Scholarly article on topic 'Reference Area Method for Mapping Soil Organic Carbon Content at Regional Scale'

Reference Area Method for Mapping Soil Organic Carbon Content at Regional Scale Academic research paper on "Earth and related environmental sciences"

CC BY-NC-ND
0
0
Share paper
Keywords
digital soil mapping / soil organic carbon / geostatistics / mapping

Abstract of research paper on Earth and related environmental sciences, author of scientific article — Yusuf Yigini, Panos Panagos

Abstract Soil organic carbon, the major component of soil organic matter, is very important in all soil processes. The decline in soil organic carbon (SOC) is recognized as one of the eight soil threats identified in the European Union Thematic Strategy for Soil Protection (COM (2006)231 final) (EC, 2006). The research challenge in this paper is to develop aggregation techniques which can ‘catch’ the variability at local level and being able to transfer this knowledge to larger scales. The latest developments included the research in upscaling modelling techniques which allow transferring the data values from local to regional scale. Digital Soil Mapping (DSM) and in particular regression kriging has been selected as the most appropriate modelling approach for the study. It has been successfully implemented at regional level and a soil organic carbon map was produced for the selected region (Slavkovsky Forest, 2400 km2, Czech Republic) by transferring knowledge from Lysina Critical Zone Observatory (0.40 km2). The Lysina CZO is situated in western Bohemia, the Czech Republic. Using digital soil mapping (DSM), demonstrated that it is possible to upscale the processes that take place at Critical Zone Observatories (CZOs) towards larger regions. According to the prediction results, the soil organic carbon values in the Slavkovsky Forest are ranging between 0% - 35.11%. The results of the linear regression procedure is promising, however, the statistical indicators are relatively low (R2=0.31) in the first step of the two stage model. The results of the study encourage applying similar approach at a wider scale. However, the ground data availability is still the key component to have more robust geostatistical models. The model and its outputs can be improved by using more ground data and high resolution environmental covariates.

Academic research paper on topic "Reference Area Method for Mapping Soil Organic Carbon Content at Regional Scale"

Available online at www.sciencedirect.com

ScienceDirect

Procedía

CrossMark

ELSEVIER

Earth and Planetary Science

Procedía Earth and Planetary Science 10 (2014) 330 - 338

Geochemistry of the Earth's Surface meeting, GES-10

Reference Area Method for Mapping Soil Organic Carbon Content

at Regional Scale

Soil organic carbon, the major component of soil organic matter, is very important in all soil processes. The decline in soil organic carbon (SOC) is recognized as one of the eight soil threats identified in the European Union Thematic Strategy for Soil Protection (COM (2006)231 final) (EC, 2006). The research challenge in this paper is to develop aggregation techniques which can 'catch' the variability at local level and being able to transfer this knowledge to larger scales. The latest developments included the research in upscaling modelling techniques which allow transferring the data values from local to regional scale. Digital Soil Mapping (DSM) and in particular regression kriging has been selected as the most appropriate modelling approach for the study. It has been successfully implemented at regional level and a soil organic carbon map was produced for the selected region (Slavkovsky Forest, 2400 km2, Czech Republic) by transferring knowledge from Lysina Critical Zone Observatory (0.40 km2). The Lysina CZO is situated in western Bohemia, the Czech Republic. Using digital soil mapping (DSM), demonstrated that it is possible to upscale the processes that take place at Critical Zone Observatories (CZOs) towards larger regions. According to the prediction results, the soil organic carbon values in the Slavkovsky Forest are ranging between 0 % - 35.11%. The results of the linear regression procedure is promising, however, the statistical indicators are relatively low (R2=0.31) in the first step of the two stage model. The results of the study encourage applying similar approach at a wider scale. However, the ground data availability is still the key component to have more robust geostatistical models. The model and its outputs can be improved by using more ground data and high resolution environmental covariates.

© 2014TheAuthors. Publishedby ElsevierB.V.This is an open access article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/3.0/).

Peer-review under responsibility of the Scientific Committee of GES-10

Keywords: digital soil mapping, soil organic carbon, geostatistics, mapping

* Corresponding author. Tel.: +39 0332 786210; fax: +39 0332 786394. E-mail address: yusuf.yigini@jrc.ec.europa.eu

1878-5220 © 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/3.0/).

