Earth’s atmospheric temperature (T) fluctuates because of “natural factors”, such as changes in solar activity and volcanic eruptions. However, simulations based on the latest climate models reveal that even when natural factors are considered, the temperature increase since the mid-19th century cannot be reproduced. Furthermore, only by adding the anthropogenic factor of greenhouse gas carbon dioxide (CO₂) emissions can results that are closer to actual observational data be obtained on the basis of the IPCC’s (Intergovernmental Panel on Climate Change) Sixth Assessment Report [1].

The report used the extremely strong statement that “there is no doubt that human influence has caused global warming.” This is because the history of climate change over the past several thousand years, physical mechanisms, and the results of advanced computer analysis suggest that “human activity is the cause”. In short, the consensus is that “humans are emitting CO₂ at a rate that exceeds the natural CO₂ cycle, enveloping the Earth like a blanket and preventing heat from escaping”.

Looking back over the Earth’s long history, including ice ages spanning hundreds of thousands of years, the relationship between T and CO₂ may not be a simple one-way street where one comes first but rather may have a mutually amplifying feedback structure. When humans burn fossil fuels, atmospheric CO₂ concentrations first increase, and this increase in CO₂ produces a greenhouse effect, which in turn causes T to increase. In other words, the causal relationship may proceed in the following order: anthropogenic CO₂ emissions → increase in CO₂ concentrations → increase in T. In this case, in modern global warming, the increase in CO₂ concentrations precedes the increase in T.

The cross correlation between T and CO₂ is among the fundamental starting points for addressing the issue of climate change. The commonly assumed relationship, as asserted by the IPCC, is that increases in CO₂ cause increases in T. However, analyses since 1990 have shown that the generally assumed relationship is in the opposite direction [2]-[10], casting doubt on this assumption. All the evidence from these analyses suggests a unidirectional relationship, with T as the cause and CO₂ as the effect. This relationship is not represented in climate models, which show an inverse correlation that is contradicted by actual measurements. These analyses revealed that CO₂ fluctuations lagged behind correlated changes in T by severalmonths [2]-[10].

Therefore, in a series of recent studies, the correlation, causality, and causes of the correlation between T and CO₂ have been investigated [11]-[18]. The cause and core of the correlation is that, in contrast to the commonly held belief that “T increases are primarily caused by increases in CO₂ due to human activity,” we have confirmed, using the latest observational data and statistical time series analysis, that “changes in T precede changes in CO₂ concentration” with a time lag. The results of these series of studies are reviewed in some detail below. To subsequently verify the results obtained and the mechanisms of global warming, we will examine several findings currently being gathered through satellite observations.

Summaryofpreviousstudies

The correlation coefficient r between two variables, x and y, is defined by Equation (1) below. r ranges from −1 to 1, where the closer r is to 1, the stronger the correlation. As a guideline, a correlation coefficient of 0.8 or higher indicates a strong correlation, a correlation coefficient of 0.7 - 0.5 indicates a fair correlation, and a correlation coefficient of 0.4 - 0.3 indicates little correlation. The correlation coefficient r can be calculated relatively easily using the built-in function in Microsoft Excel^®.

Murry Salby analyzed the temperature change (ΔT) and the rate of change in CO₂ concentration (ΔCO₂) and proposed that the relationship between ΔCO₂ and ΔT can be expressed as Equation (2) [19][20]:

d(ΔCO₂)/dt = γΔT (γ: a constant) (2)

The ΔCO₂ (annual change) values are reported by NOAA as visualized on their website [21], which they call the CO₂ growth rate. The ΔT (monthly average anomalies) and the growth rate during 1979/7 and 2023/12 are shown in Figure 1(a) [18]. The correlation coefficient r is 0.744. Although the high r can be interpreted as indicating a very good correlation, it should be noted that the same annual change value ΔCO₂ is used throughout each year. The power spectral density obtained via Fourier transforms for ΔT and ΔCO₂ is shown in Figure 1(b) [18]. A regular frequency pattern is observed in both. Furthermore, upon comparison of the double-headed arrows on the two power spectra in Figure 1(b), a phase shift is observed between these frequencies.

