurements the delta CO2is higher than 0. They also

show that for all hotspots -except for Los Angeles- the measurement sampling in time is limited. For Los Angeles the target mode is used, which explains the high number of measurements and large coverage in time of the data. Because of this limited sampling in time the average delta CO2and the entire time series

(and not per month) is studied as a means to determine whether the identified hotspot showed a significant elevation in the delta CO2. It should be noted that

only part of the spread in the datapoints is caused by the precision and errors in the data. CO2 variation across the hotspot as well as variation of CO2 in

(10)

3.1 Western part of the US

Over the western part of the US, the OCO-2 data (see Figure 3) shows roughly the same patterns as observed in the EDGAR database. Along the Westcoast a large enhancement is visible in the EDGAR data, between 33◦N and 39◦N. This is also visible in the OCO-2 data, only here this enhancement has been moved east. This is likely because the wind moves the CO2 east, where it cumulates

at the foot of the Sierra Nevada. What is surprising is that larger cities like Denver (40◦N, 105◦W) are visible in the EDGAR data, but not in the OCO-2 data.

Figure 3: Top - OCO-2 data over western US

Left - EDGAR data over western US on a logarithmic scale, Right - ODIAC data over western US

(11)

Rapid City

Rapid City is a city in South Dakota with about 72.638 inhabitants. The major industry is agriculture, next to some mining activities. (Rapid City: Economy, 2017) Rapid City has two working powerplants, with natural gas as main power source. This area is not visible in the EDGAR data. Also, the value of ODIAC is lower than the global average of 0.22 gram/m2_{/day. As the ODIAC data is}

used at 1 x 1 degree resolution, it could be that the signal from a real point source is too much diluted to be compared with the 0.25 x 0.25 degree OCO-2 data. In this resolution the hotspot could disappear in the surrounding. The two different backgrounds both indicate a clear hotspot, although the small background gives a higher delta CO2. However, the difference in elevation

be-tween the two backgrounds is clearly visible when comparing the total in delta CO2 and the corresponding standard deviation (see Figure 4, Table 1). This

difference highlights the importance of choosing correct backgrounds. We will come back to this point later

Table 1: Rapid City

Lon Lat Delta CO2: large bg in ppm SD Delta CO2: small bg in ppm SD Nr of measure-ments Nr of grid-cells EDGAR in gram/ m2/day ODIAC in gram/ m2/day -103 44 2.95 1.36 3.97 1.19 148 5 0.55 0.22 Los Angeles

Los Angeles is a megacity in California, with a lot of industry. Steel is produced, along with apparel, computer and electronic product. (Los Angeles: Economy, 2017) Figure 4 (middle) shows the individual delta CO2 for this location. The

corresponding averaged delta CO2and the total number of observations is given

in Table 2. This hotspot is visible in the EDGAR and ODIAC data. What is surprising is the deviation of Los Angeles compared to the deviation observed for Rapid City. According to ODIAC and EDGAR Los Angeles should have a much larger CO2enhancement than Rapid City in the large background. This

background is the same for both hotspots, so a larger deviation in Los Angeles was expected.

Table 2: Los Angeles

Lon Lat Delta CO2: large bg in ppm SD Delta CO2: small bg in ppm SD Nr of measure-ments Nr of grid-cells EDGAR in gram/ m2/day ODIAC in gram/ m2/day -118 34 3.08 1.64 2.38 1.59 36771 9 15.80 2.46

(12)

Figure 4: delta CO2 in time for three different hotspots using the large

(13)

Seattle

Seattle, Washington is home to aerospace, iron and steel industry, among oth-ers.(Seattle: Economy, 2017) The hotspot -as observed in Figure 3 and 4 in the OCO2 data- was expected looking at the EDGAR and the ODIAC data. This hotspot has a larger standard deviation than delta CO2for the large

back-ground (see Table 3), but looking at the time series Seattle appears to be a clear hotspot. Table 3: Seattle Lon Lat Delta CO2: large bg in ppm SD Delta CO2: small bg in ppm SD Nr of measure-ments Nr of grid-cells EDGAR in gram/ m2/day ODIAC in gram/ m2/day -121 47 1.40 1.44 2.02 1.56 505 7 6.83 1.42

(14)

3.2 Middle East

In this region a large elevation in CO2is visible in EDGAR, ODIAC and

OCO-2 (see Figure 5), the region spreading from Baghdad to Kuwait City (around 33◦N, 48◦E). In this region a large amount of oil drilling takes place. This el-evation is also clearly visible in the research of Hakkarainen et al. (2016). In Ca¨ıro (30◦N, 31◦E) no large hotspot is visible in the OCO-2 data, while ODIAC and EDGAR show a clear elevation. Nonetheless, the Nile Delta is visible in the OCO-2 data as well as some CO2elevation along the Nile.