Peer-review under responsibility of the Scientific Committee of GES-10

doi: 10.1016/j.proeps.2014.08.028

Yusuf Yigini*, Panos Panagos

European Commission Joint Research Centre Land Resource Management Unit Italy

Abstract

1. Introduction

Soil organic carbon as one of the most important environmental indicators has been used in mapping and rescaling studies. Soil organic carbon is very important in all soil processes. Organic material in the soil is essentially derived from residual plant and animal material due to action of microbes and decomposed under the influence of temperature, moisture and ambient soil conditions. The annual rate of loss of organic matter can vary greatly, depending on cultivation practices, the type of plant/crop cover, drainage status of the soil and weather conditions. There are two groups of factors that influence inherent organic matter content: natural factors (climate, soil parent material, land cover and/or vegetation and topography), and human-induced factors (land use, management and degradation).

The recent EU-funded ENVASSO project (Kibblewhite et al., 2008) investigated the feasibility of deriving indicators relating to the key threats to soil. Soil Organic Carbon (SOC) content is one of the 27 priority indicators, with baseline and threshold values, that can be rigorously defined and implemented to form a Europe-wide reference base to assess current and future soil status.

Under the European Union Thematic Strategy for Soil Protection, the European Commission Directorate-General for the Environment and the European Environmental Agency (EEA) identified a decline in soil organic carbon and soil losses by erosion as priorities for the collection of policy relevant soil data at European scale. Moreover, the estimation of soil organic carbon content is of crucial importance for soil protection and for climate change mitigation strategies. One of the key goals of the strategy is to maintain and enhance SOC levels. A recent policy document known as the Roadmap to a Resource Efficient Europe (COM(2011) 571 final) (EC, 2011), which is one of the building blocks of Europe 2020 strategy, sets the objective of increasing current levels of soil organic carbon in areas by 2020 where less than 2% of SOC has been detected (Panagos et al, 2013).

At the European level, there is a serious lack of geo-referenced, measured and harmonised data on soil organic carbon available from systematic sampling programmes. The European Soil Database, at a scale of 1:1,000,000 (King et al., 1994), is the only comprehensive source of data on the soils of Europe harmonised according to a standard international classification (FAO). At the present time, the most homogeneous and comprehensive data on the organic carbon/matter content of European soils remain those that can be extracted and/or derived from the European Soil Database in combination with associated databases on land cover, climate and topography.

Soil resources play a major role as a terrestrial sink of carbon and can contribute to climate change mitigation and adaptation. However, around 45 % of the mineral soils in Europe have low or very low organic matter content (0 to 2 % organic carbon) and 45 % have a medium content (2 to 6 % organic carbon) and soil organic matter in Europe is currently diminishing. Several factors are responsible for the decline in soil organic matter and many of them relate to human activity. These factors include conversion of grassland, forests and natural vegetation to arable land; deep ploughing of arable soils; drainage, liming, nitrogen fertiliser use; tillage of peat soils; crop rotations with reduced proportion of grasses (Jones et al, 2012).

Some 45% of soils in Europe have low or very low organic matter content (0-2% of SOC). This is particularly evident in the soils of many southern European Countries, but is also evident in parts of France, the United Kingdom, Germany, Norway and Belgium. A key driver is the conversion of woodland and grassland to arable crops. The soils of EU27 Member States are estimated to store between 73 and 79 billion tonnes of carbon (Jones et al, 2012).

Numerous environmental and socio-economic models require soil parameters as inputs to estimate and forecast changes in our future life conditions. However, the availability of soil data is limited on both national and European scales. Soil information is either missing at the appropriate scale, its meaning is not well explained for reliable interpretation, or the quality of the data is questionable (Dobos et. al, 2006). DSM has evolved as a discipline linking field, laboratory, and proximal soil observations with quantitative methods to infer on spatial patterns of soils across

various spatial and temporal scales. Studies use various approaches to predict soil properties or classes including univariate and multi-variate statistical, geostatistical and hybrid methods, and process-based models that relate soils to environmental covariates considering spatial and temporal dimensions (Grunwald, 2010).

Decision-making in the field of agriculture and environment ideally requires a precise knowledge of soils over wide regions. However, detailed soil surveys are only possible over limited area because high costs of systematic soil mapping and soil property measurements (Lagacherie et al, 2001). When dealing with areas of different sizes and with information available at different scales, policy makers and decision makers need to either upscale their evaluations and simulations from small to large scale or downscale from large to small scale (Stein et al., 2001).

This study uses data from one of the SoilTrEC Project Critical Zone Observatories (CZO) which is heavily managed forest area and used for timber production (Banwart et al., 2012). The Lysina CZO is situated in western Bohemia, the Czech Republic. CZOs are natural areas defined by the natural boundaries, in which scientists from several different fields, measure and monitor natural processes. The chosen CZO are relatively small area and well researched by the SoilTrEC researchers. Although, CZOs are small well-studied areas, It is time consuming and costly to map soil properties and to understand natural processes in the larger neighbourhood areas.