A good correlation between ΔT and ΔCO₂ is shown in Figure 1(a), but effective analysis is not possible when the time lag between these variables is less than 12 months. Therefore, we compared the 13-month monthly average ΔT with the 13-month monthly average ΔCO₂. It is necessary to download NOAA data and calculate the 13-month monthly average of ΔCO₂. As shown in Figure 2(a), the correlation coefficient r between the two is 0.664 [17], indicating that ΔCO₂ changes with lag time (months). The value at which the correlation coefficient r is maximized indicates that ΔCO₂ changes with lag time. Therefore, Equation (2) can be rewritten as Equation (3).

d(ΔCO₂)/dt = γΔT’ (ΔT’ = ΔT when a lag time is considered) (3)

The power spectrum represents the intensity of the vibrational components of a signal at each frequency. However, it does not contain information about the phase (time difference), so it cannot determine the direction of the time difference. To quantitatively analyze the time difference, a cross-correlation function or cross-spectrum (cross-spectral density) must be used. Dedicated software is required to automate calculations using the cross-correlation function or cross-spectrum.

(a)

(b)

Figure1. (a) Correlations of the global temperature anomaly (red line, scale: left axis, ˚C) and 12-month average annual CO₂ growth rates (blue bar, scale: right axis, ppm/year); (b) comparison of the power spectral density (PSD) obtained via Fourier transform for the global temperature anomaly and 12-month average annual CO₂ growth rates (the horizontal axis represents the number of data points from the start of the analysis. The number of data points is the same as the number of months [18].

(a)

(b)

Figure2. (a) Correlations between global temperature anomalies (red line, scale: left axis, ˚C) and monthlyCO₂ annual growth rates (ΔCO₂, blue line, scale: right axis, ppm/year) and (b) changes in correlation coefficients with time lag (in months) [17].

Satellite measurements of Earth’s temperature began in 1979, providing reliable information on ΔT over a large surface area. Moreover, long-term measurements of CO₂ concentrations have been ongoing since 1957 at an observatory on Mauna Loa, Hawaii, providing reliable ΔCO₂ data. As shown in Figure 2(a), these reliable data clearly show that ΔT and ΔCO₂ have been well correlated since at least 1979, but ΔT leads ΔCO₂. The relationship between r and time lags is shown in Figure 2(b). ΔT leads ΔCO₂ by approximately four months when the value of the maximized risconsidered. Whether ΔT leads ΔCO₂ on longer timescales remains a challenge for further observation. If a change in the time lag does occur, it should be possible to detect it using the power spectrum.

When time series data are analyzed using a Fourier transform, if the time lag reverses midway, the “phase difference (phase spectrum)” calculated from the power spectrum will also reverse accordingly. However, because a normal Fourier transform processes the “entire period” all at once, some ingenuity is required to capture phenomena where the properties change midway. The details are beyond the scope of this paper, but one example is a method of dividing the data into short windows and performing a Fourier transform while shifting those windows. This makes it possible to track changes such as “the phase is positive in this time period and negative in the next time period.”

ΔT and ΔCO₂ have been well correlated since at least 1979, but ΔT leads ΔCO₂ by approximately four months. The fact that ΔT precedes ΔCO₂ was also confirmed using a slightly different method [14]. ΔT and ΔCO₂ were analyzed before and after ENSO events. Owing to seasonal changes in photosynthesis and soil respiration (Rs), CO₂ concentrations follow a regular pattern, reaching their lowest values in August and their highest values in May each year. Therefore, as shown in Figure 3, we considered the CO₂ concentration and T changes over one year from September as a cycle for each year. T increases with the occurrence of El Niño. CO₂ concentrations also increase with increasing temperature. The values of ΔCO₂ and ΔT were calculated by comparing the El Niño period with the period before occurrence and were analyzed according to Equation (2). The correlation coefficient r was maximized by shifting ΔT by five months (Figure 4(a) and Figure 4(b)). ΔT and ΔCO₂ in Figure 4(b) overlap well when ΔTisshifted by five months. This indicates that ΔCO₂ is determined by ΔT, and the time lag is five months.

