This presentation was presented during the 3 Parallel session on Theme 1, Monitoring, mapping, measuring, reporting and verification (MRV) of SOC, of the Global Symposium on Soil Organic Carbon that took place in Rome 21-23 March 2017. The presentation was made by Mr. Sevinç Madenoglu, from Ministry of Food, Agriculture, and Livestock - Turkey, in FAO Hq, Rome

1. Turkey’s National Geospatial Soil Organic
Carbon Information System
Bulent SONMEZ1
Aynur OZBAHCE2
Mehmet KECECI2
Suat AKGUL2
Ece AKSOY3
Sevinc MADENOGLU1
Armagan KARABULUT ALOE2
Vecihe INCIRKUS2
Ceren GORGISEN2
Tulay TUNCAY2
Ines BEERNAERTS4
Ronald VARGAS5
1
General Directorate of Agricultural Research and Policies, Ankara-Turkey
2
Soil Fertilizer and Water Resources Central Research Institute, Ankara-Turkey
3
European Topic Center, Malaga University, Malaga-Spain
4
FAO Sub Regional Office for Central Asia, Ankara-Turkey
5
FAO Headquarters, Rome –Italy

2. Introduction
 need a Digital National Soil Information System
The analysis of soil fertility parameters belong to 7742
soil samples such as N, P, K, OM, EC, pH, CEC, sand, silt,
clay, S, Fe, Cu, Zn, Mn etc.
status of soil carbon content that are essential in
the context of climate change, sustainable soil/land
management and soil ecosystem services
National Geospatial Soil Fertility and Soil Organic
Carbon Information System Project
(UTF/TUR/057/TUR)
2

3.  The Project has been carried out by Soil Fertilizer and Water
Resources Central Research Institute and is contributed by
FAO-SEC under FAO/TURKEY Partnership Programme
between 2012-2015.
 Establishing the geographical database of certain soil
properties of the Turkey’s upper agricultural soils (0-30 cm)
and of plant nutrients has been one of the targets in this
study.
 Another important aim of this project is to supply soil
organic carbon geodatabase and map for Turkey which is
the base spatial information highly related to the climate
change, soil/land degradation and soil ecosystem services.
3
Introduction

4.  Output 1: Turkey’s thematic maps of some soil quality
and fertility parameters (N, P, K, OM, EC, pH, CEC, sand,
silt, clay, S, Fe, Cu, Zn, Mn, etc.) in the scale of 1/500 000
 Output 2: Soil organic carbon and carbon stock maps
(SOC) (Updatable)
 Output 3: WEB-GIS portal to serve the result maps
and Geodata to the end users in different access
rights (universities, researchers, governmental and public
institutions, farmers, private sector, civil society
organizations etc.)
4
Introduction

5.  The research database and maps obtained by this study has an
intention to meet the important needs of related topics and
Institutions in Turkey.
5

6. Soil sampling
Soil Sampling strategy was established by interpreting
spatially explicit a number of characteristics together
such as;
- soil properties
- land use
- land use capability classes
- poorly drained alluvial soils with high water
table
- geothermal sites etc.
using different base maps such as;
- CORINE land use
- geological maps
- digital elevation models
- 1/25 000 scale digital soil maps.
6
Method

7. Soil sampling
Figure 1: Spatial distribution of soil samples
-in the period
2008-2009
-7742 topsoil
samples
(0-30cm)
-mainly from
agricultural
lands
7
Method

8. Soil analysis
7742 soil samples
(from 81 provinces
of the country)
have been analyzed
to determine soil
organic carbon
contents of the soils
in the laboratory.
8
Method

9. 9
Geodata DSM Model Prediction
Figure 2. Digital Soil Mapping Workflow
Digital Soil Mapping

10.  The SOC content is controlled by a series environmental
factors at different spatial scales, such as;
• climate variables
- temperature
- rainfall
• topography
• soil texture and parent material
• vegetation and land-use types
• human management
- land use
- land management
- land degradation
(Hassink, 1996; Percival et al., 2000; Heviaa et al., 2003; Ganuza and Almendros, 2003; Rezaei
and Gilkes, 2005; Dai and Huang, 2006; Davidson and Janssens, 2006; Liu et al., 2006; Su et
al., 2006; Sainju et al., 2008; Wang et al., 2009)
10
Digital Soil Mapping

11. 11
Auxiliary dataset
 1/500 000 scale Turkey’s geology map were used and 296
geological formation codes were re-classified (simplified into eight
classes).
 CORINE Land Cover (2000, 100 m resolution) (CLC) was used to
classify main land use types.
Soil samples’ hydrated lime, clay and sand contents were used as
driving variables. These variables were normalized and interpolated
by Ordinary Kriging interpolation method to be used as base data in
DSM.
Climatic data sets (precipitation and temperature), covering daily
meteorological data, were derived from 265 meteorological stations
distributed throughout the country.

