hul 2004
inciples,
cators as
ment in
itors of
1-132
'ineering
er Sub-
raphical
'ersity of
ly of the
dation in
s of the
sing in
> Gorges
tification
rnational
. landuse
erranean
onmental
Systems.
nics and
iphy, 70:
izing. In
nceptual
y "Press:
egetation
In]. "B.
nd Sons,
ican. In:
tification
al de los
ion. PhD
Alcalá de
Land
| Control
anges in
y Landsat
al Issue:
itterns of
^ Project
x EVK2-
- Vicente
rsitat de
LAND COVER DYNAMIC MONITORING IN THE REGION OF COQUIMBO (CHILE)
BY THE ANALYSIS OF MULTITEMPORAL NOAA-AVHRR NDVI IMAGES
: ^ ab : . " : ; :
Luis Morales S.*", Giorgio Castellaro G.®, José A. Sobrino°, Jauad El Kharraz"
? Department of Physics, Metropolitan Technological University, POBox 9845, Santiago, Chile-luis.morales@utem.cl
® Faculty of Agronomic Sciences, University of Chile, POBox 1004, Santiago, Chile- giolucas@ctc.internet.cl
* Global Change Unit, University of Valencia, POBox 46100, Burjassot (Valencia), Spain-(sobrino,jauad)@uv.es
Commission VI, WG VI/4
KEY WORDS: Land cover, Dynamic, Multi-temporal, Statistics, Classification, Simulation, Climate, Cartography
ABSTRACT:
The current work presents a new method for the land cover dynamic monitoring based on the inter-annual variability of the NDVI
obtained from multi-temporal NOAA-AVHRR time series. It consists in using the mean annual values of NDVI and its spatial
variability in the study area. A matrix relating the statistics was designed in order to calculate the spatial homogeneity index (SHI),
which is a good way to classify the variability of the NDVI. This classification permits the identification of areas with homogeneous
behaviour in terms of inter-annual land cover dynamic. Moreover, with the aim to compare the aforementioned classification with
the superficial Biomass, the potential yields of the natural prairies were modelled in the study area at the same spatial resolution of
NOAA. A model based on simulation of the prairies growth and which integrates the main eco-physiologic processes and its
climatic regulation and maturity, was developed. The last results together with the obtained classification permit a development of
land cover cartography.
1. INTRODUCTION
In order to evaluate any process related to the vegetation
dynamics from satellite data, it is necessary to take into account
indicators that are related to the biophysical parameters which
suitably describe the vegetation physiology. These parameters
in the optical, thermal or the microwaves spectrum; can be used
altogether or to be combined generating complex indices. This
is sustained in the fact that all land cover changes, modify its
spectral behaviour, in such way that this premise allows the
monitoring of the vegetation dynamics in the land surface by
the use of satellite data (Pearson and Miller, 1972; Rouse et al.,
1976; Huete, 1988; Goward, 1989; Kaufman and Tanré, 1992;
Liu and Huete, 1995). Among other parameters, the vegetation
indices present a high correlation with some parameters related
to the vegetation, such as the total biomass and the leaf area
index (LAI), which makes it an efficient tool for vegetation
studies (Curran, 1980; Jansen, 1983; Tucker and Sellers, 1986;
Diallo et al., 1990; Price, 1992; Gong and Millar, 1995). In the
field of the cartography of the spatial-temporal variability of the
vegetation, numerous methods of analysis have been published,
being based many of them on the normalised difference
vegetation index (NDVI) (Tucker et al. 1985; Townshend et al.
1987; Loveland et al. 1991; Lambin & Strahler, 1994). In
addition some investigations have introduced the land surface
temperature (LST) as an analysis parameter for the study of the
land cover changes (Lambin & Ehrlich, 1996; Raissouni &
Sobrino, 1998; Sobrino & Raissouni, 2000). This last parameter
is less used, owing fundamentally to the difficulty to obtain it.
But, whatever it is the case, these analyses are based on the
exhaustive knowledge of the biophysical relations between the
used parameters and the ecological variables (Nemani &
Running, 1989; Goward et al. 1985; Nemani ET to. 1993; The
Friedl & Davis, 1994).
From these premises, we have developed a method for the
monitoring of the land cover dynamics based on the inter-
>
559
annual variability of the NDVI, obtained from a time series of
NOAA-AVHRR images.
STUDY AREA
CHILE (%
SUD AMERICA
Figure 1.- Study area: Limari province, which is located in the
[Vth region of Chile. This zone corresponds to a transition
between the arid climate and the Mediterranean one known as a
semi-arid zone.
2. STUDY AREA AND SATELLITE DATA
The total of the population in 1992 was of 504 387 inhabitants
(INE, 1992). In average, 30% of the total number of inhabitants
of the IVth Region live in rural areas. Some of the communes
are typically rural like: La Higuera, Hurtado river. & Paihuano,
where 100% of the population live in rural areas, but also
Vicufia, Patria Mount, Combarbalá, Punitaqui and Canela have
more than 2/3 of the population living in rural areas. In the
communes who have cities or towns, like Serena and
Coquimbo, most of the population live in urban areas. The
