Handling the changes of organic carbon stock in mineral soils
of the Europen Union
Vladimir Stolbovoy1, Nicola Filippi1, Luca Montanarella1, Fabio Petrella2,
Mauro Piazzi2, Senthil-Kumar Selvaradjou1
1 Land Management and Natural Hazards Unit, Joint Research Center EC
2 Soil Department, Institute for Forestry and Environment (IPLA)


The interest in soil organic carbon (SOC) among soil scientists and practitioners is quite common due to its extensive influence on various environmental issues pertaining to global studies. But, this topic has recently got recognition as one among the most priority issues, because of its implications on the global and pan-European policies, e.g., SOC plays a key role in major Rio conventions, e.g., climate change, biodiversity and Combat Desertification. At the pan-European level SOC takes a central stage within European Soil Thematic Strategy issues (ESTS) (see EC COMMUNICATION (2002, 179)). The SOC is a parameter driving majority of soil ecological functions, e.g, fertility, buffering capacity, absorption of dangerous chemicals, water quality, regulation of atmospheric gas composition, etc. The decline of the SOC deteriorates soil quality and is recognized to be a serious environment threat indicated by European Environment Agency (Huber 2001, 2001). In principle, the amount of SOC is an universal indicator of the soil quality. Hence, it is an urgent need to develop a common, simple, transparent and cost effective method to identify the changes of organic carbon content in the mineral soils of the European Union. The objective of this paper is to introduce a feasible method for detecting the changes of organic C in the mineral soils of EU.


A new method named “Area Frame Randomised Soil Sampling” that focus on  practical sampling in field plots and forests was developed. This method combines traditional soil survey composite sampling with randomized positioning of the sampling sites in the field. The method aims for achieving: average SOC content; unification of the sampling localization and the project’s consistency; pedological substantialization; technological simplicity; and cost effectiveness. Technically, the method exploits ISO/FDIS 10381-1:2002(E)) and particularly relevant to ISO 10381-4 devoted to “Sampling to support legal or regulatory action” (standard setting). It supplements IPCC Good Practice Guidance (IPCC, 2003).

Technical specification

A randomized sampling template is a used as a core for this method. The template represents a grid of 100 points resulted from ‘modified random sampling’ with a distance threshold. Sampling points were selected at random with a careful attention that no points are less than 6 ‘distance units’ (the grid step). Where ever it was not possible to find points more distant than 6 units, the distance threshold was progressively relaxed. This sampling approach prevents the previous sampled point to get too close to the subsequent ones, failing which would lead to partially redundancy. This occurs for example both for systematic sampling and other sampling plans (Bellhouse, 1977, 1988). Systematic sampling, or other sampling plans that avoid points too close to each other, gives a lower variance than simple random sampling (SRS). But the application of the formulae given in the section ‘uncertainty’ to such sampling plans generally overestimates the variance (Wolter, 1984).


Template parameterization

Spatial dimensions of the template are different depending on the geographical coordinates (size) of the field. To define the template parameters one have to select the extrime axis X and Y values.  In our example (Figure 1) the X axis dimension is defined as a difference between Xmax = 376255 and Xmin = 375917, which is equal to 338 m. The Y dimension would be a difference between Ymax = 5025669 and Ymin = 5025326 that is 343 m. Thus the max axis dimension is 343 m, which defines the size of the template. The grid parameters are calculated by dividing max axis by 10, which gives 34.3 m. The grid size defines the distances between soil profiles and sampling points that is shown in self explaining fashion in the Figure.

It is important to note that the location of the soil profiles and sampling sites should be fixed in European Coordinate Reference Systems (CRS identifier ERTS89 Ellipsoidal CRS) (Boucher, C., Altamini, Z., 1992). The position should be recorded with the precision of 10m in the field by means of Global Positioning Systems (GPS) to be used for return visits to the sampling site. Data can be downloaded to a portable or office computer for registration and combination with other layers of information for spatial analysis.mpling procedure

As can be seen from figure 1, each sampling site represents a soil profile surrounded by the points at which samples should be taken for the composite sample. The distances for the sampling points are easy to calculate (Figure 1).


Figure 1. Parameterization of the randomized sampling according to the field plot (red).
The crosses indicate selection of the sampling sites.