12. 12
Auxiliary dataset
 Drought index was calculated from two climatic
parameters; long-years average of annual total
precipitation and evapotranspiration.
 Total evapotranspiration was calculated (for 265
meteorological stations) over Turkey from meteorological
parameters using Penman-Monteith method.
 Minimum, maximum and average data of NDVI vegetation
indices were used for the prediction of soil organic carbon.
NDVI composite image (for 2011) was derived from NASA.

13. 13
All input data were prepared before executing
geostatistical analysis;
• All input data were prepared using transformations for
compliant projection and coordinate system and were
resampled to the same resolution (50 m) for ensuring
compatible data structure.
• All covariates were normalized before executing the
model.
• Most of the continuous covariates (slope,
temperature, precipitation, etc.) were normalized by
using Z-score normalization technique.
Data Preparation and Processing

14. 14
Digital Soil Mapping
Multiple linear Regression-Kriging
geostatistical technique was applied
to estimate regression coefficients,
calculate residuals and determine
significant predictors for soil
assessing and producing a continuous
covers for soil organic carbon
modelling.
Residuals were interpolated by
ordinary kriging technique.
For final output, regression model of
the significant predictors and
interpolated residuals were summed
up.
Regression Kriging

15. 15
Figure 3: Semivariogram of the residuals and Q-Q Plot for soil organic carbon (Aksoy, 2014).
 Significant covariates were selected by “Akaike information criterion
(AIC)” in R.
 The AIC, which is a measure of the relative quality of a statistical
model for a given set of data, is used to select a best model for our
study.
 The chosen model is the one that minimizes the Kullback-Leibler
distance between the model and the truth.
Results

16. 16
 For validation of the model, “repeated random
sub-sampling validation” model was used by
taking averages of the values comes from 25%
validation datasets.
 The validation result was calculated using R by
taking the averages of the results comes from
25% validation datasets.
 Predicted data were evaluated with repeated
random sub-sampling validation datasets and
average R² and RMSE were found for soil organic
carbon.
Results

17. 17
Soil Properties R² Standard
Error
RMSE Mean of soil
property
Soil Organic Carbon (%) 0.324 0.439 0.422 0.89
Table 1. Regression results, Standard error of regression, Root Mean Square
Error (RMSE) of residuals and Mean of soil properties at National scale (Aksoy, 2014).
Soil Organic Carbon =
(1.075 +(0.022 *"nlime")- (0.136*"nsand") - (0.092* "nclay")+ (0.162 *"nEto")+
(0.173 *"ndrought")+(0.119 *"nndviave")+ (0.941*"nmintemp")+ (0.594*
"nmaxtemp")-(1.778*"navtemp")- (0.108* "navprec") + (0.011* "ncti")+ (0.026*
"nslope") - (0.101*"cor5") -(0.083* "cor3") - (0.145 *"cor2") - (0.061*"geo5") -
(0.084* "geo4") - (0.052* "geo2") -(0.062* "geo1") - (0.218 *"Lpe"))
Table 2. Regression equations for calculations of soil organic carbon (Aksoy, 2014)
Results

18. 18
Figure 4. SOC modelling of Some Agro - ecological Zones of Turkey

19. 19
Figure 5: Soil organic carbon map of Turkey (MoFAL,2014)

20. 20
Figure 6: Soil organic carbon stock map of Turkey (MoFAL,2014)

21. Development of WEB Based GIS Service and
GIS Portal
1. A special GIS Web Client application is been developed for
the institution.
2. The application is capable of downloading the data
3. WEB-GIS portal will serve the result maps and Geodata to
the end users in different access rights.

22. Results
 Produced soil carbon data
are used for systematic
approach of land
degradation neutrality
studies within the context
of ‘Sustainable Land
Management and Climate
Friendly Agriculture
Project
(GCP/TUR/055/GFF) at
local level in Turkey.

23. 23
 With this study, soil organic carbon (%) and soil carbon
stock were modelled and mapped by digital soil mapping
techniques for Turkey.
 The study was performed with a limited number of soil
sample dataset.
 Despite the fact that the number of the soil samples was
not sufficient to represent all land use types, except for
agricultural areas, this study is a first attempt at national
scale to be updated and improved with incoming works.
 The first version of the soil organic carbon map is being
updated under the following project in collaboration with
other relevant Institutions and research data.
Follow-up Actions

24. THANK YOU