Mapping the State and Risk of Desertification in Italy by means of
Remote Sensing, Soil GIS and the EPIC Model. Methodology Validation on
the Island of Sardinia, Italy.
*Edoardo A.C. Costantini, Michele Bocci, Giovanni L’Abate, Andrea Fais, Gabriele
Leoni, Giosuè Loj, Simona Magini, Rosario Napoli, Pasquale Nino, Massimo Paolanti, Luca
Salvestrini, Fabrizio Tascone, Ferdinando Urbano (* Contact Person) .
Experimental Institute for Soil Study and Conservation, p.za D’Azeglio 30, Firenze, 50121, Italy.
e-mail: www.soilmaps.it
Abstract
Studies have already been conducted on assessing the risk of desertification in Italy at
continental, national and regional level. They combine different climatic, soil,
vegetation, and socio-economic attributes to estimate pressure on land and state of soil
and vegetation. This study is aimed at presenting the methodology used in the creation
of the new atlas of the risk of desertification in Italy. The methodology is based on the
use of indicators of pressures, state and response. Response indicators are desertified,
sensitive and vulnerable lands. Desertified lands are intended here as tracts of land
occurring in dry climates and showing “functional sterility”, that is areas where
agriculture and forestry are currently no longer either economically or ecologically
sustainable. In vulnerable lands, environmental characteristics are close to that of a
desertified area, but some factors, e.g. vegetation cover or irrigation, successfully
mitigate the process of desertification. Sensitive lands are surfaces where processes
leading to desertification are active, although lands do not yet have functional sterility.
Key words : atlas, desertification, methodology, Italy.
Introduction
Studies regarding mapping the desertification risk in Italy have already been carried out in
the ambit of maps produced at global (Eswaran and Reich, 1998) and continental scales
(DISMED, 2003). A good first appraisal of the magnitude of the problem can also be obtained
by means of the satellite image of the Mediterranean basin, where areas deprived of
vegetation are clearly identifiable all around the Mediterranean sea (Fig.1). At national scale,
a first approximation of the areas at risk of desertification in Italy, based mainly on climatic
and socio-economical data, was produced in the year 1998 by the National Technical
Services. The map indicated only small areas of southern Italy and of the major islands
(Sardinia and Sicily) as being at risk (Comitato naziona le per la lotta contro la
desertificazione, 1998). More recently, using the MEDALUS approach (MEDALUS,
MEditerranean Desertification And Land Use, Kosmas et al., 1999) the European
Environment Agency (DISMED, 2003) worked out a new map of the desertification risk in
Italy at 1:250,000 scale. The areas at risk of desertification were singled out by means of three
indices, i.e. quality of climate, soil and vegetation. The map indicated as being at risk large
parts of central and southern Italy, as well as of the Po Plain and the Alps. On the other hand,
some areas of southern Italy which were formerly considered at risk were now excluded.
Fig. 1 Satellite image showing areas deprived of vegetation all around the
Mediterranean sea. Arrow indicates the island of Sardinia.
Other maps of desertification risk have been drawn up/compiled at regional level, using
modified MEDALUS approaches and data coming from regional databases (UNCCD-CRIC,
2002).
In the year 2003, two research Institutes of the Ministry for Agricultural and Forestry
Policies, the Experimental Institute for Soil Study and Conservation (ISSDS) and the National
Institute of Agricultural Economics (INEA) were entrusted by the Italian Ministry for the
Environment with the responsibility of preparing a map of the desertification risk in Italy at
the semi-detailed scale (reference scales 1:100,000 – 1.250,000).
The aim of this work is to present the rationale of the research, the methodological
approach, and the application and validation experience. The territory used to validate the
methodology was the island of Sardinia and the desertification processes considered were: i)
soil losses by water erosion and mass movements, ii) salinization, iii) urbanization.
