The mapping of tsetse distribution is
one the most consolidated fields of application of RS and
GIS in the context of Tsetse and Trypanosomosis (T&T)
intervention. A number of studies have tackled the issue on
all scales ranging from continental to local. It is possible
to distinguish two classes of mapping, one based on the so-called
biological approach, the other using statistical techniques.
The biological approach is also called predictive
or process-based approach. This type of models studies the
population dynamics of a species in terms of the biotic and
abiotic mortalities. Some mortalities are driven by environmental
conditions that may be detected from satellite. If potential
fecundity exceeds mortality, presence is predicted. The number
and detail of input data that are required by this type of
models make them suitable only for small scale mapping.
The statistical technique is also known as descriptive
approach. It is based on direct correlation between species
presence or absence and a set of predictor variables such
as rainfall, temperature and relative humidity. A significant
number of parameters influencing tsetse distribution can be
derived from satellite measurements, thus it is possible to
estimate their presence over large geographical areas and
different environmental conditions. This method provides estimates
of absence/presence in a statistical way and allows the mapping
of tsetse distribution with a high accuracy over the entire
Unfortunately, tsetse presence/absence maps are not sufficient
to assess the disease challenge and to map the risk. In this
regard, a very important parameter is also the vector density.
A review of the available literature shows that the application
of RS techniques to map tsetse abundance is far less widespread
than the mapping of distribution. This is because the abundance
is determined by density dependent processes that do not necessarily
have satellite variable proxies. Nevertheless, studies to
map tsetse abundance do exist: for example the national map
in Togo at a grid resolution of 0.125° for all tsetse
species present in the country.
Some satellites provide observations on a daily basis; such
data can be analysed to depict time-dependent seasonal fluctuations
and even longer-term trends. The latter can possibly be put
into relation with the impact of climate change.
The historical classification of tsetse (genus Glossina),
based on morphological criteria, divides the species into
three groups, which have different habitat requirements that
are thought to reflect their evolutionary history. The fusca
group flies (subgenus Austenina) with supposedly the most
primitive male genitalia, occur mainly in the low-land rainforests
of West and Central Africa; the palpalis group species (subgenus
Nemorhina) occupy similar forest habitats throughout sub-Saharan
Africa and also extend into riverine and lakeside forests
or the moist areas between such forests; and finally the morsitans
group of flies (subgenus Glossina s.s.) occurs in a variety
of savannah habitats lying between the forest edges and deserts.
Tsetse live in habitats that provide shade for developing
puparia (pupae) and resting sites for adults. Their development
is temperature and humidity limited, like that of many invertebrates.
These factors together enabled the early map-makers to estimate
the distributional limits of many species of tsetse based
on vegetation type, meteorological records and altitude. Before
the remotely sensed satellite data came into practice, all
large scale distribution maps were affected by severe information
gaps, inconsistencies and inaccuracies due to the different
times and methods of survey. Remote sensing is now able to
depict many features of climate, vegetation and altitude,
making it possible to produce more precise maps of habitats
suitable for flies. Satellite images offer several advantages
over field surveys: the data are free from any human bias,
make remote places accessible, can be continuously produced
and refined, and show real-time information.
All satellite sensors show trade-offs between spectral, temporal
and spatial resolution. Data provided by the sensors need
to be processed, integrated and interpreted in the epidemiological
context under study. Most biological applications to date
have used passive satellite sensors, which detect reflections
or emissions ultimately arising from the sun’s activity.
Active sensors (radar) are gathering increasing interest,
mainly due to or also because of their ability to produce
images even under cloudy conditions.
Meteorological satellites, despite their poor spatial resolution,
provide images at a frequency that is sufficient to capture
habitat seasonality at national, regional and continental
level. Among these, NOAA-AVHRR (National Oceanic & Atmospheric
Administration - Advanced Very High Resolution Radiometer)
and Meteosat proved to be particularly useful for the study
of tsetse distributions. The quality of all imagery from passive
sensors is adversely affected by atmospheric contamination,
the most obvious of which is cloud (although other aerosols
are also important). The very high revisit frequency of meteorological
satellites permits to combine them by selecting for each pixel
the highest value from all of those recorded for that pixel
in each period of time. This is believed to be the most cloud
or contaminant-free value. The resulting images are called
maximum value composites (MVCs). Data from each of the sensor
channels can be used directly or processed to produce various
indices that are related more directly to important variable
such as soil surface temperature.