Figure3. Changes in the annual CO₂ concentrations (ppm) from September 2014, 2015, or 2016 to August 2015, 2016, or 2017 [14].

(a)

(b)

Figure4. (a) Deviations in ΔCO₂ and ΔT between 2015-2016 and 2014-2015; (b) the same data are plotted, but ΔT is shifted to the right by five months. The correlation coefficient r is improved from 0.47 to 0.94 by shifting [14].

As demonstrated by the findings outlined above, since satellite-based observations of T began in 1979, changes in ΔCO₂ have consistently lagged behind changes in ΔT by a period of four to five months. Specifically, when T rose, CO₂ increased; conversely, when T fell, CO₂ decreased. During the major El Niño event of 1997-1999, as illustrated in Figure 5, ΔT shifted by 0.5˚C, whereas ΔCO₂ shifted by 2 ppm. Although the Earth released CO₂ when T rose, it subsequently absorbed that CO₂ back into the system as T cooled and returned to its baseline level.

Figure5. Changes in global temperatures and CO₂ concentration growth rates during El Niño events from 1997-99.

ΔCO₂ has consistently lagged behind ΔT by a period of four to five months since 1979. Therefore, whether anthropogenic ΔCO₂ affects ΔT is questionable. This question is also supported on the basis of a mass balance in the carbon cycle. The carbon cycle budget reported by the IPCC [1] is shown in Figure 6. Some critical numbers regarding anthropogenic CO₂ are summarized here [12].

Figure6. Simplified carbon cycles and carbon equivalent estimates (Unit: GtC) obtained from the IPCC report [12].

1) The carbon cycle budget shows that anthropogenic CO₂ accounts for only 4% of the total:

anthropogenic CO₂ ratio

≒ (fossil fuel combustion)/(fossil fuel combustion + respiration and decomposition + ocean atmosphere exchange)

≒ (7.8)/(7.8 + 107.2 + 79.2)

≒ 0.04 (4)

2) The residence time of CO₂ is approximately 4 years:

CO₂ residence time

= (CO₂ in the atmosphere)/(fossil fuel combustion + respiration and decomposition + ocean-atmosphere exchange)

≒(829)/(7.8 + 107.2 + 79.2)

≒4 (5)

3) CO₂ is only 4% of greenhouse gas, and the remaining 96% is H₂O:

CO₂ concentration in the greenhouse gas

≒(0.04)/(1 + 0.04)

≒0.04 (6)

Next, the results from our papers are summarized to investigate why ΔCO₂ has lagged behind ΔT by a period of four to five months since 1979. First, ΔCO₂ and ΔT at various latitudes were analyzed [11].

There are two major differences between the Northern and Southern Hemispheres [22]. The first, as illustrated in Table1, lies in the disparity between their respective oceanic and land areas. Approximately 68% of the Earth’s total landmass is concentrated in the Northern Hemisphere, where the proportion of land is more than double that of the Southern Hemisphere. In the Southern Hemisphere, however, more than 80% of the surface area is covered by ocean. In particular, the region between 40˚ and 60˚ south latitude is renowned as a zone where powerful westerly winds sweep across the sea unimpeded by landmasses.

Table1. Proportions of ocean and land areas in the northern and southern hemispheres [22].

Because land heats up and cools down more rapidly than water does, the Northern Hemisphere, which contains a greater proportion of landmass, tends to experience larger annual temperature fluctuations than the Southern Hemisphere. The Southern Hemisphere is characterized by the ease with which large-scale ocean currents develop, such as the Antarctic Circumpolar Current, precisely because it is not obstructed by massive continents.

The second point concerns the disparity in forest area between the Northern and Southern Hemispheres. Next, on the basis of data such as the latest Global Forest Resources Assessment (FRA 2020) by the Food and Agriculture Organization (FAO) of the United Nations [23], we compiled the forest area ratios. The total global forestland area is approximately 4 billion hectares (approximately 31% of the total land area). When categorized into “tropical rainforests” and “other forests,” breakdown generally yields the following ratios (Table2).