Rationale of the work
The most widely accepted definition of desertification is, at present, the one given by the
United Nations Convention to Combat Desertification (http://www.unccd.int). It defines
desertification as “land degradation in arid, semi-arid and dry sub-humid areas resulting
from various factors, including climatic variations and human activities”. Land is here
defined as the terrestrial bio-productive system which includes soil, vegetation, other biota
and the ecological and hydrological processes that operate within the system. Land
degradation means the reduction or loss of the biological and economic productivity and
complexity of irrigated and non- irrigated agricultural land, pastures, rangeland, forest and
woodland. Arid, semi-arid and dry sub-humid areas are the territories, other than polar and
sub-polar regions, in which the aridity index, that is the ratio of annual precipitation to
potential evapotranspiration, falls within the range from 0.05 to 0.65.
This set of definitions is tailored for evaluations produced at a global or continental scale,
but it is too vague for studies at national or local scales. The scaling down needs a more
precise conceptual framework, along with more detailed delimitations of the areas susceptible
to desertification.
The concept of land degradation, which means impairment of soil qualities, must be
separated from that of land desertification, which is a particular kind of land degradation,
implying the loss of capability for a sustainable agricultural and forestry production, this loss
being irreversible or having very little chance of reversibility. A desertified area is a tract of
land occurring in dry climates and showing “functional sterility”, that is an area where
agriculture and forestry are currently no longer economically or ecologically sustainable. A
desertified area could be cultivated, but the economic and social input needed would be so
high that only rarely could it be put into practice. The level of economic and social input can
obviously vary from country to country, according to the minimal accepted tenure of life and
environmental awareness of the population. Taking as a reference the well know and widely
accepted Land Capability classification system (Klingebiel and Montgomery, 1961), soils
with functional sterility belong to the last class, the eighth, that is soils which are left set
aside, and only considered for ecological purposes. As a consequence of the relative
assessment of desertified lands, a proper evaluation should be validated with an inventory of
benchmark lands, clearly showing field evidence of desertification.
Fig. 2 South of Italy: bare lands on steeper slopes are desertified, where still
protected by woodland they are vulnerable. Fire or excessive forest exploitation would
subject vulnerable areas to rapid desertification.
In transitional zones, such as in Mediterranean environments, desertified lands are very
often intermingled with areas at risk of desertification. An area at risk of desertification is a
tract of the earth surface which is vulnerable or sensitive to the processes of desertification. In
a vulnerable land, environmental characteristics, like soils for instance, are close to that of a
desertified area, but some factors, e.g. vegetation cover or irrigation, successfully mitigate the
process of desertification (Fig. 2). On the other hand, a sensitive land is a surface where the
process leading to desertification is active, although the land does not yet have functional
sterility (Fig. 3).
With reference to the PSR framework (Pressure, State, Response), desertified lands are
the ecosystemic responses to the pressures on the environment. State factors of land
desertification are the environmental characteristics (indicators, Enne and Zucca, 2000) which
regulate the occurrence of desertification processes, namely climate, soil, vegetation, water
availability, human pressure and management. Pressure factors are the processes of
desertification, e.g. soil erosion, contamination, salinization, and drought. Urbanization is
considered a desertification process in the sense that it induces the irreversible loss of
agricultural and forestry functionality, although the land acquires other functions. A major
concern is produced by the diffusion of urbanized areas where the problem of lack of water is
more acute. This is the case of the coasts of the Mediterranean sea. If we compare the
distribution of cities around the Mediterranean basin (Fig. 4) to the satellite image of areas
deprived of vegetation (Fig. 1) we can clearly identify the tracts of the coast where there is
interaction between urbanization and loss of the vegetation cover.
Fig. 3 An area of Sardinia sensitive to desertification because of excessive grazing
and strong sheet water erosion.
Choice of the indicators of desertification
The indicators used changed according to the step of the evaluation (Table 1). The first
step was the delimitation of the area of the country at potential risk of desertification. The
indicators used were aridity index, pedoclimatic regimes and climatic regions. The second
step consisted in the mapping of the risk at reference scale.
The choice of the indicators was made taking into account the parameters suggested by
previous experiences (Enne and Zucca, 2000), as well as the actual possibility of having
specific databases at national scale.
