In the context of the Kyoto Protocol, the mandatory accounting of Afforestation and Reforestation (AR) activities requires estimating the forest carbon (C) stock changes for any direct human-induced expansion of forest since 1990. We used the Carbon Budget Model (CBM) to estimate C stock changes and emissions from fires on AR lands at country level. Italy was chosen because it has one of the highest annual rates of AR in Europe and the same model was recently applied to Italy’s forest management area. We considered the time period 1990-2020 with two case studies reflecting different average annual rates of AR: 78 kha yr-1, based on the 2013 Italian National Inventory Report (NIR, official estimates), and 28 kha yr-1, based on the Italian Land Use Inventory System (IUTI estimates). We compared these two different AR rates with eight regional forest inventories and three independent local studies. The average annual C stock change estimated by CBM, excluding harvest or natural disturbances, was equal to 1738 Gg C yr-1 (official estimates) and 630 Gg C yr-1 (IUTI estimates). Results for the official estimates are consistent with the estimates reported by Italy to the KP for the period 2008-2010; for 2011 our estimates are about 20% higher than the country’s data, probably due to different assumptions on the fire disturbances, the AR rate and the dead wood and litter pools. Furthermore, our analysis suggests that: (i) the impact on the AR sink of different assumptions of species composition is small; (ii) the amount of harvest provided by AR has been negligible for the past (< 3%) and is expected to be small in the near future (up to 8% in 2020); (iii) forest fires up to 2011 had a small impact on the AR sink (on average, < 100 Gg C yr-1). Finally the comparison of the historical AR rates reported by NIR and IUTI with other independent sources gives mixed results: the regional inventories support the AR rates reported by the NIR, while some local studies suggest AR rates somehow intermediate between NIR and IUTI. In conclusion, this study suggests that the CBM can be applied at country level to estimate the C stock changes resulting from AR, including the effect of harvest and fires, though only a comparison with results based on direct field measurements could verify the model’s capability to estimate the real C stock change.
In the context of the Land Use, Land-Use change and Forestry (LULUCF) sector of the Kyoto protocol (KP), Afforestation and Reforestation (AR) refers to the direct human-induced conversion of non-forested land (not containing forest on 31 December 1989) to forested land. Specifically, afforestation refers to land that has not been forested for a period of at least 50 years and reforestation for less than 50 years (
For the first commitment period of the KP (2008-2012) the accounting of emissions and removals from AR is mandatory for Annex I (so-called developed) countries,
Many studies have provided estimates on the C stock changes for Forest Management (FM,
The contribution of AR removals to the GHG emission reduction targets may, however, be significant and - according to the country reports to the KP - it can be even more important than FM (
The main objective of this study was to estimate the stock changes of the five forest C pools for the period 1990-2020, for AR lands and at country level. To this purpose we applied the Carbon Budget Model (CBM) developed by the Canadian Forest Service (
Following the relevant IPCC guidance (
According to the Italian National Inventory Report (NIR) submitted to UNFCCC and its Kyoto protocol (
Until 2013, poplar plantations (considered cropland according to Italian laws) were excluded from AR (NIR 2013); the last Italian NIR (2014), submitted to UNFCCC on 15th April 2014, included poplar plantations under AR activities. All these lands are considered legally bound by national legislation, even prohibiting clear cut activities on these forests (
Due to the lack of data, the annual rate of deforestation occurring on AR cannot be estimated. According to the 2013 Italian NIR (Tab. 7.3 -
The estimates of the forest expansion can be provided by two different data sources: the Italian NIR and the Italian Land Use Inventory System (
As first data source, we used the official estimates reported by Italian NIR in 2013 (
The value reported in the NIR derives from a linear interpolation between the 2005 total forest area estimated by the last National Forest Inventory (NFI: Italian National Forest and Carbon Inventory, named INFC -
The INFC forest area was equal to about 8759 kha (with a standard error SE = 0.4%) and was based on the FAO-FRA 2000 forest definition (
By applying a forest definition based on a minimum forest size equal to 0.2 ha and a crown cover greater than 20%, the first Italian NFI estimated a total forest area for 1985 equal to 8675 kha (SE = 0.9%), including forests, plantations and other wooded lands (
According to
Based on the Land Use Change Matrices for the years 1990-2011 (Tab. 7.3 -
The annual rate of AR based on the official estimates (77.8 kha yr-1 and 78.6 kha yr-1 -
where 78.6 and 79.3 are the total annual rate of AR in kha yr-1 reported by Tab. 10.10 for the periods 2008-2009 and 2009-2010 (78.6 kha yr-1) and 2010-2011 (79.3 kha yr-1).