Commonly used products include: land surface temperature of
the earth’s surface (LST), derived from AVHRR channels
4 and 5; middle infrared band (MIR) from AVHRR channel 3,
also related to the temperature of the earth surface; normalized
difference vegetation index (NDVI) derived from AVHRR channels
1 and 2 and related to plant photosynthetic activity; vapour
pressure deficit (VPD), also derived from AVHRR channels 1
and 2 and effectively a measure of atmospheric drying power;
near surface air temperature (TvX) derived from LST and vegetation
index measurements; cold cloud duration (CCD) from Meteosat
that is correlated with rainfall in convective precipitation
Monthly composite imagery usually shows strong serial correlation
and therefore data redundancy, which may be eliminated in
two different ways. Either the data are subjected to principal
component analysis (PCA) and the resultant significant principal
components are used in the analyses, or the data are subjected
to temporal Fourier analysis that describes the cycles of
temperature, vegetation growth etc., in terms of annual, biannual,
triannual and other cycles with longer or shorter periods.
This method removes data redundancy and produces a set of
orthogonal (i.e. uncorrelated) outputs whilst retaining a
description of seasonality (lost in PCA) that is of vital
interest in vector and disease mapping.
The mapping of tsetse distribution takes advantage of other
data, either directly derived from satellite (Digital Elevetion
Models, Land Cover Maps, etc.) or from a combination of satellite
and ground-based meteorological data [e.g. the Length of the
Growing Period (LGP), derived from temperature, rainfall and
evapotraspiration meteorological records].
AVHRR data were used to analyse tsetse distribution in many
studies. Most of them relied on discriminant analysis while
some others used optimal threshold distribution . The section
of this site that displays maps shows an application of RS
and multivariate statistical analysis to produce tsetse species
A review of major studies dealing with tsetse mapping by means
of remotely sensed data show that the number and type of variables
that better describe fly distribution varies among different
geographical regions and eco-climatic zones. For example,
in West Africa, thermal variables tend to be more important
in describing distributions than are vegetation index (NDVI)
or rainfall (CCD) variables . In the same region, rainfall
(CCD) data were relatively more important in describing the
abundance of flies (i.e. flies per trap per day), than they
were for describing their distribution.
One criticism of the environmental envelope approach to mapping
fly distributions is that it ignores the influence of biotic
factors that are important in determining the abundance of
flies within their geographical limits . Nevertheless, the
high mapping accuracy achieved without modelling biotic factors
supports the idea that they are less important than the abiotic
ones (though it is also possible that some satellite data
are acting as surrogates for un-quantified biotic factors
In conclusion, satellites appear able to describe with remarkable
statistical accuracy the distribution of virtually all fly
species. The main problem remains the ultimate degree of accuracy
that such an approach can provide. It is not possible to achieve
the 100 percent reliability tsetse control personnel would
like, but this level is also failed by ground surveys and
it is really essential only for vector elimination purposes.
In the more common context of tsetse control, the identification
of “hotspots” and ecological corridors of reinvasion
can be described with accuracy by remote sensing techniques.
In this line of work, an effort is needed in the validation
of the models and in the utilization of new generation satellites,
which promise higher accuracy as a consequence of more data
channels and better spatial and spectral resolution.
Sunshine, rainfall and humidity, temperature,
atmospheric pressure and wind make up the climate of a particular
area. Temperature and humidity are the most important of these
for tsetse, and light can also be important.
Glossina lives well at 25-26° but if temperature goes
much higher or much lower there may be some damage to the
fly. In general a temperature above 38° is damaging to
the adults, while a temperature below about 17° will not
allow the adult to live a normal active life.
Rainfall probably does not have any direct effect on tsetse,
but does so indirectly by affecting the humidity and maintaining
different vegetation zones. A humid atmosphere allows tsetse
to spread away from sheltered habitats, so long as the temperature
is not too high.
Remote Sensing is used at a wide range of geographical scales
to monitor meteorological and climatic conditions. Several
services exist that provide up-to date information on precipitation,
moisture, temperature and other parameters that can be relevant
in tsetse and Trypanosomosis interventions. While remotely
sensed datasets have often been used to predict areas of suitability
for tsetse species in Africa, it is yet to be fully explored
to what extent they could be useful in the implementation
phase of interventions. Anomalies in the rainfall or temperature
could be used to forecast potential movements of flies’
populations and thus support control/eradication campaigns.
Similar applications are currently being experimented to monitor
zones at epidemic risk (ex. malaria).
Many meteo-climatic datasets can be either viewed on the
web or downloaded as georeferenced datasets to be used within
a GIS system. In this field, the most relevant space instruments
are “Meteosat”, operated by Eumetsat (Europe’s
meteorological satellite agency) and the “Advanced Very
High Resolution Radiometer” (AVHRR) of NOAA (USA National
Oceanic and Atmospheric Administration). Due to the size of
the areas mapped, most of the datasets regularly produced
have a relatively low resolution (usually 1 kilometre or lower)
which makes them suitable for national or regional studies.
Nevertheless, higher resolution raw datasets can be acquired
over specific areas to obtain more detailed information. In
this regard, the most interesting sensors are MODIS, mounted
on the NASA satellites Terra and Aqua, and MERIS, on ENVISAT
satellite operated by the European Space Agency.