Fig. 4 Satellite image showing distribution of main cities all around the
Mediterranean sea.
Table 1. Indicators considered in assessing the areas sensitive and
risk of desertification and drought in Italy.
PRESSURE
STATE
Slope
Soil characteristics
Rooting depth
SURFACE
Presence of rills and gullies
DENUDATION
Normalized Difference
BECAUSE OF
Vegetation cover
Vegetation Index (NDVI)
WATER
Number of cattle and sheep
EROSION
Human activities
Protected areas
Burnt areas
URBANIZATION Urban areas
Distance from the sea
SALINIZATION Elevation
Irrigated vulnerable lands
Climatic region
Aridity index
Soil moisture and temperature regimes (following
DROUGHT
Soil taxonomy)
Mean annual number of days when the soil is dry
vulnerable to the
RESPONSE
Vulnerable lands
Sensitive lands
Desertified and
sensitive lands
Risk increase
Risk mitigation
Risk increase
Desertified lands
Vulnerable lands
Sensitive lands
Potentially
vulnerable lands
(national scale)
Vulnerable lands
(regional scale)
Delimitation of the area in Italy at potential risk of desertification
We elaborated a series of 656 meteorological stations with the aim of characterizing
aridity index and 140 stations to evaluate pedoclimatic regimes of the whole country. We also
considered pedoclimatic regimes because soil is capable of storing water and mitigates
drought risk and temperature excursions in the root zone. Soil classification according to Soil
Taxonomy (Soil Survey Staff, 1999) considers soil moisture and temperature regimes:
“aridic”, “xeric”, “dry xeric” and “ustic” soil moisture regimes can be identified for areas with
varying degrees of potential water deficit (Table 2). Moreover, soils with “thermic” and
“hyperthermic” temperature regimes refer to lands with high temperatures in the root zone,
which can enhance the decaying of the organic matter, especially under agricultural land use
(Table 3). Soil moisture regime was obtained using EPIC daily outputs (see below), soil
temperature regime from the algorithm proposed by Costantini et al. (2001), which makes use
of annual mean air temperature and soil water field capacity.
The aridity index was obtained applying the Hargreaves-Samani methodology (1982) on
the long term mean monthly temperature and rainfall values of the 656 meteorological
stations, and it was spatialized with the Inverse Distance Weighting method (Fig. 5).
The aridity index map and pedoclimatic regimes were matched with the climatic regions
of Italy (Finke et al., 1998; Righini et al., 2001). Potential desertification risk resulted in all
climatic regions with a Mediterranean type climate (Fig. 6). The area seems to match well
with the satellite image of Figure1.
Table 2. Soil moisture regimes (Soil taxonomy 1999, simplified)
Aridic: i) the Soil Moisture Control Section (SMCS) is completely dry for more than 180
d; ii) the SMCS is completely moist for less than 45 d in the 4 months following the winter
solstice.
Dry xeric: i) exclusive of aridic conditions; ii) the SMCS is completely dry for more than
89 d; iii) the SMCS is completely moist for more than 44 d in the 4 months following the
winter solstice.
Xeric: i) exclusive of aridic and dry xeric conditions; ii) the SMCS is completely dry for
at least 45 d in the 4 months following the summer solstice, iii) the SMCS is completely moist
for more than 44 d in the 4 months following the winter solstice.
Ustic: i) exclusive of aridic, dry xeric and xeric conditions; ii) one or more layers of the
SMCS are dry for more than 89 d.
Udic: remaining conditions.
Table 3. Soil temperature regimes at 50 cm depth (Soil taxonomy 1999, simplified)
Cryic: the annual mean soil temperature is lower than 8 °C and the summer temperature is
lower than 15 °C
Frigid: the annual mean soil temperature is lower than 8 °C
Mesic: the annual mean soil temperature is between 8 and 22°C
Thermic: the annual mean soil temperature is between 15 and 22°C
Hyperthermic: the annual mean soil temperature is higher than 22 °C
The Iso prefix is added when mean summer and winter soil temperatures differ less than 6
°C.