A second data source, named “IUTI estimates”, was obtained from the Italian Land Use Inventory System (
According to
To estimate the C stock change on AR lands, we used the Carbon Budget Model (CBM) developed by the Canadian Forest Service (
The CBM is an inventory-based (
The area, further distinguished by age classes, main species and management types (
The gross merchantable volume production by age classes, main species and (eventually) management types, defined by yield tables (YTs) provided by the user. These data represent the stand-level volume accumulation in the absence of natural disturbances and management practices. Tables can be directly inferred by the volume and increment data provided from NFIs or from the literature.
The present study focuses on young (less than 30 years old) forests. However, the INFC reports increment and volume data for the age classes < 20 years for only 39 out of 253 groups of forest types and regions. We therefore derived the YTs library from a large species-independent database (
The CBM spatial framework conceptually follows Reporting Method 1 (
In the present study, we considered 21 administrative units, 24 climatic units (CLUs), as defined by
Specific equations were selected to fit the species-specific values of biomass reported in the INFC and to convert the merchantable volume into aboveground biomass (
To estimate the decomposition rate of each DOM pool, the base decomposition rates defined at 10 °C for each pool is adjusted in the CBM based on the mean annual temperature in each SPU. For forested lands, DOM pools (dead wood and litter) and soil are initialized using a procedure that takes into consideration site productivity, temperature-dependent decomposition rates and disturbance history (
For the initialization of the non-forested lands, the user can define the initial C content (before afforestation) of the living biomass (
Forest expansion in Italy is mainly related to process of natural re-colonization of abandoned areas (
the initial soil C stock was equal to the average soil C content of grassland,
the average C stock of the living biomass sub-pools before AR was equal to 3% (
The user can define annual natural and anthropogenic disturbances such as fire, insects or storms and partial or clear-cut harvesting which may be applied during the model run (
The model provides annual predictions on C stocks and fluxes, such as the annual C transfers between pools, from pools to the atmosphere and to the forest product sector, as well as ecological indicators such as the net primary production. The model also reports land areas and C pools in the appropriate UNFCCC and KP land categories.
The annual rate of afforestation reported in
The total forest area detected by INFC can be distinguished between different management types (defined according to the silvicultural system and the forest structure): coppices (further distinguished between simple coppice, coppice with standards, coppices in transition to high forests, etc.), even-aged high forests, uneven-aged high forests, not-defined class and not-classified for the management type/system forest area (see Table 7.4 in
We assumed that the new forests detected during the field measurements of the INFC should have been included in the not-defined class (
The original FTs percentage distribution estimated at the regional level and based on the not-defined plus not-classified forest area was applied to the annual rates of AR defined for the official estimates, as a first possible share of species. This distribution, named “FT distribution 1”, was used to test the effect of the species composition on the model output.
Since the previous forest area (
To provide an estimate of the future potential C sink related to AR, the model was run to 2020, assuming a constant annual rate of afforestation between 2011 and 2020 (see
(1) We estimated the potential (
(2) Since fire is the main natural disturbance affecting Italian forests (
where
A summary of the different scenarios analyzed by CBM is reported in
The C stock change estimated by CBM is reported in
As expected, using IUTI estimates (excluding harvest and fire) with a total amount of afforestation in 2011 of 579 kha, we estimated a lower average C stock change of 370, 63 and 433 Gg C yr-1 for the living biomass, DOM and soil and total pools, respectively. This last figure is about 60% lower than the value from the official estimates.
Assuming a constant annual rate of AR after 2012 and excluding disturbance events, CBM estimated a total C stock change in 2020 equal to 3839 and 1298 Gg C yr-1 in official estimates (AR = 78.8 kha yr-1) and IUTI estimates (AR = 22.9 kha yr-1), respectively. Applying the FT distribution 1 to official estimates, we still detected a negligible percentage difference (1.5%) with the FT distribution 2 for the living biomass stock in 2020.
Adding the effects of the potential harvest provided by AR land to these runs, the 2020 total C stock change decreased to 3237 and 1065 Gg C yr-1, using the official and IUTI estimates, respectively. The total amount of harvest provided using the official estimates increased from about 325 400 m3 in 2005 (totally provided by broadleaves) to about 1 200 000 m3 in 2020. Due to our assumptions, about 85% of this amount was provided by broadleaves in 2020. This highlights that the silvicultural treatments applied to the new forest lands, excluding any kind of clearcuts, may decrease the potential 2020 C stock change by about 15-17% compared to the runs with no disturbance.