As can be seen from the Figure, the number of points for the composite soil sampling is 25, which is in line with the ISO recommendation. Table 1 provides a number of the sampling sites depending on the area of the field.


Area of the field

Number of sampling sites

< 5 ha


5 - 10 ha


10-25 ha


> 25 ha


Table 1. Number of sampling sites depending on the field area


Pedological substantiation of the sampling

Because of the global need for the detection of the SOC changes in soils and the necessity for the result compatibility the widely recognized manuals for soil sampling tend to over simplify soil diversity. For example, the IPCC (IPCC, 2003) suggests one standard soil depth (0-30cm) for all soil types in different use, e.g., cropland, pasture, forest. This brings substantial heterogeneity in the soil parameters resulting in the need of tremendous number of soil samples to achieve a certain confidence of the results. To avoid this inconvenience the proposed method recommends specifying soils according to the use more precisely. This approach allows reducing the sample amount and the cost of the analysis.



The cropland-based soil profile can be schematised by two principal horizons: topsoil (the plough layer) and the subsoil underlying it (Figure 2a).[1]

The plough horizon or layer indicates regular anthropogenic perturbation and physical mixing of soil material throughout, e.g. organic and mineral fertilizers, application of earth, etc. The plough horizon hosts the largest proportion of root biomass and incorporates surface crop residues that contribute to the change in organic content in soils. The plough horizon is seldom stratified due to regular tillage. The thickness of the plough horizon is different depending on conventional cultivation in the countries. Therefore, it is proposed that one sample be taken from the middle of the plough horizon, e.g., at 10-20 cm depth if plough horizon is 30 cm thick as illustrated in Figure 2a. An undisturbed soil sample to determine the bulk density should also be taken at the same depth.


Pasture soils are exposed to anthropogenic disturbances limited to a reduction in organic input because of biomass consumption through grazing, fertilization, additional grass seeding, etc. The profile of these soils has gradual change of soil characteristics with depth in line with that of natural soils. The good practice guidance (IPCC, 2003) suggests detecting changes of organic carbon stock in 30 cm topsoil. The principal structure and a scheme of soil sampling of pasture are illustrated in Figure 2b.

Column sampling the profile at 10 cm intervals is recommended. These samples will be combined into one composite sample for the laboratory analysis. ‘Disturbed’ samples, taken at the three similar sampling depths, should be combined too into a composite sample. Sampling to determine bulk density and for laboratory analysis is the same as for cropland soils.


General rules for soil sampling in forests of Europe are specified by the ICP Manual (UNECE, 2003) and can be partly adapted, e.g., sampling points should be 1 m distant from tree stems and should avoid animal holes, disturbances like wind-thrown trees and trails. However, ICP Manual centres on monitoring of changes in the point and includes details of litter horizon, which are unnecessary when total organic carbon stock is considered.

As illustrated by the principal structure of soil sampling in forests (Figure 3c) the organic (litter) topsoil is sampled in a whole and accompanied by indication of total thickness of the layer. A frame of 25 by 25 cm is recommended for collecting forest organic layer. In the field, the total fresh weight of forest organic layer should be determined. A sub-sample is collected for the determination of moisture content (weight %) in the laboratory to calculate total dry weight (kg/m2).

Mineral layers should be sampled at exactly the same locations, i.e. sample the mineral soil underneath the organic layer that has already been removed for sampling. Sampling should be done at fixed depths. The top of the mineral soil corresponds with the zero level for depth measurements. The entire thickness of the predetermined depth should be sampled and not the central part of the layer only. Auguring is preferred and pits are allowed, especially in case of stony soils where auguring are usually difficult and sometimes impossible.

For the determination of bulk density each mineral layer (0-10 and 10-20 cm) of non-stony soils should be taken from.



Figure 2. The scheme of soil profile sampling



The changes in organic carbon stock in soils should be measurable, transparent and verifiable, which is in line with some recommendations (IPCC, 2003). This conditions can be achieved if based on physically measured carbon stocks prior to (baseline occasion) and after (second occasion), e.g. the latter can be first or second commitment periods for the Kyoto Ptotocol (UNFCCC, 1998), etc. Changes derived from models are complimentary and valuable to define potential for carbon sequestration. Area frame randomized soil sampling ensures a reproducibility of the sampling sites. The target is the estimate of the changes in organic carbon stock and its standard error rather than the estimate of organic carbon stock in soils.


