Cambiamenti dell’uso del suolo e aumento dell’attività franosa nel territorio di Rocchetta Sant’Antonio (Appennino Dauno)

Abstract

Lamanna C. et al., Land use change and increase in landslide activity in the Rocchetta S. Antonio area (Daunia
Apennines). (IT ISSN 0394-3356, 2009)
We examine temporal variations in landslide activity in the context of land use changes and precipitation characteristics in the north-
western part of the Apulia Region (Southern Italy). The period considered is from 1976 to 2006. The main study area (72 km2) includes
the municipal territory of Rocchetta S. Antonio (FG), assumed to be a representative portion of the Daunia Apennines.
Despite the moderate elevation (from 250 to about 1000 m), the Daunia mountains are known for their susceptibility to landsliding, with rainfall events and unwise man activity being the main triggering/causative factors. Only the steepest, highest elevation areas include a significant percentage of arboreous land. Elsewhere, especially where clay-rich units predominate, the vegetation cover is represented mainly by cultivated land (cereals) and locally by grass land.
The Rocchetta S. Antonio territory is distant only few tens of km from the Adriatic coast, and, as the entire Daunia region, is characteri-
sed by a Mediterranean semiarid (sub-Apennine) climate with total annual rainfall that typically varies from 600 to 750 mm.
The clay-rich flysch formations are widespread in Daunia and that makes slopes susceptible to landsliding. The presence of intensely
deformed geological units is another factor predisposing to instability.
To investigate the differences between historical (1976) and recent (2006) landsliding we first compiled two landslide activity maps. The
1976 landslide inventory was obtained through a systematic interpretation of 1:25000 scale aerial photographs and the 2006 inventory
was based on the interpretation of high resolution (1 m pixel) IKONOS satellite imagery. For simplicity a distinction was made only
between active and inactive landslides. We followed the geomorphic criteria typically adopted for the recognition of landslides and
their state of activity from air-borne imagery. To facilitate the interpretation the satellite imagery was first orthorectified and pan-sharpened. Then natural colour imagery was draped over a detailed DTM (with 5 m grid size) generated from recent topographic maps at 1:5000 scale. For comparative analysis the information on landslides was presented by quantifying area frequency of landslides (per-
centage of the total study area occupied by landslides) and their density (number of landslides per km2). In particular, the comparison
of the 1976 and 2006 landslide inventories shows large increase in number of active landslides (from 5.7/km2 to 34.4/km2 in terms of
density), which corresponds to 160 % increase in aerial frequency.
To assess land use change occurred in the same 30 year period we used again 1976 aerial photos and produced a map of historical
land use. A map of recent land use was prepared using ASTER satellite imagery. This imagery was preferred over IKONOS data,
because the acquisition period of the ASTER imagery (summer, after harvest) resulted more suitable for land use classification based
on automatic supervised procedure. The following three major groups or classes were distinguished: i) agricultural land (sown fields
with mainly cereal cultivation ploughing), which represents by far the predominant group ii) arboreous land, including dense shrub and iii) other (including uncultivated and bare land, grass land, pasture, and man-made). The selection of these three groups was driven by the presence of few predominant classes of land cover and land use, as well as by their potential impact on shallow landsliding.
To quantify the temporal variations in land use from 1976 to 2000 the map data were expressed as percentage of areal frequency. The
results show that about 52% of the land has been used as sown fields in the seventies and that by the year 2000 the percentage
increased to 75%.
The influence of the land use change on mass movement activity is evident when considering the distributions of active landslides on
sown fields. In particular, the results demonstrate that the recent (2006) very high density (7.6%) of active landslides in the areas that
have become sown after 1976 (new sown) can be linked to the originally high susceptibility of those areas to landsliding (2.9%), exceeding even the area frequency of failures in the fields already sown in 1976 (old sown, 1.8%). In fact, new cultivated land, which is on steeper slopes than older fields, had a landslide density 55% higher than did the older fields in 2006. Furthermore, the increase in the frequency of active landsliding observed on both old and new sown fields has been greater than elsewhere (remaining areas). All this demonstrates that land use changes occurred between 1976 and 2006, and in particular the increases in the areal extent of sown
fields, have negative influence on the stability of slopes in the study area.
Clearly, the frequency of slope failures in time results from an interaction of many variables, and it is well known that the most signifi-
cant temporal controls are often those related to rainfall-induced processes. Although relations between climatic events and landslide
occurrence can be very complex, our field observations indicated that in Daunia highest landslide activity typically occurs in late winter
time. Indeed, the majority of shallow slope failures registered in 2006 were triggered by late February and March rainfalls that followed
a wet winter period. A comparison of the precipitation characteristics in 1976 and 2006 did not show major differences. In particular,
by taking into account potential evapotranspiration in the months preceeding the two landslide inventorying periods we demonstrate
that potential water input to the soil differed by about 20%.
In summary, the results of this work demonstrate clear relations between the land-use changes in the period 1976-2006 and the recent
high landslide activity. We thus stress that for hillslope areas profoundly altered by man in the recent decades, the effects of changes in both land use and rainfall patterns have to be considered to avoid a risk of overestimating (or underestimating) the relative impact of climate change or variability on landslide activity.