Å
Bolzano
0
0
Aosta
Milan
0
0
Turin
0
Trento
Trieste
0
0
Venice
0 50 100
Bologne
0
0
Genoa
Florence
0
200
300
400
Km
Ancona
0
0
Perugia
Pescara
0
0
Rome
Naples
0
0
Potenza
Bari
0
0
Cagliari
0
Catanzaro
0
Palermo
ARIDITY INDEX
0.20 - 0.50
0.50 - 0.65
0.65 - 1.00
>1.00
METEOROLOGICAL STATION
Fig. 5 Aridity index and meteorological stations
Å
k
0 50100
k
k
300
400
Km
F
G
k k
F
G
FG
G
F
G G
F
FF
G
kk
FG
G
F
F
G
F
G
kG
F
F
G
kk
k
k
F
G
/
"
/
"
F
G
k
kk
/
"
Fkk k
G
k "
/
#
F
G
0#
F
G
0
#
0
/
"
F
G
/
"
F
G
#
0
F
#
F
G
0 G
F
G
/
"
F
G
F
#
/G
"
0
F
G
/
"
F
"G
/
F
G
/
"
/ "
"
/"
/
200
/
"
0
#
#
0 #
0
F
G
/
"
#
0
.
!
CLIMATIC REGION
PEDOCLIMATIC RISK
Temperate mountain
Warm Temperate sub-continental
0 - NO RISK
Temperate suboceanic
k
1 - THERMIC - (((DRY XERIC)))
Mediterranean mountain
F
G
2 - THERMIC - ((DRY XERIC))
Mediterranean sub-oceanic
#
0
3 - THERMIC - (DRY XERIC)
Mediterranean sub-continental
/
"
4 - THERMIC - DRY XERIC
Mediterranean sub-tropical
.
!
5 - HYPERTHERMIC - DRY XERIC
Fig. 6 Pedoclimatic drought risk and climatic regions of Italy
Calibration of the Epic model
The EPIC (Environmental Policy Integrated Climate, former Erosion-Productivity Impact
Calculator, Sharpley and Williams, 1990) model (version 3090) was used on a daily time step.
Daily inputs were minimum and maximum air temperature, rainfall, and radiation. Potential
evapotranspiration was calculated according to Priestley-Taylor (1972). Monthly means came
from long-term series of data, daily statistics from long-term or a limited number of years.
The reference crop was a stable meadow, grown without irrigation and maintained at about
0.20 m high. Optimal temperature for plant growth was 20 °C for warmest climate and 15 °C
for the others. Soil inputs were horizon depth, texture, rock fragments, bulk density, water
content at field capacity and wilting point, and organic carbon. Field capacity and wilting
point were measured in laboratory or estimated according to Baumer’s method (provided by
EPIC). Soil bulk density was measured in the field or estimated by EPIC. Soil horizons depth
was arranged to have four layers corresponding to the moisture control section of each studied
soil. EPIC was initialized at field capacity and, using the weather generator, EPIC ran for a
50-year period of time. Daily soil moisture of the layers forming the control section was
extracted from the output file of EPIC with a software created with visual basic language
A set of linked spreadsheets of Excell were used to: i) draw graphs of the climate and soil
water content of each layer; ii) elaborate daily data and classify soil moisture and temperature
regimes of the 50 years generated by Epic. The methodology was tested in experimental
fields, where soil water content was measured weekly or biweekly (Costantini et al., 2002)
and applied at regional and national scales on benchmark soils and meteorological stations.
Use of the Epic model
Soil moisture regime and average cumulative days per year when the soil moisture control
section is completely dry were estimated by means of the EPIC model, the National Soil
Database and long term climatic data. The EPIC model was also used to estimate the mean
annual number of days when the soil moisture control section is dry. A multiple regression
deriving the number of days from long term mean air temperature, annual rainfall and
available water capacity (difference between soil water at field capacity and wilting point;
AWC) was found (R2 = 0.55; F<0.00001; n=260). The regression was applied to the AWC of
13,000 soils and to air temperature and rainfall of the soil sites, obtained through ordinary
kriging of the long term annual values of 1067 (total rainfall) or 944 (mean temperature)
meteorological stations. The indicator “mean annual number of days whe n the soil moisture
control section is dry” was finally added to each soil of the National Soil Database.