The computation stems from a few parameters that have to be measured in the field, determined in laboratory and taken from other sources, e.g., cadastral information on the field location and area. The list of necessarily parameters includes: the carbon content in soil, the soil bulk density, the thickness of the soil layer, the content of coarse fragments and the area of the field. The computation routine follows steps below:

Step 1
: Calculation of soil organic carbon density SCDsite[2] for the sampling site:



SOC content  is a soil organic carbon content, % of mass or ,
Bulk Density  is a soil bulk density  ,

Depth  is a thickness of the sampled layer (cm) ,

frag  is  volume of coarse fragments, % of mass or .

The SCDsite provides an average value for the sampling site, which is derived from taking a composite sample combining a number of sub-samples. According to ISO 10381-4 at least 25 sub-samples should be obtained (see Figure 2).

Step 2
: Calculation of mean (arithmetic average) soil carbon density () for field:



SCDsite is as indicated in Equation 1,

n is a number of sampled sites within the plot.

Step 3
: Calculation of reference soil organic carbon stock (SOCrefstock) for the field:



 as indicated in Equation 2,

Ap is an area of the field.

Step 4
: Calculation of changes (ΔSOCstock) in organic carbon stock in soils[3]:



SOCrefstock  is as indicated in Equation 3,

SOCnew is a new soil organic carbon stock (second occasion), which is computed similar to SOCreference,

forg  is C applied with organic fertilizers,

flim is C applied with lime.



Uncertainty is a parameter associated with the result of measurement that characterizes the dispersion of the values that could be reasonably attributed to the measured quantity (IPCC, 2003) defines. The uncertainty of the changes in organic carbon stock in soils can be characterized by standard error of the changes value that can be computed by the steps below:

Step 5: Calculation of standard error for mean soil carbon density :





is the average of  for the sites sampled in the field ,

n is a number of sampling sites within the field.

Step 6: Calculation of standard error of the changes of organic carbon stock
in the field:



 is as indicated in Equation 5,

Ap is an area of the field.

Step 7: The overall result in weight of SOC and its standard error is:

Expressing the result inaccuracy in terms of standard error allows to avoid the normality assumptions.


Table 2 shows average SCD and SOCstock for three tested plots in Piemonte. The test was done twice on each plot to simulate second occasion sampling and followed the sampling protocol (Stolbovoy et al., 2005). The highest SCD (12.91 KgC/m2) was observed for pasture, which was explained by the  increase of C content in soil with height. Forest soils have the lowest SCD (4.12 KgC/m2), which might be due to the forest planting on poor dehumified soils. In this case, one can assume that forest planting will stimulate recovering of carbon content in soils. Cropland soils have some medium SCD (9.03 KgC/m2).  This data can suggest that establishing of grassland on cropland might lead to increase of carbon accumulation in soil of Piemonte. Actually, this conclusion is in line with general analysis for carbon sequestration options in soil of Europe.  

Second sampling illustrates distribution of SCD similar to that of the first sampling.  The difference for both samplings varies from 14.5 tC/ha for pastures to 5.2 tC/ha for forests.

Land Use

First sampling

Second sampling


Average SCD,


Average SCD,



% of the first sampling






















Table 2. Parameters of carbon content in soils of the tested cropland in Piemonte.
(Repeated area frame randomized soil sampling)


The cost efficiency is a critical issue for the carbon detection in soils. Figure 3 provides tentative cost analysis of the carbon determination in the laboratory depending on the area of the plot. The calculation is done for two common European conditions: 1) average carbon sink in agricultural soils is 1.5 tC/ha; 2) - the lab cost of carbon determination is 16 euro. As can be seen, the lab cost decreases with the increase of the field size: e.g., the lab cost to detect 1 tC in the field having 1 ha is nearly 50 euro; this cost is less than 2 euro if the area of the plot is about 50 ha.


Figure 3. Dependence of the lab cost for carbon determination on area of the field
(area frame randomized soil sampling).