Figure 5: Top - OCO-2 data over Middle East

Left - EDGAR data over Middle East on a logarithmic scale, Right - ODIAC data over Middle East

(15)

Teheran

Teheran houses more than half of Iran’s industry, including electronics, weaponry, cement and chemical products. (Tehran, Economy, 2017) The hotspot shows up clearly in the EDGAR and the ODIAC data, as well as in the OCO-2 data. (see Figure 6, Table 4)

Table 4: Teheran Lon Lat Delta CO2: large bg in ppm SD Delta CO2: small bg in ppm SD Nr of measure-ments Nr of grid-cells EDGAR in gram/ m2/day ODIAC in gram/ m2/day 49 37 2.13 1.38 1.78 1.61 127 4 7.08 0.74 Baghdad 2

This hotspot as observed in the OCO-2 data (Figure 5 and 6, Table 5) lies about 100 km southwest of the city of Baghdad, which houses most of Iraq’s industry. (Baghdad Governorate Profile, 2015) EDGAR indicates a clear hotspot. It is questionable whether this point can be called a hotspot. In the surrounding of this area a large elevation over a large area is visible. It’s possible that the CO2

concentration in this area is not emitted by a point source within this hotspot, but by various sources within the larger area. To determine whether ODIAC gives a clear hotspot further study is necessary.

Table 5: Baghdad 2 Lon Lat Delta CO2: large bg in ppm SD Delta CO2: small bg in ppm SD Nr of measure-ments Nr of grid-cells EDGAR in gram/ m2/day ODIAC in gram/ m2/day 44 32 2.44 1.31 1.97 1.35 121 3 4.30 0.61 Khuff

A clear hotspot in the OCO-data was found near Khuff (see Figure 5 and 6, Table 6), a small city about 250 km west of Medina. The EDGAR and ODIAC data indicate no emissions are expected. This is surprising, since the OCO-2 data clearly indicates a hotspot.

Table 6: Khuff Lon Lat Delta CO2: large bg in ppm SD Delta CO2: small bg in ppm SD Nr of measure-ments Nr of grid-cells EDGAR in gram/ m2/day ODIAC in gram/ m2/day 37 25 2.07 1.06 2.42 1.06 170 3 0.27 0.06

(16)

Figure 6: delta CO2 in time for three different hotspots using the large

(17)

Figure 7: Top - OCO-2 data in southern Africa

Middle - EDGAR data in southern Africa on a logarithmic scale Bottom - ODIAC data in southern Africa

(18)

3.3 Southern part of Africa

In the EDGAR and ODIAC data a clear elevation is visible around Pretoria, at 27◦S, 27◦E (see Figure 7). This elevation is not visible in the OCO-2 data. Along the east coast a general elevation is visible in the OCO-2 data, as well as in the northern part. This could be caused by biomass burning like forest fires. Knysna

EDGAR: 0.09 gram/m2_/day

ODIAC: 0.03 gram/m2_/day

Knysna is a small town in the Western Cape. According to EDGAR and ODIAC, almost no emission is produced. Nonetheless, the OCO2 data appears to show slightly elevated CO2 values (Figure 7 and 8, Table 7)

Table 7: Knysna Lon Lat Delta CO2: large bg in ppm SD Delta CO2: small bg in ppm SD Nr of measure-ments Nr of grid-cells EDGAR in gram/ m2/day ODIAC in gram/ m2/day 23 -34 1.88 1.63 1.93 1.54 210 6 0.09 0.03 Richardsbaai

EDGAR: 0.3 gram/m2_/day

ODIAC: 0.14 gram/m2/day

Richardsbaai is a town in KwaZulu-Natal, with a large coal terminal. The EDGAR data is low, but indicates some emission. The ODIAC data is not clear enough on this resolution. This would be an interesting case for a follow-up study as OCO-2 shows a clear hotspot (see Figure 7 and 8, Table 8)

Table 8: Richardsbaai Lon Lat Delta CO2: large bg in ppm SD Delta CO2: small bg in ppm SD Nr of measure-ments Nr of grid-cells EDGAR in gram/ m2/day ODIAC in gram/ m2/day 31 -29 2.47 1.45 2.42 1.46 378 4 0.30 0.14

(19)

Quelimane

EDGAR: 0.01 gram/m2/day ODIAC: 0.00 gram/m2/day

This hotspot was found 80 kilometers south of Quelimane, Zambezia in Mozam-bique According to the EDGAR and the ODIAC data no emission is present. The standard deviations in the OCO-2 data are low, but the deviation in small background is small. The time series show a lot of measurements, spread over the entire period. (see Figure 7 and 8, Table 9) This emission in the OCO-2 data could be caused by biomass burning.

Table 9: Quelimane Lon Lat Delta CO2: large bg in ppm SD Delta CO2: small bg in ppm SD Nr of measure-ments Nr of grid-cells EDGAR in gram/ m2/day ODIAC in gram/ m2/day 36 -19 1.53 1.08 1.11 1.13 580 4 0.01 0.00

(20)

Figure 8: Left - Large background Right - Small background

(21)

4 Conclusion and Discussion

OCO-2 clearly shows a number of localized elevations that have been positively identified as hotspots, proving OCO-2 is able to detect anthropogenic point sources. Elevated CO2values are clearly visible over large cities like Los Angeles

and Teheran, that are also present in the EDGAR and ODIAC data. Also a number of hotspots have been found in the OCO-2 data that do not correspond to large cities and are not present in the EDGAR or ODIAC data.

