Mapping and monitoring the wide spread of Sarcoproterium Spinosum across the Mediterranean Basin: Challenges and Opportunities

Dwarf-shrubs patterns of Sarcopoterium Spinosum (SS) cover wide Mediterranean landscapes. Following Global warming SS is suspected to further spread into areas of disturbed ecosystems due mainly to fires, and into fields abandoned due to decreasing water resources. On one hand, SS patterns may decrease runoff and soil erosion and allow use of these woody shrubs for heating and cooking, but on the other hand, they decrease primary productivity, and slow down natural succession processes and ecological recovery. Despite their wide current extents, their potential future  expansion, and their ecological implications, mapping SS patterns and monitoring their spatio-temporal change received relatively  limited  attention by the environmental remote sensing community. Part of the explanation concerns difficulties in detecting  these plants during the winter due to their spectral similarity to other green plants  and their spectral resemblance  to bare soil and other dry plants  in their vicinity during the summer.

At the Landsat TM resolution (30 meters) there were developed phenological algorithms which allow estimation of dwarf-shrubs coverage based on their slower drying rates compared with herbaceous growth. This method is applicable  in areas  characterized by short rainy season and long dry and hot summer. However, such  mid-resolution phenological techniques are highly affected by the selection of appropriate dates according to rainfall distribution at the end of the winter, and has limitations concerning the density and the size of  individual dwarf-shrubs’ patches. Global and wide regional land cover mappings utilizing diverse sensors in the VIS/NIR/SWIR and SAR spectral regions processed by  different spectral/temporal and spatial techniques disregarded  dwarf-shrubs in general and Sarcoproterium Spinosum in particular.  Yet, there is a possibility that some areas of this cover category are implicitly  included in the broad “shrubs” classification.

High spatial resolution hyperspectral imagery was found to allow detection of few dwarf-shrub species in general and SS in particular. However, hyperspectral  mapping at the required spatial resolution is still expensive and does not allow frequent mapping of wide areas. The current growing availability of high resolution RGB and NIR imagery  (e.g., WorldView) may be instrumental for detecting dwarf-shrubs and SS. The use of spectral indices such as NDVI and  red-edge, of color transformations (such as HIS) and of texture techniques had shown  potential  for serving this purpose. Recent implementation of Deep learning methods  on RGB 30 cm. resolution orthophotographs showed good potential  for discriminating SS patterns at three densities.

During my presentation I will review the different techniques as implemented along a semi-arid to arid gradient at the South-Eastern corner of the Mediterranean Sea. Their results and  limitations will be discussed together with methodological improvements required for achieving better regional spatio-temporal coverage of the SS phenomenon and by that contribute to better understanding their wide regional ecological  implications in the context of Climate Change and Desertification.