Application of the method to the island of Sardinia
The National Soil Database, having data of 792 profiles in Sardinia, was queried to select
soil characteristics used to qualify the state indicators “rooting depth”, “presence of rills and
gullies”, and “mean annual number of days when the soil is dry” (Figs. 7 and 10).
Soil information was generalized using land components of the land subsystems
(1:250,000), linked to the National Soil Database. Land subsystems are geographical units
with a characteristic pattern of lithologies, morphologies and land uses (Costantini et al.,
2003). A land component is a part of a land system, not delineated in the GIS but stored in the
database, formed by a combination of morphology, lithology and land use, with legends
created on the basis of pedolandscape perception at the reference scale.
NDVI analysis of “natural areas” (woodlands and pastures) was calculated on Landsat TM
images at 30 m resolution. The result was classified in different land typologies by means of
field evidence and sample points on aerial photographs. Desertified and sensitive lands were
selected from the classification (Fig. 8).
An original land cover database was created at 1:100,000 by means of aerial
photointerpretation, which was used in combination with the other databases as well as in the
delineation of urban areas and irrigated lands. A Digital Elevation Model at 20 m resolution
was used to calculate slope, distance from the sea and elevation.
The remaining state indicators came from specific databases.
Å
Å
LEGEND
0
25
50
LEGEND
No data
Not pasture or
natural lands
Not affected
Not affected
Vulnerable
Sensible
Sensible
Desertified
100
Kilometers
0
25
50
100
Kilometers
Fig. 8 Desertified and sensitive lands
Fig. 7 Lands sensitive and vulnerable to
because of lack of vegetation cover.
erosion because of soil characteristics.
State indicator “NDVI analysis”
State indicators “rooting depth”,
“presence of rills and gullies”, and “slope”
Å
Å
LEGEND
LEGEND
Water bodies
No data
Not affected
Not affected
Vulnerable
Vulnerable
Sensible
0
25
50
100
Kilometers
Fig. 9 Lands sensitive and vulnerable to
salinization. State indicators “distance
from the sea”, “elevation”, and “irrigated
lands”
0
25
50
100
Kilometers
Fig. 10 Lands vulnerable to drought.
State indicator “mean annual number of
days when the soil is dry”
PHOTO-ATLAS
3
Site: Irgoli
Sensible area because of grazing
on steep slopes and shallow soils
Site: Tertenia
Sensible area because of fires on
steep slopes and shallow soils
Site: Teulada
Sensible area because of fires and
overgrazing on shallow soils
0
25
50
100
Kilometers
Fig. 11 GIS of the “true points”
Validation with “true points” on the field
The validation of the method was obtained by means of a set of field observations, stored
in a specific database. Each observation was georeferentiated and briefly described in terms of
state and causes of desertification. A picture of the landscape and an aerial photograph of the
site was added to the site information. The observation was linked to the geographic database
and therefore had information about geology, morphology, land use and soils of the sites (Fig.
11).
Conclusions
The elaboration of the atlas of the desertification risk in Italy is not completed yet,
however, the first results are promising. NDVI analysis resulted a powerful tool to single out
desertified and sensitive lands, especially because it was combined with a land use database,
however its use was limited to the not ploughed soils.
The evaluation of sensitive and vulnerable lands because of water erosion, through the
analysis of soil characteristics, gave important information both in agricultural and not
agricultural areas, but envisaged a numerous and well organized soil database, harmonized
and linked to a proper geography at the national scale.
The use of the EPIC model permitted the evaluation of the drought risk at the national
level, through the estimation of the soil moisture and temperature regimes, as well as at the
regional level, by means of the estimation of the long term mean annual number of days when
the soil is dry. The evaluation of potentially vulnerable lands through the first pedoclimatic
indicators, in particular, resulted far more accurate than that obtained with the traditional
climatic indicator, the aridity index. On the other hand, the indicator “mean annual number of
days when the soil is dry” was particularly useful for providing information about the land
vulnerability in agricultural areas, where NDVI analysis could not be performed.