There are a number of elements in this study that would be useful to improve. In particular the way we determine the background can be further improved. Large and small background areas were defined in this study. The regions of interest were chosen as large backgrounds, and these were divided in four equal pieces to obtain the small backgrounds. This method of choosing background areas is an efficient and easy way of finding and comparing the hotspots, but this method is not ideal. A few of the hotspots in the appendix are close to the edges of these standard background areas. This means that the hotspot is compared to data that isn’t close to the hotspot, and the hotspot is not compared to points that are close and should be included in the background. It would be best to define the background based on the location of the hotspot, by defining the background as a box of a fixed size with the hotspot exactly in the middle. The quantitative results of the OCO-2 data for a hotspot (delta CO2 and

standard deviation) differed a lot between the two different background areas. This is most likely due to the fact that in calculating the monthly mean, all data in the background was used. In this way the hotspot and possible other hotspots are included in the monthly mean, so the presence of hotspots in the background influences the visibility of the hotspot of interest. If the number of hotspots differs greatly between two different background areas, the delta CO2

and standard deviation for a hotspot will differ as well. It would be best to create a background that excludes hotspots in calculating the background monthly mean. Also interesting would be to compare the data to a daily background mean instead of a monthly background mean, like Hakkarainen et al. (2016) did. This way the data might correct for trends more accurately.

The standard deviation in delta CO2is derived over an entire hotspot. This

is very efficient, but has a few disadvantages. CO2 variation across the hotspot

and variation of CO2emission over time contribute to the observed spread, and

could in this way increase the standard deviation.

The ODIAC data was used on a 1 x 1 degree grid. Using the ODIAC data at higher spatial resolution could show more point sources like powerplants. Comparing the OCO-2 data to this data could give more matches between the data. This could also be helpful for the EDGAR data. The EDGAR data stems from 2010 and might be outdated for at least a few categories, like emission from biomass burning caused by forest fires. This might be one of the reasons some hotspots from the OCO-2 data aren’t visible in the EDGAR data.

This method is very labor intensive, so it would be useful to automate a part of the process. All hotspots were selected manually, of course this could be more efficient. For a next study it would be advisable to have an algorithm

(22)

that extracts the most clear hotspots and gives all available data about them, at which point they can be manually checked.

The data of OCO-2 in this study is only represented as the total amount in the vertical column and the difference between a monthly background mean and the hotspot, both in parts per million. It would be interesting to calculate the absolute value of emission in g/m2_{/day, the unit in which the EDGAR and}

ODIAC data are also represented. This would require a detailed inversion setup at high resolution in order to resolve the small footprint of the sources.

Nevertheless this study has shown the potential of OCO-2 for CO2 hotspot

detection and this method should be further expanded using the recommenda-tions listed above.

5 References

Andres, R. J., Boden, T. A., Bron, F. M., Ciais, P., Davis, S., Erickson, D., ... & Oda, T. (2012). A synthesis of carbon dioxide emissions from fossil-fuel combustion. Biogeosciences.

Baghdad Governate Profile. (2015). Retrieved 21 June 2017 from

http://www.ncciraq.org/images/infobygov/NCCI Baghdad Governorate Profile.pdf. Bovensmann, H., Buchwitz, M., Burrows, J. P., Reuter, M., Krings, T., Gerilowski, K., ... & Erzinger, J. (2010). A remote sensing technique for global monitoring of power plant CO2 emissions from space and related applications.

Atmospheric Measurement Techniques, 3(4), 781.

Cherubini, F., Peters, G. P., Berntsen, T., Strmman, A. H., & Hertwich, E. (2011). CO2emissions from biomass combustion for bioenergy: atmospheric

decay and contribution to global warming. Gcb Bioenergy, 3(5), 413-426. Duren, R. M., & Miller, C. E. (2012). Measuring the carbon emissions of megacities. Nature Climate Change, 2(8), 560-562.

European Commission, Joint Research Centre (JRC)/Netherlands Environ-mental Assessment Agency (PBL). Emission Database for Global Atmospheric Research (EDGAR), release version 4.3.1. Retrieved June 2017 from

http://edgar.jrc.ec.europa.eu/overview.php?v=431, 2016.

Guan, D., Liu, Z., Geng, Y., Lindner, S., & Hubacek, K. (2012). The gigatonne gap in China's carbon dioxide inventories. Nature Climate Change, 2(9), 672-675.

Gregg, J. S., Andres, R. J., & Marland, G. (2008). China: Emissions pattern of the world leader in CO2 emissions from fossil fuel consumption and cement

production. Geophysical Research Letters, 35(8).

Hakkarainen, J., Ialongo, I., & Tamminen, J. (2016). Direct spacebased observations of anthropogenic CO2 emission areas from OCO2. Geophysical

Research Letters, 43(21).

IPCC, 2014: Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergov-ernmental Panel on Climate Change [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland, 151 pp.

(23)

Janardanan, R., Maksyutov, S., Oda, T., Saito, M., Kaiser, J.W., Ganshin, A., Stohl, A., Matsunaga, T., Yoshida, Y., & Yokota, T., (2016). Comparing GOSAT observations of localized CO2 enhancements by large emitters with

inventory-based estimates, Geophys. Res. Lett., 43, 3485-3493.

