Russian forest sequesters substantially more carbon than previously reported

Russia has been reporting almost no changes in forested area, growing stock volume (GSV) and biomass to the United Nations Framework Convention on Climate Change (UNFCCC)¹ and the Food and Agriculture Organization of the United Nations (FAO) Forest Resources Assessment (FRA)² since the collapse of the USSR and the decline in the Soviet Forest Inventory and Planning (FIP) system. According to the State Forest Register (SFR)³, which is the main repository of forest information, and national reporting to the FAO FRA², the GSV and the above ground biomass (AGB) increased by 1.1% and 0.6% (Table S1), respectively, during 1990–2015, yet studies using remote sensing (RS) indicate increased vegetation productivity⁴, tree cover (annual rate + 0.417% over 1982–2016)⁵, increased AGB (+ 329 Tg C yr⁻¹ over 2000–2007⁶), total biomass (annual rate + 0.44% or + 153 Tg C yr⁻¹ over 1990–2007⁷), and forest ecosystem carbon pools (ca + 470 Tg C yr⁻¹ over 2001–2019⁸). This inconsistency in estimates can be explained by an information gap that appeared when Russia decided to move from the FIP to another system for the collection of forest information at the national scale – the National Forest Inventory (NFI).

The FIP involves revisiting every forest stand (on the ground for managed forests or using RS techniques for remote non-commercial forests) on a 10–15-year interval, with the measurement of forest parameters combined with the formulation of forest management directives. After the collapse of the USSR, the inventory within the FIP system slowed down substantially. For example, more than 50% of the forest area was surveyed by the FIP more than 25 years ago⁹. For these reasons, the reliability of information on forests in Russia has deteriorated since 1988, which is the year when FIP-based reporting¹⁰ involved the largest inventory efforts in recent decades. According to this report¹⁰, the total GSV of Russian forests was 81.7 × 10⁹ m³ (without shrubland, bias corrected¹¹). This value is used here as a reference to quantify biomass stock changes in Russia with respect to the current decade.

In contrast, NFI is a state-of-the-art inventory system based on a statistical sampling method. It was initiated in 2007 and the first cycle was completed in 2020. The NFI data processing is ongoing, but the first official press release¹² suggests that Russian forest accumulated 102 × 10⁹ m³ over its lifespan until 2014. Once finalized, the NFI will be verified before adoption as the official source of information to the SFR and national reporting. The NFI has received some criticism¹³ because of the relatively sparse sampling employed and the stratification method used, which is partially based on outdated FIP data.

In Russia, the long intervals between consecutive surveys and the difficulty in accessing very remote regions in a timely manner by an inventory system make satellite RS an essential tool for capturing forest dynamics and providing a comprehensive, wall-to-wall perspective on biomass distribution. However, observations from current RS sensors are not suited for producing accurate biomass estimates unless the estimation method is calibrated with a dense network of measurements from ground surveys¹⁴. Here we calibrated models relating two global RS biomass data products (GlobBiomass GSV¹⁵ and CCI Biomass GSV¹⁶) and additional RS data layers (forest cover mask⁹, the Copernicus Global Land Cover CGLS‐LC100 product¹⁷) with ca 10,000 ground plots (see Material and Methods) to reduce nuances in the individual input maps due to imperfections in the RS data and approximations in the retrieval procedure^18,19. The combination of these two sources of information, i.e., ground measurements and RS, utilizes the advantages of both sources in terms of: (i) highly accurate ground measurements and (ii) the spatially comprehensive coverage of RS products and methods. The amount of ground plots currently available may be insufficient for providing an accurate estimate of GSV for the country when used alone, but they are the key to obtaining unbiased estimates when used to calibrate RS datasets²⁰. The map merging procedure was preferred over a plot-aided direct estimation of GSV or AGB from the RS data because of the usually poor association between biomass measured at inventory plots and remote sensing observables²¹. In addition, models relating biomass and remote sensing observables that are trained with spatially inhomogeneous datasets (Figure S1) tend to be biased in regions not represented by the dataset of the reference biomass measurements.

We estimate the total GSV of Russia for the year 2014 for the official forested area (713.1 × 10⁶ ha) to be 111 ± 1.3 × 10⁹ m³, which is 39% higher than the 79.9 × 10⁹ m³ (excluding shrubland) figure reported in the SFR³ for the same year. An additional 7.1 × 10⁹ m³ or 9% were found due to the larger forested area (+ 45.7 10⁶ ha) recognized by RS⁹, following the expansion of forests to the north²², to higher elevations, in abandoned arable land²³, as well as the inclusion of parks, gardens and other trees outside of forest, which were not counted as forest in the SFR. Based on cross-validation, our estimate at the regional level (81 regions of Russia – Table S2, Figure S2) is unbiased. The standard error varied from 0.6 to 17.6% depending on the region. The median error was 1.6%, while the area weighted error was 1.2%. The predicted GSV (Fig. 1) with associated uncertainties is available here (https://doi.org/10.5281/zenodo.3981198) as a GeoTiff at a spatial resolution of 3.2 arc sec. (ca 0.5 ha).