This effect was null before 2004 (
A different FT composition (based on the FT distribution 1 applied to official estimates) may reduce the total amount of harvest by less than 4% on average.
The total amount of harvest provided by AR using the official estimates increases from about 89 Gg of dry matter in 2005 (42.5 Gg using the IUTI estimates) to about 328 Gg of dry matter in 2020 (127 Gg using the IUTI estimates). These figures account for about 3% (1% in IUTI estimates) of the total amount of harvest provided in 2005 by Italian forests, equal to about 6550 Gg of dry matter (
The possible amount of harvest provided by plantations is also small. Indeed, the total amount of plantations (excluding poplar) reported at the national level by INFC is equal to about 56 kha. The inventory also detected the total amount of plantations (including poplars) established after 1990 equal to about 60% of the amount of plantations (accounting also for the not classified area). This suggests that, even assuming that these new forests are totally represented by not-poplar plantations, the total amount of plantations established between 1990 and 2005 is equal to about 33 kha,
In order to directly compare our results with the values reported by NIR, we reported in
In
For the period 2008-2010, including the effect of fire and harvest (red line in
Other differences may be due to the effect of fire on the dead wood and litter pools. Indeed, as highlighted for the FM area, due to the approach applied by Italy (based on a linear regression with the aboveground biomass), a reduction in biomass C pool due to fire causes a corresponding reduction in the dead wood pool which represents an immediate release to the atmosphere. In the CBM model, fire disturbances move part of the living biomass to the dead wood pool where it will slowly be released to the atmosphere through decay (
The differences between the annual rates of AR based on the official estimates and IUTI reflect the difference in the trend of total forest area reported by the two NFIs (used by official estimates) and estimated by IUTI. For IUTI, based on the data reported by
where
By contrast, the forest area estimated by the NFIs, excluding shrub lands, other wooded lands and poplar plantations, is equal to 7089 kha (
Since both these studies also provide detailed information at the regional level, the total forest area reported for 1985 (by the first Italian NFI), 1990 (by IUTI), 2000 (by IUTI), 2005 (by INFC) and 2008 (by IUTI) can be compared with the estimates provided by regional forest inventories available for 8 of 21 administrative regions (
Despite the differences in the forest definitions applied at the regional level (reported in
The different AR rates could also be partially due to a possible underestimation of the 2005 forest area reported by the second NFI, due to the omission, in the second stage of the INFC, of field surveys on points classified as not-forested during the first stage (
Furthermore, the localization of the forest border, the possibility to distinguish young trees from shrubs in orthophotos, the question whether actual or potential minimum tree heights are assessed, the training of the personnel conducting the assessments, as well as other issues specified in the field protocols and assessment instructions, have to be considered in an exhaustive comparison of the two studies (
Additional information to analyze AR estimates of NIR and IUTI is provided by three local studies on forest expansion. For the Trentino region,
The observation period is different and particularly for the Trentino region the initial year (1973) could affect the resulting AR, as afforestation is not a linear phenomenon.
The case study of Veneto is limited to a portion of the region.
The three studies are based on photointerpretation without field data, and the diachronic classification of the ortophotos could have led to an overestimation of forest area for the past.
Overall, the comparison of AR rates reported by the 2013 NIR and IUTI with other independent sources for the period 1990-2005 gives mixed results: the regional forest inventories seem to support the AR rates reported by the NIR, while the local studies suggest AR rates somehow intermediate between NIR and IUTI. Based on the data provided by IUTI, recently confirmed by the preliminary results provided by the new Italian NFI, it is likely that the annual rate of AR in Italy in the last decade is decreasing compared with the ’ÂÂ80s and the ’ÂÂ90s.
We used the CBM to estimate the C stock changes resulting from AR activities in Italy for the period 1990-2020, including the potential effect of harvest and natural disturbances. We ran the model for two cases studies, based on different sources of data: an average annual rate of AR of about 78 kha yr-1 (using the 2013 official estimates, reflecting the Italian National Inventory Report) and 28 kha yr-1 (using the IUTI estimates, based on Italian Land Use Inventory System). Furthermore, we compared these two different AR rates with independent sources: eight regional forest inventories and three local studies.