Keppel-Aleks, G., Wennberg, P. O., O’Dell, C. W., & Wunch, D. (2013). Towards constraints on fossil fuel emissions from total column carbon dioxide. Atmospheric Chemistry and Physics, 13(8), 4349-4357.

Kort, E. A., Frankenberg, C., Miller, C. E., & Oda, T. (2012). Spacebased observations of megacity carbon dioxide. Geophysical Research Letters, 39(17). Los Angeles: Economy. (n.d.). Retrieved 21 June 2017 from http://www.city-data.com/us-cities/The-West/Los-Angeles-Economy.html.

Mandrake, L., ODell, C., Wunch, D., Wennberg, P. O., Fisher, B., Osterman, G. B., & Eldering, A., (2015). Lite Files, Warn Levels, and Bias Correction De-termination, NASA Jet Propulsion Labora- tory, California Institute of Technol-ogy, Version 1. Retrieved June 2017 from: http://disc.sci.gsfc.nasa.gov/OCO-2/documentation/oco-2-v7/OCO2XCO2 Lite Files and Bias Correction 0915 sm.pdf

Oda, T., & Maksyutov, S. (2011). A very high-resolution (1 km 1 km) global fossil fuel CO 2 emission inventory derived using a point source database and satellite observations of nighttime lights. Atmospheric Chemistry and Physics, 11(2), 543-556.

Oda, T., & Maksyutov, S. (2015). ODIAC Fossil Fuel CO2 Emissions

Dataset (ODIAC2016), Center for Global Environmental Research, National In-stitute for Environmental Studies, doi:10.17595/20170411.001. (Reference date : 2017/05/22)

Le Qur, C., Moriarty, R., Andrew, R. M., Peters, G. P., Ciais, P., Friedling-stein, P., ... & Boden, T. A. (2015). Global carbon budget 2014. Earth System Science Data, 7(1), 47-85.

Rapid City: Economy. (n.d.). Retrieved 21 June 2017 from http://www.city-data.com/us-cities/The-Midwest/Rapid-City-Economy.html.

Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories: Reporting Instructions. Retrieved June 2017 from

http://www.ipcc-nggip.iges.or.jp/public/gl/guidelin/ch1ri.pdf

Schneising, O., Heymann, J., Buchwitz, M., Reuter, M., Bovensmann, H., & Burrows, J. P. (2013). Anthropogenic carbon dioxide source areas observed from space: assessment of regional enhancements and trends. Atmospheric Chemistry and Physics, 13(5), 2445-2454.

Seattle: Economy. (n.d.). Retrieved 21 June 2017 from http://www.city-data.com/us-cities/The-West/Seattle-Economy.html.

Solomon, S., Plattner, G. K., Knutti, R., & Friedlingstein, P. (2009). Ir-reversible climate change due to carbon dioxide emissions. Proceedings of the national academy of sciences.

Tehran, Economy. (2017, June 21). Retrieved 21 June 2017 from http://en.tehran.ir/Default.aspx?tabid=100.

Wunch, D., Wennberg, P. O., Osterman, G., Fisher, B., Naylor, B., Roehl, C. M., ... & Griffith, D. W. (2017). Comparisons of the Orbiting Carbon

(24)

Observatory-2 (OCO-2) X CO 2 measurements with TCCON. Atmospheric Measurement Techniques, 10(6), 2209.

Zhao, Y., Nielsen, C. P., Lei, Y., McElroy, M. B., & Hao, J. (2011). Quan-tifying the uncertainties of a bottom-up emission inventory of anthropogenic atmospheric pollutants in China. Atmospheric Chemistry and Physics, 11(5), 2295-2308.

(25)

6 Appendix

6.1 EDGAR categories

Agricultural Aviation Biomass burning Rice Cultivation -Irrigated -Continuously Flooded -Intermittently Flooded -Single aeration -Multiple aeration -Rainfed -Flood prone -Drought prone -Deep Water -Water depth 50 to 100 cm -Water depth 100 cm Agricultural Soils Civil Aviation

Changes in Forest and Other Woody Biomass Stocks Abandonment of Managed Lands -Tropical Forests -Temperate Forests -Boreal Forests -Grasslands/Tundra CO2 Emissions and Removals from Soil Prescribed

(26)

Chemical

solvents Combustion Manufacturing Energy buildings Chemical Industry

-Ammonia Production -Nitric Acid Production -Adipic Acid Production -Carbide Production

Solvent and Other Product Use -Paint Application

-Degreasing and Dry Cleaning -Chemical Product,

Manufacture and Processing

Manufacturing Industries and Construction (ISIC) -Iron and Steel

-Non-Ferrous Metals -Chemicals

-Pulp

-Paper and Print

-Food Processing, Beverages and Tobacco Other Sectors -Commercial/Institutional -Residential Agriculture/Forestry/Fishing -Stationary

-Off-road Vehicles and Other Machinery -Fishing

Energy

industry Fugitives Mineral

Public

Electricity and Heat Production: -Public Electricity Generation -Public Combined Heat and Power Generation (CHP)