The objective of this paper is to show to what extent the natural processes can be extended from reference areas to larger areas and how the local (CZO) information is transferable to larger areas (Regions, Catchments) using geostatistical methods.

2. Material and Method

2.1. Study Area

The Lysina CZO is an experimental catchment to study the resilience of soil functions to acid deposition and other impacts in heavily managed forest land used for timber production. It represents managed land as an important economic asset where soil is under threat from industrial pollution. The Lysina CZO is situated in western Bohemia, the Czech Republic and more specific in the Protected Area Slavkovsky Forest, 120 km W from Prague, 10 km N from Marienbad. The catchment covers an area of around 0.40 km2 and is covered with a Norway spruce (Picea abies) monoculture (Navratil et al., 2007). Altitude of the site is ranges between 829 and 949 m above sea level with a mean slope gradient of 11.5%. The region was not glaciated during Wisconsinan time, and soils are developed from residuum bedrock. Soils in the catchment are Dystric Cambisols. The catchment is drained by a perennial stream that begins at about 900 m above sea level. Lysina represents sites with acid-vulnerable soil and water environment due to base-poor rock (Kram et al., 2012). The main focus is on description of long-term hydro-biogeochemical patterns in the magnesium-poor and acid-sensitive Lysina catchment. Research activities have started in 1988 and include among others: study of element fluxes and pools, wet and dry deposition, internal recycling in trees, soil exchange processes, chemical weathering, nutritional status of trees and toxic metals speciations, modeling predictions of hydrologic, hydro-chemical and soil chemical status.

The selected larger area which receives geostatistical knowledge from Lysina CZO is Slavkovsky Forest. The Slavkovsky forest area is located in North and North West part of Czech Republic in the region of Bohemia. The mean altitude is about 700m above sea level. The parent material is mainly granite on which podzolic have been developed. Minimum and maximum temperatures are for January - 51° to - 02 °C and for July 10.5-21.5 °C. The annual average precipitation is about 1094 mm. The natural vegetation of the area is mainly beech and spruce forests, peat bogs, and pine forests on serpentine (Neuhauslova & Moravec 1997). The forest area of Slavkovsky is about 2,000 km2, while there is a protected area (of high interest for biodiversity) inside the forest of about 617 km2.

2.2. Input Data

In Lysina CZO, 32 sample points provided by SoilTrEC researchers (Kram & Lamakova) were used as input data for mapping Soil Organic Carbon (SOC) content in the CZO area. Basic statistics are shown in the Table 1.

Table 1. Soil organic carbon basic statistics in Lysina CZO.

Count Minimum Maximum Mean Standard Deviation

32 2.1% 45.0% 11.2% 8.7%

Auxiliary data for the area were derived from a DEM with a 100m resolution. Topographic parameters such as Elevation, Slope, Aspect, and Topographic Wetness Index (TWI) were used as predictors in the regression model to produce a SOC surface map in Lysina CZO. The TWI is a mathematically simple parameterisation of soil moisture status. The index is based and calculated on slopes and depends therefore on digital terrain data. It describes the tendency of water to accumulate at any point in a catchment (Haas, 2010). It is worthwhile to underline that the most important feature affecting soil organic carbon is land cover/use which is the same (Forest) for the whole Lysina CZO and the Slavkovsky area.

In the Slavkovsky Forest, SOC Map has been produced by using the rules were transferred from the regression model which is used to map SOC in Lysina CZO. Only 5 points of the LUCAS are within the Slavkovsky region and have been used to calculate and to propagate the errors. Lucas is a soil survey project which has approximately 20,000 topsoil samples were collected in 25 European Union (EU) Member States with the aim to produce the first coherent pan-European physical and chemical topsoil database, which can serve as baseline information for an EU wide harmonized soil monitoring (Toth et al, 2013).

2.3. Method

The principle of the method is transferring the statistical model from smaller reference area (Lysina) to a larger area (Slavkovsky Forest) (Figure 1).

