Within the Land Degradation Assessment in Drylands (LADA) framework, an
analysis of land cover changes in Senegal was carried out.
The analysis covered 60 land cover classes present in the
Senegal 2005 Land Cover legend.
Particular attention was given to the most
sensitive areas such as forest, agricultural land and urban areas, using
images taken in two dates: 1990 and 2005.
The 1990's Landsat TM images have been compared with the 2005 Landsat
ETM set (see image above), and whenever a change was detected, a polygon to delimitate the
change was drown and labelled (see example below).
The basis for the analysis was the GLCN-Senegal land cover database represented
by almost 22000 polygons.
The analysis result is summarized in the graph below:
It shows the total area (sqkm) by main land cover class for the 2 dates.
There is a clear increase of agricultural and urban areas and a general
decrease in natural vegetation, except for the trees closed to open (100-40%)
aggregated class (NV_TCO).
Also, tree plantations (TP) show a growth from 1990 to 2005. Finally,
both seasonal & permanent waterbodies (WAT) and bare soil (BS) remain
almost unchanged.
The map above (click
here for a larger view) shows classes of change for agricultural
areas represented by dots.
Notes
- The scale used to detect the change was 1:100000, while the digitization
was performed at a 1:50000 scale. A general rule was applied:
where the change was not clear enough, it was not considered a change.
- The different seasonality of compared images could have created differences
not present in reality.
- Whenever the noise caused by the black strips made impossible to perform
the visual interpretation of the 2005 image, then the 2000 one was used.
- It is evident that whenever cloud coverage was present, no change was detected.
A
workshop has been held in Dakar, Senegal on 10 - 12 June 2009, to present
and discuss the results of the land cover mapping and change analysis
activities conducted in collaboration with the
Centre de Suivi
Ecologique de Dakar (CSE).