-Public Heat Plants

Solid Fuels Mineral Products -Cement Production -Lime Production

-Limestone and Dolomite Use -Soda Ash Production and Use -Asphalt Roofing

-Road Paving with Asphalt Non road Oil Production Road Railways Other Transportation -Pipeline Transport -Off-road Oil -Exploration -Production -Transport -Refining/Storage

-Distribution of Oil Products

Road

Transportation -Cars

-Light Duty Trucks -Heavy Duty Trucks and Buses -Motorcycles -Evaporative Emissions from Vehicles Transformation Total Manufacture

of Solid Fuels and Other Energy Industries -Manufacture of Solid Fuels -Other Energy Industries

(27)

6.2 Us West

1

Table 10: Rapid City

Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -103.0 44.25 402.51 3.39 4.58 6 -103.25 44.25 401.10 3.09 4.28 48 -103.25 44.50 404.05 2.94 3.86 75 -103.25 44.75 404.20 2.27 2.89 13 -103.50 44.75 403.10 5.10 6.28 6 Mean 2.95 1.36 3.97 1.19 148

Table 11: Rapid City EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0028 0.0184 0.0002 0.0067 0.0514 0.0787 0.0826 0.0006 Mineral Non road Oil Production Road

Transfor-mation Total ODIAC

0.0078 0.0133 0.2874 0.0010 0.5512 0.224

1_{In calculating the delta CO}

2 for the individual gridcells a different code was used than

described in the main part of the report. This could lead to a small difference between the mean given in this appendix and the mean calculated from the individual gridcells. In most cases the difference was (much) less than 0.1 ppm, only in the hotspots Los Angeles, Albequerque, Dixie, Khorramabad, Bariq and Gochas this difference was slightly higher, but still less than 0.3 ppm. Unfortunately there was not enough time anymore to look into this. The numbers quoted in the tables below for the mean are obtained using the method in the main text.

(28)

Table 12: Los Angeles Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -119.25 34.0 402.33 2.11 1.77 136 -119.0 34.0 402.66 2.57 2.43 26 -119.0 34.25 403.03 2.94 2.80 7 -119.0 34.5 402.89 2.80 2.67 7 -118.75 34.0 404.34 1.79 1.44 110 -118.50 34.0 402.99 1.67 1.12 187 -118.25 34.0 402.26 3.30 2.57 34946 -118.0 34.0 403.82 3.41 2.67 981 -118.0 33.75 402.68 2.55 1.78 371 Mean 3.08 1.64 2.38 1.59 36771

Table 13: Los Angeles EDGAR & ODIAC in gram CO2/m2/day

Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0007 0.0902 0.0002 1.2457 4.4957 5.8825 1.7673 0.0982 Mineral Non road Oil Production Road

0.1637 0.0738 0.0097 1.7841 0.1855 15.8031 2.456 Table 14: Seattle Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -122.0 46.75 403.76 5.75 6.65 3 -122.0 47.0 402.38 1.59 2.37 39 -122.25 47.0 401.75 1.85 2.67 69 -122.5 47.0 399.89 2.47 2.79 6 -122.0 47.25 402.34 1.75 2.58 7 -122.25 47.25 402.06 1.17 1.74 172 -122.25 47.5 402.74 1.17 1.72 209 Mean 1.40 1.44 2.02 1.56 505

(29)

Table 15: Seattle EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0028 0.2375 0 0.4048 1.6676 2.1562 0.1868 0.032 Mineral Non road Oil Production Road

0.0559 0.0691 0.2298 1.7101 0.0613 6.8328 1.423

Table 16: Yosemite Park

Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -119.25 35.75 402.89 3.39 3.05 145 -119.25 36.0 403.80 4.27 3.74 177 -119.5 36.0 404.24 2.27 2.08 66 -119.25 36.25 404.52 3.01 2.51 53 -119.5 36.25 403.60 3.46 2.89 141 -119.5 36.5 403.76 2.96 2.37 110 -119.5 36.75 404.15 3.45 2.91 27 -119.75 36.75 403.98 2.41 2.16 97 -119.75 37.0 404.04 3.90 3.64 30 -119.75 37.25 404.19 2.81 2.73 5 -120.0 37.25 403.03 2.94 2.80 3 Mean 3.34 1.69 2.90 1.74 854

Table 17: Yosemite park EDGAR & ODIAC in gram CO2/m2/day

(30)

Table 18: Cuymaca Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -116.75 32.75 406.97 3.50 2.78 12 -116.75 33.0 406.26 2.78 2.08 25 -116.75 33.25 406.82 5.07 4.48 21 Mean 3.71 1.48 3.06 1.51 58

Table 19: Cuymaca EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0010 0.0502 0.0021 0.2789 1.018 1.3970 0.0330 0.0219 Mineral Non road Oil Production Road

0.0368 0.0267 0 0.9998 0.0414 3.9065 0.394

Table 20: San Francisco

Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -121.0 36.0 406.97 1.48 1.18 94 -120.75 36.0 406.26 1.85 1.73 92 -121.0 36.25 406.82 1.72 1.48 81 Mean 1.66 1.43 1.43 1.39 267

Table 21: San Fransisco EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0023 0.0688 0.0003 0 0.0004 0.0018 0 0 Mineral Non road Oil Production Road

(31)