Figure 1. SOC Modelling Workflow for Lysina CZO to Slavkovsky Forest Area

In the first step of the overall process, SOC map of Lysina CZO were produced by applying the regression kriging sub-model (Part 1 of Figure 1). The application of regression kriging generates coefficients for each of the predictors (namely Regression rules). In practice, the regression rules are the knowledge of different processes affecting soil organic carbon and taking place at CZO level. In the second step, the transfer sub-model receives the knowledge (regression rules) from the reference area (CZO) and applies to larger area. In the model, the constant is the land cover in Lysina and Slavkovsky Forest Area (Forest). The transfer sub-model uses additional soil organic carbon data (coming from National/European datasets) for developing the 'Regression map' of the large areas. In the model which is estimating SOC in CZO (Lysina), the low R2 (0.31) indicator value reflect the complex nature of the soil organic carbon distribution, determined by other important predictors, which could not be used in the models due to lack of variation (Land Cover, Climate, NDVI-vegetation). Those predictors (Land cover, Climate, vegetation) are very important for identifying the soil organic carbon variation. Even if it was not possible to use them in CZO level, they were used in the extrapolation part of the transfer sub-model. This part is very important for developing the 'Error map' in the larger area (Slavkovsky Forest Area).

3. Results

The final soil organic carbon prediction map for Slavkovsky Forest area is the sum of the 'regression map' and the 'error map'. Error map was produced by kriging of the residuals to minimize error variance. The regression model equation predicts SOC content in Lysina CZO and also in Slavkovsky Forest (Equation 1).

Equation 1.

SOC = 1.02 + (Aspect * 0.08) + Elevation * 0.0012 + Slope * 0.015 + (TWI * 0.003)

The output map of Soil organic carbon (SOC) content in Lysina CZO and the upscaled region (Slavkovsky Forest) are demonstrated in figure 2 and the basic statistics are shown in the Table 2. The SOC values in the Slavkovsky Forest are ranging between 0 % - 35.11%.

Lysina --> Slavkovsky Forest

High : 35.11

-Reichenbach

-TO-ZD"

Sources: Esri, USGS. NOAA

u Kilometers

Figure 2. Soil Organic Carbon Map of Slavkovsky Forest Area

Since the main land cover type is forest (with many peat blogs as referred in the study site description), the SOC content is high in Slavkovsky area as expected.

Table 2. Statistical indicators for spatial extrapolation exercises (SOC %)

Layers/Model R2 Maximum Minimum Mean Standard Deviation

Lysina SOC Map 0.31 32.77 0 11.26 5.87

Lysina > Slavkovsky Forest - 35.11 0 17.91 6.54

3.1. Validation

The model application has been validated using independent dataset. In 2003, the University of Agriculture (Prague, Czech Republic) has delivered to the Joint Research Centre 275 sample data on soil organic carbon covering the whole country for the development of DanubeSIS dataset (DanubeSIS, 2004). The DanubeSIS dataset is currently hosted in the European Soil Data Centre (ESDAC) (Panagos et al., 2012) and is used for modeling purposes. The DanubeSIS samples "falling" in the upscaled study area were used as validation. A basic validation was carried out by using DanubeSIS data points to compare model outputs for Slavkovsky Area. The linear relationships were determined between predicted and measured values. The Pearson Correlation found as 0.61 (p=0.03) to the SOC prediction map of Slavkovsky Forest.

4. Conclusion

In this modelling study, it was demonstrated that how and to what extent the natural processes can be extended from reference areas to larger areas and how the local (CZO) information is transferable to larger areas (Regions, Catchments) using geostatistical methods. The results of the study encourage applying similar approach at a wider scale, and in other regions. However, the ground data availability is still the key component to have more robust geostatistical models. The model and its outputs can be improved by using more ground data and high resolution environmental covariates.

A new combined modelling approach (digital soil mapping, transfer of knowledge) was tested here to produce soil organic carbon map in Lysina CZO and to transfer the knowledge and to extend the predictions to larger area (Slavkovsky Forest Area). The results of the linear regression procedure is promising, however, the statistical indicators are relatively low (R2 = 0.31). The regression combined with kriging of the residuals named regression-kriging produced here acceptable statistical results and reasonable spatial distribution of the SOC.

The methodology applied in this study to map SOC in the CZO and in the selected larger area, uses datasets based on existing predictors at European scale, thus it can be repeated anywhere in the Europe and even at the global scale as per their availabilities, where the sufficient input soil data is available. Digital elevation information and derived parameters are among the best environmental predictors which were already used in our models. However, the correlation between these data layers and SOC depends highly on the data quality and the other environmental conditions. In terms of a well-defined spatial soil inference system, more potential pre-existing input data could be used to run the model and refine the procedure to have better outputs to fit to our needs.