In conclusion, it must be stressed that the success of the proposed methodology relied on
the relevance and straightforwardness of the indicators used at the different scales, as well as
on the field survey and storage inside a GIS of tangible benchmark sites, where desertification
state and risk were clearly identified, explained and shown using pictures.
References
Comitato nazionale per la lotta contro la desertificazione (1998) Comunicazione nazionale alla convenzione
delle nazioni unite per la lotta contro la desertificazione nei paesi gravemente colpiti dalla siccità e/o
desertificazione,
con
particolare
urgenza
in
Africa
(UNCCD)
[online]
www.desertification.it/doc/TESTOFIN.htm#_ftn52 (verified on March 2004).
Costantini E.A.C., Castelli F., Iori M., Magini S., Lorenzoni P., and Raimondi S. (2001) Regime termico del
suolo in alcuni campi sperimentali del nord, centro e sud Italia. Atti convegno SISS, Gressoney Saint Jean,
1999, (CD-ROM computer file), ISNP, Roma.
Costantini E.A.C., Castelli F., Lorenzoni P., Raimondi S. (2002) Assessing soil moisture regimes with traditional
and new methods. Soil Sci. Soc. Am. J. Soil Sci. Soc. Am. J., 66, 6, 1889-1896.
Costantini E.A.C., Magini S., and Napoli R. (2003). A Land System database of Italy. 4th European Congress on
Regional Geoscientific Cartography and Information Systems. Proceedings vol.1. p.124-126
DISMED (2003) Map of Sensitivity to Desertification and Drought in the Mediterranean Basin – Italy [online]
www.ibimet.cnr.it/programmi/Pcase/dismed_products (verified on March 2004).
Enne G., and Zucca C. (2000) Desertification indicators for the European Mediterranean region. State of the art
and possible methodological approaches, ANPA, Rome, Italy, pp. 261.
Eswaran H. and Reich P. (1998): Desertification: A global assessment and risk to sustainability. In: Proc. Of
16th Int. Congr. Soil Science, Montpellier, France. CD ROM.
Finke, P., Hartwich R., Dudal R., Ibanez J., Jamagne M., King D., Montanarella L., and Yassoglu N. (1998)
Georeferenced soil database for Europe. EUR 18092, Ispra, Italy.
Hargreaves, G.H., and Samani, Z.A. (1982) Estimating potential evapotranspiration. Tech. Note, J. Irrig. and
Drain. Engrg., ASCE, 108(3):225-230
Klingebiel A.A., and Montgomery P.H. (1961) Land capability classification. USDA-SCS, Handbook n.210,
Washington D.C. USA.
Kosmas C., Kirkby M., Geeson N. (1999) The MEDALUS project. Mediterranean Desertification and land use.
Manual on key indicators of Desertification and mapping environmentally sensitive areas to desertification.
EUR 18882, Bruxelles, Belgium.
Priestley, C.H.B., and Taylor R.J. (1972) On the assessment of surface heat flux and evaporation using large
scale parameters. Mon. Weath. Rev. 100:81-82.
Righini G., Costantini E.A.C., and Sulli L. (2001) La banca dati delle regioni pedologiche italiane. Boll. Soc. It.
Scienza del Suolo, 50, Suppl., 261-271.
Sharpley, A.N., and Williams J.R. (1990) EPIC-Erosion/Productivity Impact Calculator. USDA Tech. Bull.
1768.
Soil Survey Staff (1999) Soil Taxonomy: A basic system of soil classification for making and interpreting soil
surveys. 2nd ed. USDA-NRCS Agric. Handb. 436. U.S. Gov. Print. Office, Washington, DC.
UNCCD-CRIC (committee for the review of the implementation of the convention) (2002) Italy national report.
Report of the The Ministry of Environment and Territorial Protection. Rome, Italy, 43 pp.