A new area frame randomized soil sampling to detect the changes of organic carbon stock in mineral soils is developed. The field-testing shows the following advantages of the method: instrumental (GPS) positioning of the sampling sites, uniform design of the sampling strategy ensuring project’s consistency, pedological substantialization of the sampling scheme, technological simplicity and cost effectiveness. First field-testing of the Area Frame Randomised Soil Sampling illustrates that the changes ranged from 11 to 14 % of initial C stock can be verified for the tested plots. This difference illustrates that the method allows achieving the reliable reproducibility of the second occasion measurements. In addition, the method allows to check applicability of specific carbon sequestration measures for the field, e.g., cropland conversion to grassland can be verified for the 20yr period (approximate sink would be 40 tC/ha).




Bellhouse D.R., 1977, Some optimal designs for sampling in two dimensions, Biometrika 64, pp. 605-611.

Bellhouse D.R., 1988, Systematic sampling,  Handbook of Statistics, vol. 6, ed. P.R. Krisnaiah, C.R. Rao, North-Holland, Amsterdam,  pp. 125-146.

Boucher, C., Altamimi, Z., 1992, The EUREF Terrestrial Reference System and its First Realization. Veröffentlichungen der Bayerischen Kommission für die Internationale Erdmessung, Heft 52, München 1992, pp. 205-213.

EC (2002). Communication of 16 April 2002 from the Commission to the Council, the European Parliament, the Economic and Social Committee and the Committee of the Regions: Towards a Thematic Strategy for Soil Protection [COM (2002) 179 final]. (At: http://europa.eu.int/scadplus/printversion/en//lvb/l28122.htm; last accessed: 11.11.2004)

Huber, S., B. Syed, A. Fredenschuß, V. Ernstsen and P. Loveland, 2001. Proposal for European soil monitoring and assessment framework. European Environment Agency, Technical report 61, Copenhagen, 58 pp.

Intergovernmental Panel on Climate Change (IPCC), 2003. Penman J., M. Gytarsky, T. Hiraishi, T. Krug, D. Kruger, R. Pipatti, L. Buendia, K. Miwa, T. Ngara, K. Tanabe and F. Wagner (Eds). Good Practice Guidance for Land Use, Land Use Change and Forestry. IPCC/OECD/IEA/IGES, Hayama, Japan.

UNECE, 2003. ICP Forests Manual on methods and criteria for harmonized sampling, assessment, monitoring and analysis of the effects of air pollution on forests, 2003. Part IIIa Sampling and Analysis of Soil and Part IIIb Soil Solution Collection and Analysis. United Nations Commission for Europe Convention on Long-Range Transboundary Air Pollution, International Co-operative Programme on Assessment and Monitoring of Air Pollution Effects on Forests.

Soil quality – Sampling – Part 1: Guidance on the design of sampling programmes, 2002. ISO/FDIS 10381-1:2002(E).

Soil quality – Sampling – Part 4: Guidance on the pprocedure for investigation of natural, near-natural and cultivated sites, 2002. ISO/FDIS 10381-4:2002(E).

Stolbovoy, V., L. Montanarella, N. Filippi, S. Selvaradjou, P. Panagos and J. Gallego, 2005. Soil Sampling Protocol to Certify the Changes of Organic Carbon Stock in Mineral Soils of European Union. EUR 21576 EN, 12 pp. Office for Official Publications of the European Communities, Luxembourg.

UNFCCC, 1998. Report of the Conference of the Parties on its Third Session, Held at Kyoto from 1 to 11 December 1997. Addendum. Document FCCC/CP/1997/7/Add1. Available on the Internet: http://www.unfccc.de

Wolter K.M., 1984. An investigation of some estimators of variance for systematic sampling. Journal of the American Statistical Association, Vol. 79 No 388, pp. 781-790.



[1] If no-till and non-plough crops are adopted the soil profile turns to have gradual changes of soil characteristics with depth. For this case soil sampling should follows that of pasture. 

[2] SCD refers to carbon concentration  or  related to a layer of soil, e.g., 0-0.3, 0-05, 0-

1.0, 0-2.0 m. The SCD should not be confused with carbon content in soils, which is fraction of carbon by

weight of soil expressed in per cent or .

[3] This equation describes the changes of organic carbon stock due to sequestration from the atmosphere.  




    n. 1-3 anno 2005