Table2. Proportion of “Tropical Rainforests” and “Other Forests” [23].

Note: Among the “other forests” category, “subarctic forests” (boreal forests), which occupy a particularly vast area across the Northern Hemisphere, alone account for approximately 27% of the total.

The Northern Hemisphere is characterized by its extensive landmass, which encompasses the vast coniferous forest belts of Siberia and Canada; consequently, the proportion of “other forests” is overwhelmingly high. Tropical rainforests account for approximately 25% of the total, covering regions such as Central America, parts of North Africa, and portions of Southeast Asia. The remaining 75% consists of “other forests,” comprising vast temperate forests as well as boreal forests, the largest forest zone on Earth. In contrast, while the landmass of the Southern Hemisphere is limited, the vast majority of its existing forests are located within the tropics. Tropical rainforests constitute approximately 70% of this total, with massive rainforest ecosystems, such as those covering the majority of the Amazon Basin, the Congo Basin, and Indonesia, concentrated in these regions. “Other forests” make up the remaining 30%, consisting of temperate and arid-zone forests found in countries such as Australia, Chile, Argentina, and South Africa. This creates a contrasting structural dynamic: the Northern Hemisphere serves as a “treasure trove of vast boreal and temperate forests,” whereas the Southern Hemisphere acts as a “stronghold of tropical rainforests”. In particular, the “other forests” of the Northern Hemisphere, specifically its boreal forests, play a role that is comparable to that of tropical rainforests, not only as carbon dioxide sinks but also in terms of the immense quantities of carbon stored within their soils.

On the basis of the aforementioned differences between the Northern and Southern Hemispheres, we can examine the variations in ΔCO₂ and ΔT at various latitudes, as presented in Table3. Among these observations, points #4 and #5 in Table3 are crucial for addressing the following question: “Why does ΔCO₂ lag behind ΔT by four to five months?” Specifically, the magnitude of ΔCO₂ at approximately 50˚ north latitude is greater than that at the equator.

Table3. Summary of the results obtained in study [11].

Furthermore, the following results have been obtained [12].

1) δ¹³C and ΔCO₂ are inversely correlated, which is known as the Suess effect [24]. However, natural CO₂, rather than anthropogenic CO_2, may affect the Suess effect more because anthropogenic CO₂ accounts for only 4%, as shown above.

2) The extent of the correlation between d(ΔCO₂)/dt and ΔT differs depending on the latitude and between the land and sea.

3) During El Niño, d(ΔCO₂)/dt follows ΔT with a time lag of several months, and CO₂ emission and absorption at the Earth’s surface respond to ΔT.

4) The concentrations of CO₂, CH₄, and N₂O gases increase annually, but seasonal changes are observed. These concentrations decrease from spring to summer but increase from fall to winter.

5) Rs is interpreted to be activated in spring because of increasing temperatures and to generate CO₂, CH₄, and N₂O in fall because of biological processes after a time lag.

6) The temperature patterns have changed over the last 2000 years, as reflected by the ice age and warm periods. Therefore, CO₂ has evolved to breathe slowly in response to ΔT.

7) Plant decomposition and Rs are accompanied by microbial processes and increased soil fertility, and Earth is becoming greener because of rising CO₂ and increasing fertility.

On the basis of these results, we conclude that changes in plant decomposition and Rs due to global temperatures primarily control global CO₂ cycles. The impact of CO₂ emissions from fossil fuel combustion on global warming is extremely low. For these reasons, the man-made global warming hypothesis needs to be carefully reinvestigated.

As summarized above, d(ΔCO₂)/dt follows ΔT with a time lag of several months, and CO₂ emission and absorption at the Earth’s surface respond to ΔT. The increase in ΔCO₂ with increasing ΔT can be regarded as thermallyinducedCO_2, which may be significantly related to Rs.

