Field and Basin Scale Analyses of Asar Imagery for Soil Moisture Estimation in the Campidano Plain, Sardinia

There is a strong interest in assessing the potential of space-based monitoring and mapping of state variables that are critical to hydrological and agricultural applications. Our study consists of the acquisition of ASAR imagery in single and alternating polarization modes and at different incidence angles over the Campidano plain in south-central Sardinia (Italy), the island's most important agricultural region. In tandem to image acquisition, ground data (surface soil moisture and roughness) is being collected. The project will investigate soil moisture dynamics and detection at both the large scale (multitemporal analysis for the Campidano region) and the small scale (retrieval algorithms tested on individual field plots). This paper will focus mainly on the field scale work, where ground data and imagery from the period Jun-Nov 2005 have been used to assess a semi-empirical model for surface soil moisture and roughness inversion from the radar signal. 1. PROJECT DESCRIPTION AND OBJECTIVES OF THE OVERALL STUDY There has been steady progress in our ability to estimate or map land surface state variables from spaceborne sensors. In the visible, infrared, and (passive) microwave bands, data products that provide information about vegetation, soil moisture, roughness, temperature, and other parameters are now routinely available. Active microwave SAR (synthetic aperture radar) instruments have recognized advantages over other sensors (e.g., finer spatial resolution, day/night sensing, cloud penetration), but progress in using SAR for land surface applications has been slower due to the interacting effects of near-surface soil moisture, surface roughness, and vegetation on the radar backscattering signal. Important advances are expected with the advent of new SAR sensors (e.g., ALOS-PALSAR and RADARSAT-2) that will provide polarimetric imagery, from which it should be possible to isolate the effects of two or more parameters on the backscattering response. In advance of these new sensors (just launched in the case of the Japanese ALOS and soon-to-be-launched in the case of the Canadian RADARSAT-2), the ASAR (advanced SAR) sensor aboard the European ENVISAT satellite introduced significant improvements over the previous generation of sensors, such as alternating polarization and a range of possible incidence angles and spatial resolutions. In this project (ESA-AO 537), our main interest is in the detection of near-surface soil moisture, a critical state variable for hydrological and agricultural applications and in water resources management. We have a dataset of ASAR imagery from 2003 onward over the Campidano plain in south-central Sardinia, and have begun collecting field data in 2005 at an agricultural research station situated in the Campidano [1]. The main objectives of the work are: 1. To generalize an existing semi-empirical model for retrieving surface roughness and soil moisture from SAR data in order that the model can exploit the alternating polarization and incidence angle capabilities of ASAR. The new algorithm is being tested at the field scale with ground truth from 2005 (presented here) and 2006 (forthcoming work). 2. To perform a multitemporal analysis on the multiyear ASAR dataset, complemented with image and map data from other sources, at the regional (basin) scale that comprises a large part of the Campidano plain. The aim here is to detect large-scale spatio-temporal changes and patterns in soil wetness and their correlation with other environmental parameters. 3. To implement and test retrieval algorithms based on polarimetric SAR imagery. Field scale data to support this task is currently being collected (spring-summer 2007 field campaign) and will continue into 2008. 4. The study of hydrological simulation models and data assimilation algorithms that can make best use of observation data from various sources, including (and especially) remotely sensed measurements. In a hydrological model, periodic observations of near-surface soil moisture can be used to update the model's boundary conditions; these in turn drive surface and subsurface _____________________________________________________ Proc. ‘Envisat Symposium 2007’, Montreux, Switzerland 23–27 April 2007 (ESA SP-636, July 2007) partitioning of water and energy fluxes. Combining simulation and observation data in this way is expected to lead to improved predictions and a better characterization of uncertainty. In this presentation we will highlight the work undertaken so far towards objective 1. Objective 2 constitutes the research to be undertaken by a recent PhD student while objective 3 awaits the availability of ALOS and RADARSAT-2, already planned for our Campidano study region. These latter 2 phases of the project will be briefly described. Objective 4 ties in with ongoing hydrological modeling work based on a detailed, process-based numerical model of coupled surface–subsurface flow and is described elsewhere (e.g., [2, 3]). 2. DESCRIPTION OF THE STUDY AREA The island of Sardinia faces acute water management problems due to its susceptibility to long-term droughts and the possibility that this may be exacerbated by climate change. The Campidano plain (Fig. 1) is Sardinia’s most important agricultural region. Intensive campaigns for acquisition of ground data are being undertaken at several small fields that are part of CRAS’ agricultural research station, situated between the towns of Ussana and Donorì (Fig. 2). For the 2007 field campaign we have added several other fields, vegetated and bare, from a large farm adjacent to the research station. Figure 1. Location of Campidano plain (basin scale) study area (red circle). 3. FIELD DATA Field data collected on two bare soil plots (of dimension 2 and 4 ha) during campaigns in 2005 and 2006 includes surface roughness and soil moisture (Fig. 3) as well as laboratory analyses to determine additional soil properties (grain size distribution, porosity, retention curves). A meteorological station at the site provides rainfall data (Fig. 4). Additional meteorological data for the entire Campidano is available from several regional meteorological stations. Other data (topography, land cover, irrigation, geology, etc) is available at both local and regional scales. Figure 2. Location of the CRAS agricultural research station (field scale study area) near Ussana. Colors indicate different crops that are under cultivation.