References

1. EC, 2006, Communication from the Commission to the Council, the European Parliament, the European Economic and Social Committee of the Regions Thematic Strategy for Soil Protection, (COM(2006) 231 final) (2006)

2. Kibblewhite, MG, Jones, RGA, Montaranella, L, Baritz, R., Huber, S., Arrouays, D., Micheli, E. and Stephens, M. (eds) (2008) Environmental Assessment of Soil for Monitoring: Volume VI Soil Monitoring System for Europe. EUR 23490 EN/6 Office for the Official Publications of the European Communities, Luxembourg, 72 pp.

3. EC, 2011, Communication from the Commission to the Council, the European Parliament, the European Economic and Social Committee of the Regions Roadmap to a Resource Efficient Europe (COM(2011) 571 final) (2011).

10. 11.

Panagos P., Hiederer, R., Van Liedekerke, M., Bampa, F. (2013) Estimating soil organic carbon in Europe based on data collected through an European network Ecological Indicators 24, pp. 439-450.

King D, Daroussin J, Tavernier R 1994: Development of a soil geographic database from the Soil Map of the European Communities.

CATENA, 21 (1), 37-56.

Jones A., Panagos P., Barcelo S., Bouraoui F., Bosco C., Dewitte O, Gardi C., Erhard M., Hervas J., Hiederer R., Jeffery S., Lükewille A., Marno L., Montanarella L., Olazabal C, Petersen J. E., Penizek V., Strassburger T., Toth G., Van Den Eeckhaut M., Van Liedekerke M.,Verheijen F., Viestova E., Yigini Y., 2012, The state of soil in Europe, JRC reference report. Joint Research Centre of the European Commission, Ispra

Dobos, E., Carre, F., Hengl, T., Reuter, H., and Toth, G., 2006. Digital Soil Mapping as a Support Production of Functional Maps. Office for Official Publications of the European Communites, Luxemburg. EUR 22123 EN, 68p.

Grunwald S., Chapter 1. Current state of digital soil mapping in the Wester USA. In: Boettinger, J.L.; Howell, D.W.; Moore, A.C.; Hartemink, A.E.; Kienast-Brown, S. (Eds.) Digital Soil Mapping Bridging Research, Environmental Application, and Operation, 2010, 473p., Springer Dordrecht Heidelberg London New York, ISBN 978-90-481-8862-8.

Lagacherie, P., Robbez-Masson, J.M., Nguyen-The, N., Barthes, J.P., 2001. Mapping of reference area representativity using a mathematical soilscape distance. Geoderma 101, 105-118.

Stein, A., Riley, J., Halberg, N., 2001. Issues of scale for environmental indicators. Agric. Ecosyst. Environ. 87, 215-232. Banwart S., Menon M., Bernasconi S.M., Bloem J., Blum W.E.H., Souza D.M.D., Davidsdotir B., (...), Zhang B. (2012). Soil processes and functions across an international network of Critical Zone Observatories: Introduction to experimental methods and initial results. Comptes Rendus - Geoscience, 344 (11-12) , pp. 758-772.

Kram P., Hruska J., Shanley J.B.Streamwater chemistry in three contrasting monolithologic Czech catchments (2012) Applied Geochemistry, 27 (9), pp. 1854-1863.

Navratil T., Kurz D., Kram P., Hofmeister J., Hruska J.Acidification and recovery of soil at a heavily impacted forest catchment (Lysina, Czech Republic)-SAFE modelling and field results(2007) Ecological Modelling, 205 (3-4) , pp. 464-474. Neuhäuslova, Z., and J. Moravec, editors. 1997. Map of potential natural vegetation of the Czech Republic. Academia, Prague. Haas J., 2010, Soil moisture modelling using TWI and satellite imagery in the Stockholm region. University essay from KTH/Geoinformatik och Geodesi. Master's of Science Thesis in Geoinformatics TRITA-GIT EX 10-001 School of Architecture and the Built Environment Royal Institute of Technology (KTH) 100 44 Stockholm, Sweden

Toth G., Jones A., Montanarella L (eds), 2013, LUCAS Topsoil Survey, Methodology, Data and Results. European Commission Joint Research Centre, Institute for Environment and Sustainability. JRC Trchnical Reports, Report EUR 26102 EN. DanubeSIS, 2004. Soil Information system on Danube River Basin.Address: http://eusoils.jrc.ec.europa.eu/projects/danubesis/index.htm (Accessed November 2013)

Panagos P., Van Liedekerke M., Jones A., Montanarella L. (2012). European Soil Data Centre: Response to European policy support and public data requirements Land Use Policy, 29 (2) , pp. 329-338.