The average C stock change estimated by our model between 1990 and 2020, excluding harvest or natural disturbances, averaged 1738 Gg C yr-1 (official estimates) and 630 Gg C yr-1 (IUTI estimates). The results based on the official estimates are consistent with the estimates reported by Italy for the period 2008-2010. Due to a different amount of area burned and AR rate, as well as to different model assumptions on the dead wood and litter pools, the C sink estimated by CBM for 2011, was about 20% lower than the C sink reported by the KP LULUCF tables (
Furthermore, our analysis suggests that:
the rates of AR are a major source of uncertainty in the estimation of AR stock changes; CBM results are comparable with NIR’s ones when the same AR rate is used;
the different assumptions about the forest composition of AR area have a small impact on estimates of the AR sink;
due to the young age of the new forests, the potential amount of harvest provided from AR land has been negligible for the historical period (likely less than 3% of the total harvest in 2005) and is expected to be small in the near future (up to about 8% in 2020);
forest fires for the historical period, distributed by year according to the relative proportion of the AR area compared to the total forest area, had a relatively small impact on the AR sink (on average, emission from forest fires were less than 5% of the sink);
the selection of the yield tables applied by the model was based on the volume data reported by the INFC for even-aged forest; the same selection could also be based on increment data reflecting the current growth of forest, but this requires more specific information on these stands;
the comparison of NIR and IUTI estimates for AR with eight regional forest inventories confirms the afforestation rate of NIR for the period 1990-2005, while the comparison with three local studies suggests AR rates somehow intermediate between NIR and IUTI; based on the data provided by IUTI, recently confirmed by the preliminary data provided by the new Italian NFI, it is likely that the annual rate of AR in Italy in the last decade is decreasing compared with the ’ÂÂ80s and the ’ÂÂ90s.
In conclusion, this study suggests that the CBM can be applied at the country level to estimate the C stock change related to AR, including the effect of harvest and natural disturbances, even if only a comparison with results based on direct field measurements could verify the model’s capability to estimate the real C stock change.
We thank Marina Vitullo (Italian Institute for Environmental Protection and Research) for the useful comments and suggestions for improving this paper and Raoul Abad-Vinas (Joint Research Center) for the support provided to analyses the NIR and KP LULUCF tables. We also thank two anonymous reviewers who provided useful suggestions to improve the manuscript.
The views expressed are purely those of the authors and may not in any circumstances be regarded as stating an official position of the European Commission or Natural Resources Canada.
Italian administrative regions and NUTS2 code (
Left panel: total forest area (excluding poplar plantations, shrub land and other wooded lands) estimated by the first (NFI 1985) and the second (INFC 2005) NFI (used by official estimates) and by IUTI (IUTI estimates) and a linear interpolation between these data. Right panel: annual rate of AR applied during the model run, according to data reported by the Italian NIR (
Schematic representation of the main input data required by CBM in order to define the Spatial Units (SpUs). The dashed line identifies the general classifiers and the dotted line delimits the information provided by the forest inventory, split between each SpU (by
Comparison between the FT distribution 1 (based on the FT distribution of the INFC not-defined plus not-classified forest area) and the FT distribution 2 (based on the previous distribution corrected according to the weighting factors reported in
(Upper panel): total amount of area burned (black line) reported by official statistics (
C stock change (Gg C yr-1) estimated by CBM for official estimates (based on the NIR’s assumptions) and IUTI estimates (based on the IUTI’s assumptions). The figure reports the (i) living biomass, (ii) litter, dead wood (DOM) and soil, and (iii) total C stock change, excluding disturbances (no harvest and no fire) and including the potential amount of harvest provided by AR (reported in the right panels in Gg of dry matter yr-1). The vertical dotted line in the left panels divides historical data (before 2011) and the future AR rate used in our study.
Comparison between the total C sink (in Gg CO2 yr-1) estimated by CBM with the official estimates (based on NIR assumptions,
Comparison between the forest area reported by 8 regional forest inventories (RFI), by the first (NFI 85) and the second (INFC) Italian NFIs and by IUTI (for 1990, 2000 and 2008). The dotted line highlights the linear interpolation between the two NFIs (approach used by the NIR) and the dashed lines highlight the linear interpolations between the 1990-2000 and the 2000-2008 IUTI estimates.
Percentage distribution at the regional level of the total annual rates of AR applied in official estimates (assuming a fixed percentage distribution for the entire period) and in IUTI estimates (assuming different distributions before and after 2000). (*): Due to rounding the sum may be slightly lower than 100%.