We investigated the impact of selected global conditions on Rs, drawing upon the latest NASA databases [13]. The variation in the annual carbon flux derived from Rs (g∙C∙m⁻²) in relation to the annual mean temperature (˚C) over the period from 1961 to 2017 is shown in Figure 7. Rs has an approximately linear relationship with temperature. As noted above, ΔT influences Rs, thereby triggering changes in CO₂ production. The results presented in Figure 7 demonstrate a positive correlation between Rs and temperature, thereby substantiating the aforementioned hypothesis.

Figure7. Change in annual C flux (g∙C∙m⁻²) from soil respiration (Rs) versus mean annual temperature (˚C) between 1961 and 2017 (coefficient of determination (r²): 0.172). The regression red line is y = 23.3x + 582.0 [13].

While CO₂ concentrations have been increasing annually, seasonal variations have been observed in previous studies [13]. Specifically, CO₂ concentrations decrease from spring to summer but increase from autumn to winter. Rs is interpreted as becoming active in the spring in response to rising temperatures and subsequently generating CO₂ through biological processes in the autumn following a certain time lag [13]. The seasonal variations in Rs flux (mean value: μmol∙m⁻²∙s⁻¹) for spring, summer, autumn, and winter at the observation sites within the United States are listed in Table4. The values on the vertical (y) axis in Figure 8 represent ΔRs (=Rs flux − mean Rs flux) at each observation site. These results corroborate the interpretation of previous studies, namely, Rs becomes active in the spring, reaches a maximum value in the summer, and declines in the winter. Furthermore, these findings suggest that a time lag occurs in the fluctuations in CO₂ concentrations.

Figure8. Mean seasonal Rs flux (μmol∙m⁻²∙s⁻¹) in spring, summer, autumn, or winter in the U.S. The vertical values are the ΔRs (Rs flux − averaged Rs flux) at each site [13].

Table4. Mean seasonal Rs flux (μmol∙m⁻²∙s⁻¹) in spring, summer, autumn, or winter at US sites. (ΔRs = Rs flux − averaged Rs flux) [12].

In theRscontrolprocess, the increase in temperature due to the modern warm period increases CO₂ emissions becauseoftheincreaseinRs, as presented in Figure 9 [13]. One of the main factors is that the Rs control process in the midlatitudeforestzone changes significantly because of temperature changes, such as those in temperate forests in Olympic National Park, WA (USA) (Figure 10). The emitted CO₂ can be considered thermally induced CO₂. As a result, the CO₂ concentration in the atmosphere increases. Therefore, although there is a cross-correlation between temperature and CO₂ concentration, atemperature-leadingtimelag is observed because it is a process mediated by Rs. Even though anthropogenic CO₂ has decreased, reducing total atmospheric CO₂ concentrations during the modern warm period is difficult. Additionally, this means that an increase in anthropogenic CO₂ since the Industrial Revolution has contributed too little to affecting the global CO₂ concentration.

Figure9. The Rs control process causes ΔCO₂ because of ΔT followed by changes in Rs [13].

Figure10. Temperate forests in Olympic National Park, WA (USA) (photographed by the author, March 4, 2025).

The temperature tends to increase more in the north (20 N - 90 N) than in the south (20 S - 90 S), as shown above (see Table3). Additionally, compared with that in the tropics, the rate of change in the CO₂ concentration at a sine latitude of 0.75 (≒50 N) corresponds to the degree of temperature change. For these reasons, temperate forests play a critical role in controlling Rs.

SatellitedatawithrespecttoRsandCO₂emissionsinhigh-latituderegions

The impact of T fluctuations on CO₂ emissions is more pronounced in temperate forests than in tropical rainforests, where T variability is lower. Because temperate forests experience lower T than tropical rainforests, their rate of Rs is slower, resulting in carbon being sequestered within the soil over extended periods. However, as global warming causes T to increase, the decomposition rate of carbon-containing components sequestered in the soil accelerates. The impact of ΔT is more significant in temperate forests than in tropical rainforests, which exhibit less T variability. Therefore, the effects of global warming on the CO₂ balance differ between temperate forests and tropical rainforests, as determined by the following equation.