Table 22: Death Valley Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -116.5 36.75 403.05 1.81 1.61 62 -116.75 36.75 403.66 2.00 1.53 88 -116.75 37.0 403.20 2.12 1.74 55 Mean 1.96 1.41 1.61 1.44 205

Table 23: Death Valley EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0002 0.0287 0 0 0.0003 0 0 0 Mineral Non road Oil Production Road

0 0.0032 0 0.1243 0 0.1569 0.018 Table 24: Sullivan Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -101.0 35.75 402.40 2.40 2.65 52 -101.0 36.0 403.04 3.04 3.28 63 Mean 2.75 0.92 3.00 0.92 115

Table 25: Sullivan EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0026 0.0345 0 0 0.0002 0.0007 0 0 Mineral Non road Oil Production Road

(32)

Table 26: Albequerque Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -107.75 34.75 406.86 2.57 2.45 12 -107.75 35.0 407.87 3.57 3.46 3 Mean 2.88 2.01 2.76 2.01 15

Table 27: Albequerque EDGAR & ODIAC in gram CO2/m2/day

Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0004 0.0783 0 0 0.0026 0.0110 0 0 Mineral Non road Oil Production Road

0 0.0100 0 0.1151 0 0.2174 0.030 Table 28: Dixie Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -111.5 36.5 397.74 2.02 2.05 6 -111.5 36.75 400.59 3.31 3.35 33 Mean 2.82 1.63 2.82 1.72 39

Table 29: Dixie EDGAR & ODIAC in gram CO2/m2/day

Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0.0844 0 0.0002 0.0093 0.0367 0 0 Mineral Non road Oil Production Road

(33)

Table 30: Long Beach Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -124.0 46.0 405.58 4.15 4.46 21 -124.0 46.25 405.25 3.55 3.83 10 Mean 4.02 1.59 4.32 1.71 31

Table 31: Long Beach EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0008 0.0007 0 0.0014 0.0215 0.0760 0 0.0001 Mineral Non road Oil Production Road

0.0002 0.0267 0 0.7370 0.0002 0.8723 0.136

Table 32: Olympic National Park

Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -123.5 46.75 405.30 4.87 4.96 7 -123.5 47.0 404.46 4.03 4.11 6 -123.75 47.0 403.49 3.06 3.14 5 -123.75 47.25 404.73 4.30 4.38 3 Mean 4.06 0.97 4.14 0.97 21

Table 33: Olympic National Park EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0009 0.0091 0 0.0077 0.0316 0.0389 0 Mineral Non road Oil Production Road

(34)

Table 34: Portland Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments -122.75 44.5 403.22 2.78 2.87 9 -122.75 44.75 403.03 2.63 3.02 75 -123.0 44.75 403.27 2.92 3.00 5 -123.25 45.0 403.85 3.75 3.52 4 -123.5 44.5 400.72 3.92 4.33 14 -123.5 44.75 403.55 3.00 3.41 38 Mean 2.78 2.03 3.15 2.82 145

Table 35: Portland EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0098 0.0166 0.0002 0.0503 0.3547 0.4523 0.1355 0.0041 Mineral Non road Oil Production Road

(35)

6.3 Middle East

Table 36: Teheran Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 49.5 37.0 400.584 3.38 3.43 6 49.5 37.25 403.516 1.68 1.20 43 49.75 37.0 403.807 2.44 2.15 45 49.75 37.25 402.481 2.11 1.79 33 Mean 2.13 1.38 1.78 1.61 127

Table 37: Teheran EDGAR & ODIAC in gram CO2/m2/day

0.0405 0.0010 0.0138 7.0762 0.737 Table 38: Kashan Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 51.25 33.75 401.997 4.99 4.39 49 51.25 34.0 402.102 5.92 5.23 28 51.25 34.25 401.997 1.28 0.74 50 Mean 3.76 2.95 3.16 2.82 127

(36)

Table 39: Kashan EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0013 0.0032 0 0.0218 0.2625 0.4237 0.0075 0.0010 Mineral Non road Oil Production Road

0.0037 0 0 0.0069 0.0013 0.7329 0.308 Table 40: Sabsevar Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 58.25 36.25 407.080 5.26 4.70 18 58.25 36.5 402.676 2.97 2.41 14 58.25 36.75 404.553 3.46 3.13 8 58.25 37.0 404.371 2.90 2.57 3 Mean 3.96 2.12 3.48 2.07 43

Table 41: Sabsevar EDGAR & ODIAC in gram CO2/m2/day

Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0021 0.0019 0 0.0042 0.1340 0.2272 0 0.0002 Mineral Non road Oil Production Road

0.0007 0 0 0.0076 0.0003 0.3781 0.205

Table 42: Dinar Kuh

Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 47.5 32.75 403.130 2.47 2.19 36 47.5 32.5 406.561 4.20 3.59 20 47.25 32.5 399.603 2.92 2.34 64 47.25 32.25 402.667 3.33 2.83 20 47.5 32.25 405.458 4.09 3.47 16 Mean 3.06 1.31 2.56 1.22 156

(37)

Table 43: Dinar Kuh EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0002 0.0012 0 0 0.0055 0.0085 0 0 Mineral Non road Oil Production Road