Houghton et al.²⁴ estimated forest biomass based on RS and FIP data in Russia for the year 2000. Average forest biomass density varied between 80.6 and 88.2 Mg ha^-1 depending on which forest mask was used. Our estimate for the year 2014 of 107 Mg ha^-1 (using the conversion factor of GSV to AGB from²⁴ 0.6859) is 21–33% higher than the one by Houghton et al., but this is consistent with expected biomass increases over time, i.e., 14 years after the Houghton et al. estimate.

Assuming an unchanged total forest area (721.7 × 10⁶ ha) in 1988 and 2014, we conclude that Russian forests have accumulated 1,163 × 10⁶ m³ yr^-1 or 407 Tg C yr^-1 in live biomass of trees on average over 26 years. This gives an average GSV change rate of + 1.61 m³ ha^-1 yr^-1 or + 0.56 t C ha^-1 yr^-1. The sequestration rate obtained, however, should be treated with caution because different methods have been applied in 1988 and 2014 (see “Caveats and Limitations” section). To provide some context for the magnitude of these numbers, one can compare the Russian forest gain to the net GSV losses in tropical forests over the period 1990–2015 according to FAO FRA²⁵ (-1,033 × 10⁶ m³ yr^-1 in the regions with a negative trend: South and Central America, South and Southeast Asia, and Africa). A similar divergence in the carbon sink between Tropical and Boreal forest was recognized by Tagesson et al.²⁶.

In terms of carbon stock change, our estimates are substantially higher than those reported by Pan et al.⁷ for 1990–2007 (+ 153 Tg C yr^-1) based on FIP data. The biomass carbon estimates by Liu et al.⁶ are instead in line with our results. There is an increase in the annual rate of AGB in Russia of + 329 Tg C yr⁻¹ (annual variation from 214 to 400 Tg C yr⁻¹) over 2000–2007⁶. Interestingly, another boreal country – Canada – has demonstrated neutral or negative trends (from 0 to -14 Tg C yr⁻¹) for the same time span using the same estimation method⁶.

We can observe different spatial patterns in the change in the GSV density between 1988 (FIP¹⁰, bias corrected¹¹) and 2014 (our estimate), which can be explained by climate change, CO₂ fertilisation and changes in disturbance regimes (Fig. 2). The average linear trend in the annual temperature increase during 1976–2014 in Russia is + 0.45 °C per 10 years²⁷. The temperature increase is statistically significant in every region except for western Siberia (Fig. 2–3). Significantly increased temperature extremes and an increase in the number of days without precipitation is observed in the south of European Russia, Baikal, Kamchatka, and Chukotka²⁷ (Fig. 2–1). Some regions in the south of the European part of Russia are colored in dark blue, but they, as a rule, have a small share of forested area, which is often linked to water bodies and, therefore, suffers less from increased drought (Fig. 2–1). Central and eastern Siberia suffer from an increase in disturbances, which offsets the climate stimulation effect (Fig. 2–4). The forested area in the Nenets region (Fig. 2–2) is 4 times larger in 2014 based on the RS forest mask compared to the SFR in 1988 (where forest was accounted for up until a certain latitude at that time), where the increase in area resulted in a decrease in the average GSV.

Focusing specifically on national reporting of managed forest to the UNFCCC, 72% of forested area in Russia is considered to be managed¹. We multiplied the GSV density by the managed forest area for each administrative region (Table S3). The difference in GSV estimation (between ours and the one from the SFR report) is 23.6 × 10⁹ m³ (Table S3) or 33% higher. From the GSV of managed forests in 2014 and based on the same area in 1988, we can estimate the sequestration rate of live biomass of managed forests as 354 Tg C yr^-1 , which is considerably higher than the figure of 230 Tg C yr^-1 in the current report¹.

This proof of concept demonstrates the relevance of complementing recent NFI data with remote sensing map products. Our study demonstrates that the already considerable value of forest inventory data can be further enhanced in a forest resources mapping scenario. In addition, we seek to promote greater access to these data by opening up their access to the larger scientific community. Through the integration of RS estimates of GSV and forest inventory data from Russia, we confirm that carbon stocks increased substantially during the last few decades in contrast to the figures provided in official national reporting. Russian forests play an even more important global role in carbon sequestration than previously thought, where the increase in growing stock is of the same magnitude as the net losses in tropical forests over the same time period.

Source: Ecology - nature.com