ESA       
   workshops       
envmail@esa.int

 

Monitoring vineyards in the Frascati wine production area by L-band SAR: first results

Domenico Solimini(2) and Alessandro Burini(1)

(1) GEO-K, Via del Politecnico, 1, I-00133 Roma, Italy
(2) Tor Vergata University, Via del Politecnico, 1, I-00133 Roma, Italy

Abstract

Providing services to farming estates, like monitoring conditions, phenological state and potential yield of vineyards by radar would be a significant step towards operational applications of Earth observations from space.

The radar sensitivity to tree biomass stems from the different volumes of woody matter which affect the wave-plant-soil interaction mechanisms. On the other hand, developing crops like rice, wheat, maize, sunflower, colza, or alfalfa, considerably change the number, dimensions and shapes of the scattering elements, thus modifying their radar return. The situation is different when the plants have a stable woody structure which is only slightly modified by developing fruits. In this case, monitoring the fruit biomass by radar observations is made difficult by the small variations of backscattering with respect to a strong, almost saturated (at least at C-band) background. Measurements on vineyards are even more difficult, given the high number of poles and metallic wires supporting the runners. Moreover, the biophysical quantities of main interest to Precision Viticulture (PV), like phenolic and sugar ripeness levels, are not directly measurable by radar, although the plant vigor index has been correlated with the Leaf Area Index (LAI), which, in turn, can provide yield estimations.

A preliminary experiment was carried out in cooperation with ESA/ESRIN to investigate on the sensitivity of ERS SAR measurements to developing grapes. To this end, 18 vineyards were selected, also with the support of QuickBird images, in the Frascati area at about 15 km from the center of Rome, Italy. Sets of SAR images acquired in 1999 and 2001 by ERS-2 were analyzed to produce time series of the backscattering coefficient averaged over the parcels. A number of uncultivated natural soils randomly distributed in the neighborough of each parcel provided the backscattering time series references to reduce the effects both of the terrain characteristics and of climatology, mainly resulting in soil moisture variations. This procedure also reduced the interference by the weed infesting the vine rows. The method allowed fine peculiar trends of the relative backscattering coefficient of vineyards to be identified in the time series. Checks against farming records allowed identification of the effect of the cultivation practices, like ploughing and arrowing, that changed the roughness of the soil and reduced the weeds. Also on the basis of this experience, an experiment was supported and organized by ESA, within the BACCHUS-DOC project, in fall 2005. The DLR E-SAR airborne SAR overflew the studied Frascati area on 5 October, when grapes were still present, and on 25 October, after vintage. The study area was imaged at C-band at co- and cross-polarization and at L-band in a fully polarimetric mode. The adverse meteorological conditions affecting the first flight prevented sound conclusions on the backscattering effect of grapes from being drawn from the processed data.

The ascending ALOS PALSAR imaged the Frascati study area in May, June and September 2007. The first acquisition is fully polarimetric at about 25 degrees incidence angle, whereas the other two are dual-pol at about 40 degrees. Given the different ground resolution, the dual-pol images have been re-sampled from 20 m pixel to 30 m and the geometric distortion due to the different incidence angle has been corrected by means of several GCP, finally obtaining a satisfactory co-registration over the vineyard area. The initial analysis has been carried out on the vh-hh backscattering ratio to reduce the effect of the changing soil conditions. The backscattering of different parcels, consisting of bare soil, forest and grassland have been assumed as a reference for the vineyards' radar signatures. The time trend of the cross-polar to co-polar relative backscattering ratio of the vineyards depends on the type of cultivar, i.e., tent, NS (almost perpendicular to the incidence plane) rows and EW (almost parallel to the incidence plane) rows and its variation with time appears to be correlated with the vine's phenological development stage and agricultural phases. Finally, the statistical relation between the radar signature and the commonly used vigor index derived from optical multispectral images has been studied to assess the potential of L-band backscattering in monitoring some of the biophysical parameters utilized in the wine production practices.

 

 

  Higher level                 Last modified: 07.05.06