0 0 0 0.0034 0 0.0188 0.696 Table 44: Khorramabad Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 46.75 33.5 401.753 2.18 2.12 13 47.0 33.5 403.443 4.09 3.58 11 47.0 33.75 402.844 3.19 2.47 59 47.25 33.5 402.513 2.17 1.58 18 47.25 33.25 404.982 3.81 3.23 30 Mean 2.96 2.48 2.48 2.44 131

Table 45: Khorramabad EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0171 0.0071 0 0 0.0913 0.1592 0 0 Mineral Non road Oil Production Road

0.0334 0 0 0.0054 0 0.3136 0.231 Table 46: Baghdad1 Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 43.5 34.0 400.662 1.55 0.99 68 43.75 33.75 400.569 1.48 0.92 99 43.75 33.5 401.532 1.44 0.90 117 44.0 33.5 400.445 2.53 2.56 66 Mean 1.65 1.35 1.22 1.42 350

(38)

Table 47: Baghdad1 EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0016 0.0106 0.0007 0.1544 0.0661 0 0 Mineral Non road Oil Production Road

0 0 0.0600 0.2949 0 0.5883 0.122 Table 48: Baghdad2 Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 48.25 29.5 400.931 4.24 3.66 7 48.25 29.75 403.585 2.41 1.81 55 48.5 29.75 402.454 2.22 1.92 59 Mean 2.44 1.31 1.97 1.35 121

Table 49: Baghdad2 EDGAR & ODIAC in gram CO2/m2/day

Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0034 0.0120 0.0767 0.5069 0.9116 0.3560 0 0 Mineral Non road Oil Production Road

0.4659 0 0 1.9628 0 4.2951 0.609

Table 50: Kuwait City

Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 48.25 29.5 400.931 4.24 3.66 7 48.25 29.75 403.585 2.41 1.81 55 48.5 29.75 402.454 2.22 1.92 59 Mean 2.44 1.31 1.97 1.35 121

(39)

Table 51: Kuwait City EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0007 0.0104 0 0.0569 0.5811 0.0513 0 0 Mineral Non road Oil Production Road

0.0004 0 0 0.2527 0 0.9666 2.112 Table 52: Mosul Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 42.75 36.25 401.045 2.40 1.91 44 42.75 36.5 403.441 4.23 3.73 21 43.0 36.25 402.008 2.73 2.73 16 43.0 36.5 400.395 2.48 2.48 13 Mean 2.75 1.13 2.42 1.12 94

Table 53: Mosul EDGAR & ODIAC in gram CO2/m2/day

Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0024 0.0154 0.2001 0 0.8833 0.3344 0.0439 0 Mineral Non road Oil Production Road

0.0017 0 0.0176 0 0 2.5561 0.173 Table 54: Sinjar Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 42.0 35.25 407.135 2.57 2.14 16 42.0 35.5 407.391 2.82 2.39 18 42.0 35.75 402.155 1.95 1.99 16 Mean 2.43 0.67 2.17 0.53 50

(40)

Table 55: Sinjar EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0.0133 0 0 0.0689 0.0295 0 0 Mineral Non road Oil Production Road

0 0 0 0.0263 0 0.1380 0.002 Table 56: Bariq Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 41.75 18.75 406.127 2.82 2.19 31 41.75 19.0 401.744 2.80 2.18 31 Mean 2.51 1.49 1.93 1.40 62

Table 57: Bariq EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0002 0.0009 0 0 0.0991 0.0081 0 0 Mineral Non road Oil Production Road

0 0 0 0.3193 0 0.4277 0.262 Table 58: Ibb Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 43.5 13.5 404.670 2.35 3.34 33 43.5 13.75 404.440 2.41 3.35 28 Mean 2.38 1.09 3.34 1.07 61

(41)

Table 59: Ibb EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0006 0.0004 0 0 0.0227 0.1600 0 0 Mineral Non road Oil Production Road

0 0 0 0.0693 0 0.2565 0.055 Table 60: Khuff Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 37.0 25.25 407.862 2.05 2.77 80 37.0 25.5 404.519 2.60 2.36 37 37.0 25.75 402.569 1.89 1.52 53 Mean 2.07 1.06 2.42 1.06 170

Table 61: Khuff EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0.0023 0 0 0.0119 0.0010 0 0 Mineral Non road Oil Production Road

0 0 0 0.2593 0 0.2745 0.061

Table 62: Jebel Jar

Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 38.25 24.25 400.752 2.48 2.77 30 38.25 24.5 401.326 2.37 2.48 42 38.5 24.75 404.931 2.95 3.01 43 38.5 25.0 405.804 1.57 2.14 38 Mean 2.41 1.43 2.60 1.29 153

(42)

Table 63: Jebel Jar EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0.0060 0 0 0.0200 0.0016 0.5914 0 Mineral Non road Oil Production Road

0 0 0 0.2193 0 0.8384 0.262 Table 64: Khaybar Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 40.0 25.0 405.874 2.88 2.66 23 40.0 25.25 405.486 2.65 2.47 92 40.0 25.5 404.007 2.24 2.22 102 39.75 25.75 405.176 1.65 1.79 129 Mean 2.18 1.23 2.17 1.18 346

Table 65: Khaybar EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0.0016 0 0 0.0184 0 0 0 Mineral Non road Oil Production Road

