Land surface parameter monitoring with ERS scatterometer data over the Sahel: A comparison between agro-pastoral and pastoral areas
Résumé
This paper deals with ERS wind scatterometer (WSC) data for land surface monitoring over the Sahel on a regional scale. Past studies have shown that WSC is particularly well suited for monitoring Sahelian pastoral regions consisting of large homogeneous herbaceous steppes (Jarlan et al., 2002a and Jarlan et al., 2002b). On the other hand, Sahelian agro-pastoral areas are more heterogeneous in terms of WSC spatial cell resolution owing to different land use caused by anthropic pressure. The objective of this paper is to assess the suitability of WSC data in such a heterogeneous landscape. To this end, WSC data for an agro-pastoral area are compared with results presented in Jarlan et al., 2002b which focused on 4 pastoral sites, located in Mali. The agro-pastoral area concerned is in the Fakara region (Niger), for which numerous in situ measurements are available for the 1994–2000 period over a 500 km2 area. WSC temporal signatures acquired at a 45° incidence angle, σ0(45°), show the typical behaviour observed over the Sahelian areas. σ0(45°) values are higher than those observed by Jarlan et al. (2002b) over Northern Sahelian sites, especially during the dry season, and exhibit a lower yearly amplitude variability. The effect of land use heterogeneity on the radar signal is investigated by comparing WSC profiles acquired at a 21° incidence angle, σ0(21°), with ERS SAR data during the year 1995. The observations are interpreted by combining a radiative transfer model and a vegetation growth model in a similar approach to the one developed by Frison et al., 1998 and Jarlan et al., 2002a, Jarlan et al., 2002b and Jarlan et al., 2003. The results indicate that at an incidence angle of 21°, soil contribution is largely predominant, representing 90% of the radar signal, thus explaining the similar behaviour of radar signatures observed despite different land uses. At a 45° incidence angle, even though annual vegetation and bare soil are the two main factors influencing the temporal evolution of the radar signal, the bare soil contribution is still the most significant. It is shown that over the present study site, characterised by low annual vegetation fraction cover and high σ0(45°) values during the dry season, vegetation makes no significant contribution to the yearly amplitude of the radar signal, σadr0(45°). The latter can be explained by the bare soil contribution alone. Consequently, limited soil moisture due to small field capacity accounts for the small σadr0(45°) variability observed. An inversion method to retrieve the volumetric soil moisture content, Hv, using σ0(45°) data is then developed. It appears that vegetation has a negligible influence over estimated Hv. As a result, the soil moisture content can be retrieved from WSC data alone, by representing the present scene as being composed of bare soil only. These results differ from those obtained over pastoral Sahelian sites, where both vegetation and soil moisture play significant roles in radar response. Moreover, the comparison between WSC σ0(45°) retrieved soil moisture with existing soil moisture derived from passive microwave radiometer SSM/I measurements shows a fairly good overall match.