CO₂ Balance = Photosynthesis Rate − (Rs + Others)(7)

As T increases, the rate of Rs increases. Generally, the magnitude of the T dependence of Rs is characterized by the following equation [25]:

( R 1 and R 2 are the respiration rates at temperatures T 1 and T 2 (˚C), respectively.)

A Q₁₀ of 2 means that the Rs rate doubles for every 10˚C increase. In many ecosystems, the Q₁₀ for Rs is ~1.5 - 2.0 (see Table5) [26]. Other studies have also shown a tendency for Q₁₀ values to increase with increasing latitude [27]. On the basis of data from chamber observations, a method involving placing a container over the ground to measure emissions spanning thousands of sites, the report establishes the statistical fact that Q₁₀ values are higher at higher latitudes.

Table5. Comparison of Rs and Q₁₀ in temperate and tropical rainforests [26].

With respect to Rs and CO₂ emissions, phenomena observable through satellite data, observational techniques and analytical methods have been steadily accumulating in recent years. Analyses utilizing gross/net primary production (GPP/NPP) data from NASA’s MODIS instruments (the Terra and Aqua satellites) have revealed a phenomenon at mid-to-high latitudes: while increasing T in early spring stimulates photosynthesis, increasing T in early autumn and during nighttime hours drives an increase in Rs at an even faster rate [28].

Data from OCO-2 and OCO-3 (Orbiting Carbon Observatory satellites), which directly measure atmospheric CO₂ concentrations, reveal whether a specific region acts as a “CO₂ sink” or a “CO₂ source”. Analyses have revealed a phenomenon occurring in temperate forest regions during recent periods of high T, in which CO₂ emissions resulting from Rs exceed CO₂ uptake through photosynthesis, leading to localized increases in CO₂ concentrations [29].

The NOAA Arctic Report Card (2024 edition) presents data indicating that in the permafrost regions of Alaska and Siberia, CO₂ emissions from the soil during the autumn and winter are beginning to exceed CO₂ uptake by vegetation during the summer [30].

The Global Carbon Project (GCP) compiles data on variations in the terrestrial carbon balance by latitude [31]. It provides statistical data concerning the impact that accelerated decomposition of soil organic matter in high-latitude regions has on the global carbon balance.

According to data analysis from OCO-2 and GOSAT (“Ibuki”), the increase in CO₂ concentrations during winter is greater in the high latitudes of the Northern Hemisphere than at the equator [32][33]. The ΔCO₂ values at 50 N and in the tropics between 2019 and 2022 are compared in Figure 11, and the approximate values are shown in Table6. While precise comparisons of ΔCO₂ are difficult, it is evident that compared with those at the equator, CO₂ levels at 50 N show a more pronounced increase in 2019. Although this difference may amount to only approximately 2.5 ppm, as shown above, it represents a significant disparity when viewed against the global average annual increase in CO₂ of 2 ppm.

Table6. Comparison of the ΔCO₂ values at 50 N and in the tropics between 2019 and 2022.

(a)

(b)

Figure11. Global mapping of greenhouse gases retrieved from GOSAT Level 2 products, (a) duration: 4/1/2019-4/30/2019 and (b) duration: 4/1/2022-4/30/2022 [32].

Since 2015, land CO₂ uptake north of 20 N has decreased by half to 1.13 ± 0.24 GtC∙yr⁻¹ by 2023. Moreover, the tropics recovered from the 2015-2016 El Niño carbon loss, gained carbon during the La Niña years (2020-2023), and then switched to carbon loss during the 2023 El Niño (0.56 ± 0.23 GtC∙yr⁻¹) [34].

ConfirmationofGlobalGreeningbySatelliteData

Owing to global warming, CO₂ emissions from soil in high-latitude regions are increasing, as summarized in the previous section. Satellite data also revealed that the world is greener than it was in the early 1980s. The updated maps in Figure 12 show that the trend has continued [35]. This map shows where greenness increased (green) and decreased (brown) across the planet between 2000 and 2018. Specifically, the trends in the leafareaindex (LAI) and the amount of leaf area relative to the ground area during the growing season are shown. There is a clear greening trend in boreal and Arctic regions, which is a result of increasing T. For example, Svalbard in the high Arctic has experienced a 30% increase in greenness. The greening was concurrent with an increase in the mean summer temperature from 2.9˚C to 4.7˚C between 1986 and 2015 [35].