0 0 0 0 0 0.216 0.039 Table 66: Medina Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 40.25 24.5 404.129 1.27 0.64 1.34 0.62 76

(43)

Table 67: Medina EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0.0062 0 0 0.1993 0.0016 0 0 Mineral Non road Oil Production Road

0 0 0 0.0767 0 0.1044 0.075 Table 68: Ca¨ıro Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 31.25 29.5 401.870 1.54 1.33 247 31.25 29.75 403.125 2.31 2.05 217 31.25 30.0 402.460 1.91 1.84 166 Mean 1.89 1.37 1.70 1.40 630

Table 69: Ca¨ıro EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0494 0.4615 0 0.7043 5.2931 3.4268 19.1724 0.0055 Mineral Non road Oil Production Road

(44)

6.4 South Africa

Table 70: Kaapstad Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 18.75 -34.5 397.940 2.31 1.21 1.94 1.21 7

Table 71: Kaapstad EDGAR & ODIAC in gram CO2/m2/day

Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0 0 0 0.0025 0.0015 0 0 Mineral Non road Oil Production Road

0 0 0 0.0175 0 0.0415 0.448 Table 72: Knysna Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 22.75 -34.25 400.146 2.15 2.00 10 23.0 -34.25 403.545 3.13 3.34 11 23.25 -34.25 401.426 2.00 2.16 92 22.75 -34.0 400.670 1.91 1.71 14 23.0 -34.0 401.832 1.06 1.15 39 23.25 -34.0 401.352 2.02 1.90 44 Mean 1.88 1.63 1.93 1.54 210

(45)

Table 73: Knysna EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0.0069 0.0002 0.0001 0.0193 0.0118 0 0.0001 Mineral Non road Oil Production Road

0 0.0006 0.0001 0.0455 0 0.0877 0.028 Table 74: Richardsbaai Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 31.5 -28.75 398.595 2.61 2.89 122 32.25 -28.5 401.698 2.59 2.14 22 32.25 -28.25 401.659 2.40 2.33 107 32.5 -28.25 401.338 2.42 2.35 127 Mean 2.47 1.45 2.42 1.46 378

Table 75: Richardsbaai EDGAR & ODIAC in gram CO2/m2/day

Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0007 0.0018 0.0012 0 0.0748 0.04517 0.0102 0 Mineral Non road Oil Production Road

0 0.0010 0.0946 0.0715 0 0.3014 0.135 Table 76: Quelimane Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 36.0 -18.75 404.191 2.40 2.47 7 36.25 -18.75 401.760 1.60 1.15 347 36.25 -18.5 401.561 1.33 1.94 219 36.5 -18.5 399.643 2.52 2.07 7 Mean 1.53 1.08 1.11 1.13 580

(46)

Table 77: Quelimane EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0.0008 0.0046 0 0.0013 0.0006 0 0 Mineral Non road Oil Production Road

0 0 0 0.0024 0 0.0108 0.000

Table 78: Kgori Safaris

Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 24.75 -19.75 400.661 3.99 4.26 5 25.0 -19.75 398.560 2.68 2.87 27 25.0 -19.5 400.015 3.83 3.90 7 Mean 3.01 1.09 3.20 1.16 39

Table 79: Kgori Safaris EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0.0009 0 0 0.0024 0.0004 0 0 Mineral Non road Oil Production Road

0 0 0 0.0052 0 0.0088 0.000 Table 80: Garborone/Pretoria Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 27.0 -24.25 406.030 3.71 3.90 6 27.25 -24.25 403.538 1.24 1.37 56 27.5 -24.5 403.158 1.07 1.18 100 27.5 -24.75 402.456 1.91 2.32 77 27.5 -25.0 401.669 0.72 0.79 95 27.5 -25.25 404.735 4.66 5.67 6 Mean 1.33 1.87 1.52 2.16

(47)

Table 81: Garborone/Pretoria EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0.0009 0.0040 0.0034 0.0152 0.0092 0 0 Mineral Non road Oil Production Road

0 0 0 0.0588 0 0.0916 0.052

Table 82: Lower Zambezi National Park

Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 30.0 -15.25 407.120 4.36 3.64 5 30.0 -15.5 406.143 3.38 2.67 14 30.5 -15.25 398.484 1.01 0.79 23 Mean 2.26 1.73 1.81 1.54 42

Table 83: Lower Zambezi National Park EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0.0013 0.2285 0 0.0004 0.0002 0 0 Mineral Non road Oil Production Road

0 0 0 0 0 0.2306 0.001 Table 84: Gochas Lon Lat Mean absolute value in ppm Delta CO2: large back-ground in ppm SD Delta CO2: small back-ground in ppm SD Number of measure-ments 18.25 -25.25 403.477 5.30 5.77 6 18.25 -25.5 401.695 2.18 2.93 19 Mean 2.68 1.50 3.42 1.50 25

(48)

Table 85: Gochas EDGAR & ODIAC in gram CO2/m2/day Agri-cultural Aviation Biomass burning Chemical solvents Combustion manu-facturing Energy buildings Energy industry Fugitives 0 0 0 0 0 0.0003 0 0 Mineral Non road Oil Production Road

Finding local anthropogenic CO2 hotspots in OCO-2 data