Italian Regions | Official estimates (NIR) | IUTI estimates (IUTI) | |
---|---|---|---|
1990-2020 | 1990-2000 | 2001-2020 | |
Abruzzo | 5 | 4 | 8 |
Basilicata | 3 | 3 | 5 |
Calabria | 5 | 0 | 2 |
Campania | 4 | 5 | 1 |
Emilia-Romagna | 7 | 2 | 4 |
Friuli-Venezia | 4 | 7 | 5 |
Lazio | 7 | 2 | 1 |
Liguria | 3 | 4 | 5 |
Lombardia | 6 | 5 | 0 |
Marche | 4 | 6 | 9 |
Molise | 1 | 9 | 2 |
Piemonte | 10 | 5 | 2 |
Puglia | 2 | 9 | 8 |
Sardegna | 6 | 4 | 0 |
Sicilia | 3 | 8 | 20 |
Toscana | 12 | 4 | 12 |
Bolzano-Bozen | 4 | 11 | 9 |
Trento | 4 | 1 | 1 |
Umbria | 4 | 5 | 4 |
Valle d’Aosta | 1 | 1 | 0 |
Veneto | 5 | 4 | 2 |
Italy* | 100 | 100 | 100 |
Main species associated to each forest type (FT), seed dispersal capacity assigned to each FT (1: anemochory/wind-dispersed species; 0: non-anemochory species; 0.5: mixed groups of species), light tolerance index based on
Main species | FT | Seed dispersal capacity | Light tolerance | Weighting factor |
---|---|---|---|---|
Larch and stone pine forests | LD | 1.0 | 1.0 | 2.0 |
Norway spruce forests | PA | 1.0 | 0.5 | 1.5 |
Silver Fir forests | AA | 1.0 | 0.0 | 1.0 |
Scots pine and Mountain pine | PS | 1.0 | 1.0 | 2.0 |
Black pine forests | PN | 1.0 | 1.0 | 2.0 |
Mediterranean pine forests | PM | 0.0 | 0.5 | 0.5 |
Other coniferous forests | OC | 0.5 | 0.5 | 1.0 |
Beech forests | FS | 0.0 | 0.5 | 0.5 |
Oak forests ( |
QR | 0.0 | 1.0 | 1.0 |
Turkey Oak forests | QC | 0.0 | 1.0 | 1.0 |
Chestnut forests | CS | 0.0 | 0.5 | 0.5 |
Hornbeam forests | Oca | 1.0 | 0.5 | 1.5 |
Riparian forests | RF | 1.0 | 1.0 | 2.0 |
Mixed deciduous broadleaved forests | OB | 0.5 | 0.5 | 1.0 |
Holm oak forests | QI | 0.0 | 0.5 | 0.5 |
Cork oak forests | QS | 0.0 | 0.5 | 0.5 |
Other evergreen forests | OE | 0.5 | 0.5 | 1.0 |
Summary of the scenarios analyzed by CBM, based on different assumptions about the annual rate of AR (case studies 1 and 2), FT distribution (Original and Corrected) and disturbance events (potential harvest and fire).
AR assumptions | Case study | FT share | Harvest | Fire |
---|---|---|---|---|
NIR | 1 | Corrected | No | No |
NIR | 1 | Original | No | No |
NIR | 1 | Corrected | Yes | No |
NIR | 1 | Original | Yes | No |
NIR | 1 | Corrected | Yes | Yes |
IUTI | 2 | Corrected | No | No |
Reference year and main parameters defining the forest definition (minimum area, forest cover, width and potential tree height) applied by the National forest inventories, by IUTI and by the regional inventories reported in
Inventory | Referenceyear | Min. area(m2) | Forestcover (%) | Min. width(m) | Min. tree height (m) | Reference |
---|---|---|---|---|---|---|
INFC | 2005 | 5000 | 10 | 20 | 5 | |
NFI | 1985 | 2000 | 20 | 20 | - | |
IUTI | 1990, 2000, 2008 | 5000 | 10 | - | 5 | |
Emilia Romagna | 1985 | 5000 | 10 | 20 | 5 |
|
Marche | 2000 | 2000 | 20 | 20 | - |
|
Sicilia | 2009 | 5000 | 10 | 20 | 5 |
|
Toscana | 1991-1993 | 5000 | 10 | 20 | 5 | |
Trentino | 2003 | 5000 | 10 | 20 | 5 |
|
Umbria | 1991 | 2000 | 20 | 20 | - |
|
Val d’Aosta | 1993 | 2000 | 20 | 20 | - |
|
Veneto | 1986 | 5000 | 10 | 20 | 5 |
|