Figure12. Trends in the glowing season mean leaf area index (2000-2018, 10⁻³/m²/year) [35].

Research teams, including those at NASA, have published analytical results corroborating greening by using satellite data [36][37]. Over the approximately 35 to 40 years spanning from the early 1980s to the present day, it has been confirmed that the LAI of the Earth’s vegetation has increased by 25% to 50%. The increase in LAI resulting from this greening is estimated to be approximately twice the land area of the entire United States, or approximately 1.5 times the size of the Amazon rainforest. A map of the world highlighting these newly greened areas is featured in NASA’s explanatory materials [36]. Furthermore, this greening is attributed to the “CO₂ fertilization effect”, which is the process by which increasing concentrations of CO₂ stimulate plant photosynthesis [37].

With the Google Earth Engine (time series comparison), decades’ worth of satellite imagery overlaid as a time lapse can be viewed. If you visit “Google Earth Engine Timelapse” and zoom in on northern Alaska or the vicinity of the Siberian taiga boundary, you can observe how areas that were once dominated by snow and bare rock gradually become covered in green vegetation, specifically shrubs and grasses, year after year.

In the “Terrestrial Snow Cover” and “Tundra Greenness” sections of the NOAA Arctic Report Card, data are presented to demonstrate the correlation between rising temperatures and vegetation change characteristics at high latitudes, as shown in Figure 13 [38].

Figure13. Magnitude of the maximum NDVI trend calculated as the change per decade using ordinary least squares regression for Arctic tundra (solid colors) and boreal forest north of 60˚ latitude (muted colors) during (a) 1982-2023 based on the AVHRR GIMMS 3-g+ dataset and (b) 2000-24 based on the MODIS MCD13A1 v6.1 dataset. In each panel, the circumpolar tree line is indicated by a black line, and the 2024 mean August sea-ice extent is indicated by light shading [36].

The normalized difference vegetation index (NDVI) is the most widely used indicator of “greenness,” utilizing the absorption of red light and the reflection of near-infrared light by plant leaves. The LAI is an indicator of the total leaf area per unit of ground area, thereby reflecting changes in vegetation density. This method is under development [39]. Analysis utilizing Japanese satellite data, such as that from “Shikisai” (GCOM-C), is also currently advancing [40].

Summarized in bullet points, the conclusions are as follows:

1) ΔT precedes ΔCO₂.

2) Global warming increases Rs, leading to increased CO₂ emissions from the soil.

3) The higher the latitude is, the greater the temperature dependence of CO₂ emissions from the soil.

4) The temperature dependence of CO₂ emissions in temperateforests is greater than that in tropicalrainforests.

5) At high latitudes, greening is more temperature dependent because of global warming and the fertilization effect of CO₂.

6) The role of soil respiration in the global carbon cycle has long been recognized. However, amidst the emphasis placed on the impact of anthropogenic CO₂ emissions on global warming, the role of soil respiration tends to be overlooked. The amount of CO₂ generated by soil respiration far exceeds that generated by anthropogenic emissions; furthermore, compared with tropical rainforests, temperate forests exhibit greater temperature dependence in terms of soil respiration because of the magnitude of their temperature fluctuations. Consequently, vegetation, particularly temperate forests, is believed to play a decisive role in influencing global warming.

The role of temperate forests in relation to ΔT and the subsequent ΔCO₂ was highlighted by our recent research findings. Furthermore, this role has now been corroborated by several results derived from satellite observations. Further reports based on satellite measurements and analyses are anticipated in the future.

ENSO: El Niño-Southern Oscillation

IPCC: Intergovernmental Panel on Climate Change (the United Nations body)

NOAA: National Oceanic and Atmospheric Administration

NASA: National Aeronautics and Space Administration

LAI: Leaf Area Index

Rs: Soil Respiration

r: correlation coefficient