I'm not sure I'd really call the article
an attack. It's also odd that they seek to rebut the article's 49% figure, which is an EU-wide average, rather than the figure for Sweden itself, which is 36%.
The paper uses remote sensing data - that is, they rely on satellite imaging data automatically classified into different land-use types by an algorithm. This is a common and growing area of applied research, both for checking official figures and for research in areas where fieldwork doesn't happen for whatever reason. The authors have attempted to validate this data in a few ways. One is by using high-resolution photos and classifying them themselves, and the other is by validating with an independent data source.
Sweden has its own national forestry dataset, with different methodology which the authors didn't use for validation, using UN FAO data instead (it's quite common to use internationally standardised data sources when conducting international analyses). But they haven't attacked it (or even mentioned it). According to the Swedish National Forest Inventory website
, they define a forest thus:
Forest land is a broad land use class defined in the Swedish Forestry Act. The definition the same as the UN's Food and Agriculture Organization (FAO) definition of Forest Land and is internationally accepted. Forest land is defined as land with trees hiher than 5 meter and a canopy cover of more than 10 percent, or trees able to reach these thresholds in situ.
The remote sensing data, on the other hand, uses area of tree cover which is "defined as the canopy closure for all vegetation taller than 5 m in height. The authors then convert tree cover to forest cover on a per-country basis:
Specifically, we computed for 15 tree-cover classes—from 10% to 80% in 5% steps—the corresponding forest areas and selected the class that minimizes the difference between the national forest-area statistics collected in the FAOSTAT report for the year 2015 (hereafter FAOSTAT-2015); using the last published dataset is a common approach. To match the FAO definition of forest, we used a minimum mapping unit (MMU) of approximately 0.5 ha with a moving-window kernel…<snip>
The results of this analysis show that national forest-areas change considerably according to the choice of the minimum tree-cover threshold and that this threshold varies by country, making it inappropriate to use a single threshold for the whole of Europe.
It should be noted that the GFC definition of forest is land-cover based, whereas the national forest inventories employ a land-use definition. For example, orchards are considered as forests in the GFC, whereas they are excluded from national forest inventories. Conversely, bare ground which has been affected by harvest operations is still called forest if it is expected to revert to forest by national forest inventories (land-use approach). Thus, the GFC maps can be used to produce a map of forest cover, with some caveats48,49,50.
For Sweden, the difference between the FAO data and their estimate is <5%, but that doesn't mean that the two data sources are looking at the same areas. You could cut down 90% of the trees in a dense Swedish forest and the official figures would show no change, whereas the GFC data would probably notice the difference.
They also compare the remote-sensing estimates with official harvest figures from the UN FAO:
For each EU26 country, we compared the harvested forest area derived from the GFC maps and the amount of harvest volume removals reported by FAOSTAT. Harvest removals (that is, ‘total roundwood production’) are provided by FAOSTAT for each European country for the years 2004–2018, further corrected to account for possible inconsistencies, according to a previous analysis17. Harvest removals are expressed as volumes.
Note that these figures relate to the amount of wood harvested, not forest area, so discrepancies will arise where harvesting is selective (leaving much of the area forested) rather than by clear-cutting. The area figures also don't account for differences in densities of wood per unit area that might arise from differences in tree species, climate, planting, etc.
Here's what they have to say about this validation for Sweden in particular:
The lack of correlation between the GFC data and harvest-removal data is probably due to: (1) when large disturbance events occurred, salvage logging (for sanitary reasons) had the priority on clear cut, the area of which was indirectly reduced (for this reason, probably, the GFC does not highlight the effect of the two windstorms that occurred in 2005 and 2007); (2) remote-sensing estimates and harvest statistics at the country scale may not show a statistical correlation because the biomass density per unit of area differs greatly over the country in space (that is, between the northern or southern part of Sweden); and (3) for this country, final felling covered (in terms of area) about 37% of the area annually affected by fellings between 2000 and 201556. This area is not statistically correlated with the total amount of wood removed during the same period, as reported by the same data source (r = 0.48). Despite that, official statistics on the notified area (larger than 0.5 ha), affected by final felling are consistent with the GFC (see Supplementary Fig. 3) and highlight that the size of this area increased by 13% in 2018 in comparison with the previous year, and compared with the average of the period 2011–2015, increased by nearly 17%. Considering that these statistics only report the “notified area larger than 0.5 ha”56, whereas the GFC probably includes a broader share of management practices, we can infer that in Sweden the GFC maps adequately represent the variation in the relative amount of area affected by final felling.
It's not de facto surprising that wood harvesting might increase:
Among exogenous drivers, the expansion of activities on the basis of demand for wood products (economic drivers) might have affected the forest sector, as reported in official statistics from UNECE and FAO33 and Eurostat75. In fact, forest harvest is unlikely to increase when there is no rise in market demand for wood products. In northern and central–eastern Europe, where the relative contribution of the forest sector to GDP is the largest (2.1% and 1.3%, respectively, in 2010)1, the higher demand from sawmills during the last years was probably one of the major drivers of the increasing timber harvest33. For example, in Croatia sawn-hardwood production grew by 89% in the five years to 2017, and in the Czech Republic and Slovakia particleboard production grew by 10% and 6.5%, respectively, in 2017 compared with the previous year33. In addition, fuelwood removals increased at the EU26 level from around 70 Mm3 to about 99 Mm3 (+41%) between 2000 and 201576. UNECE33 also confirms a substantial increase of EU harvest in 2013–2017 compared to 2007, with three countries standing out: Poland (+19.5%), Finland (+12.2%) and Sweden (+7.5%).
International trade, sometimes linked to political factors, may also affect the harvest demand at the national level. This was, for example, the case in some north European countries (such as Finland and Estonia), where, since 2009, the collapse of exports of roundwood from Russia indirectly affected internal harvest demand. Conversely, in some central European countries (such as the Czech Republic, Hungary and Slovenia), exports have strongly increased since 2014, encouraged not only by increasing roundwood demand coming from Germany (where imports increased by 30% since 2014), but also by from other EU26 countries (such as the UK and Croatia), and more recently, from China.
The lack of standardised data in the EU is a known problem, as the associated editorial highlights: How Europe can fix its forest data gap
. The Commission is working on a new forestry strategy for next year, and planning to set up an EU-wide forestry observatory that will use remote-sensing data to compile official statistics that are standardised across countries, rather than the current piecemeal approach.
The editorial also cites some stats suggesting that the observed increase in deforestation in Sweden isn't necessarily surprising:
Paradoxically, the increase in harvested forest area has been driven, in part, by demand for greener fuels, some of which are produced from wood biomass. That includesbioenergy, which comprises about 60% of the EU’s renewable energy. This increase in biomass products can, in turn, be traced to the EU’s bioeconomy strategy, a policy that has promoted the use of forest resources for energy, as raw materials for industries and to create jobs.
The bioeconomy strategy has been a success in one respect: total economic output from the EU’s forests between 2012 and 2016 rose by 25%, from €43 billion to €54 billion— and the increase doubled to 50% in Poland and Sweden. But economic success has come at an ecological cost.