Soil surface infiltration capacity classification based on the bi-directional reflectance distribution function sampled by aerial photographs. The case of vineyards in a Mediterranean area
Introduction
Hydrological studies conducted on farmed catchments in the Mediterranean area confirmed that intense rainfall events combined with often discontinuous soil cover by crops cause intense overland flow and erosion (Llorens and Gallart, 1992, Wainwright, 1996). Previous studies have demonstrated the dominant influence of soil surface features on overland flows (Casenave and Valentin, 1992, Andrieux et al., 1996) and pesticide transport (Lennartz et al., 1997, Louchart et al., 2001), especially by determining the flow partitioning between infiltration and runoff. Such soil surface features vary strongly in space, especially between fields, and evolve at variable rates in time, being determined by agricultural management practices difficult to foresee and a highly irregular rainfall regime (Léonard and Andrieux, 1998, Andrieux et al., 2001). Therefore spatially distributed hydrological modelling is required to understand and describe flooding events, agricultural pollution and water resource management (De Roo et al., 1992, Moussa et al., 2002). Such spatial modelling, that has to include the temporal dimension, requires quantitative and easily accessible input information on soil surface variables at the scale of one or more catchments.
Remote sensing is often considered as the favourite tool to provide such spatial and temporal input information. For agricultural areas with spatially continuous crops, remote sensing has proven to provide information on various surface feature variables. Reviews like Blanchard et al. (1999) and Van de Griend and Engman (1985) summarise the results obtained over the last 20 years focusing on the use of remote sensing for hydrological modelling. They were mostly reporting the advances in surface roughness, humidity and soil organic matter characterisation.
However, these studies highlighted also the limits of remote sensing techniques for an operational use in hydrological modelling of a vineyard environment. Firstly, we note that the studies generally focus on a single soil surface variable (Arrouays et al., 1996, Cialella et al., 1997, Mathieu et al., 1997, Zhangshi and Lee, 1997), whereas the surface classes used by the hydrological models are defined by a whole set of variables like for example stoniness, roughness, crusting, presence of litter, weed cover. A second limiting aspect is the inadequacy between the spatial and temporal resolutions as provided by the available remote sensing devices and those required by the hydrological models (Blanchard et al., 1999). Thirdly and lastly we observe that little effort has been directed towards sparse crops such as vineyards. Most of the few remote sensing studies applied to vineyards were focusing on the vegetation canopy rather than on the underlying soil surface. They generally exploit the spectral features of the radiometric signal while ignoring the influence of the spatial discontinuity (Wildman, 1979, Minden and Philipson, 1982, Trolier et al., 1989 using Landsat TM; Johnson et al., 1998, Carothers, 2000). The few studies that investigated the soil surface signal over vineyards are based either on hyperspectral data or on microwave data: Hill et al. (1994) used spectral unmixing techniques at a 20-m spatial resolution to estimate soil erosion risk of Mediterranean vineyards from AVIRIS hyperspectral sensor. Company et al., 1994, Company et al., 1995, used airborne SAR and ERS data to describe soil surface roughness over Mediterranean vines. They were severely hampered by the effect of vine rows and their compass orientation on radar backscatter. It follows from these studies that satisfactory results may only be expected by directly assessing the soil surface signal.
Assessing this signal means to assess the surface's bi-directional reflectance. The directional reflectance distribution of natural surfaces is known to be non-lambertian and strong directional dependence is expected for reflectance of rough surfaces. To take this into account, the Bi-directional Reflectance Distribution Function (BRDF) of surface types have to be studied. BRDF models exist and are generally used for two types of applications (Chopping, 1998): (1) reflectance normalisation (Hu et al., 2000) and (2) information extraction from variation of reflectance upon changing pixel observation and illumination geometry (Lacaze et al., 1999, Chopping, 2000, Scarth and Phinn, 2000).
In this study, we will investigate how to detect the combined variables determining the soil infiltration-runoff flow partitioning on the basis of simple radiometric data. For this purpose, we used helicopter aerial photography that provides the required high spatial resolution (a pixel size of 0.25 × 0.25 m) and flexibility to frequently repeat the observations. The prior extraction from this imagery of the illuminated soil fraction is based on the dedicated automatic methods developed by Wassenaar et al., 2001, Wassenaar et al., 2002.
In the following, we will first describe the study area and the different groups of soil surface types with distinct infiltration capacities that are encountered in this environment. After analysing the radiometric behaviour of these groups, we will propose a method to classify the observed soil surface signal into pre-established radiometric classes representing these groups based on simple BRDF features. Analysing the sensitivity of this method and the separability of the radiometric classes completes this. After concluding, it will shortly be shown how the consideration of spatial and temporal information can improve the quality of the results obtained.
Section snippets
Vineyard soil surface infiltration capacity classes in the Roujan catchment
The small Roujan catchment (0.91 km2) is located in the Hérault valley 60 km west of Montpellier, Southern France. This catchment was selected because it represents quite well the conditions encountered in the larger Peyne valley (about 100 km2). The valley of the Peyne river is a good paradigm of the Languedoc-Roussillon coastal vine region. Moussa et al. (2002) demonstrated that the hydrological behaviour of the Roujan catchment is largely determined by agricultural practices through their
Radiometric differences between SSC
Fig. 4 shows the values obtained for the parameters ρ0 and Θ of the red band. Although taken into account, parameter k is not represented here because it proved to be less discriminating. Several important observations can be made on the basis of this figure. The mineral soil SSCs show quite well defined clusters for each class, but intermediate situations cause more or less fuzzy boundaries. This corresponds to the actual gradual transition between soil surface features.
A stony surface for
Conclusion and perspectives
We demonstrated that several classes of soil surface features with a distinct infiltration rate can be distinguished on the basis of simple radiometric information, acquired at a very high spatial resolution (0.25 m) but spectrally as well as directionally limited (two broad reflective bands and one direction). Crusted mineral soil surfaces as well as surfaces more than half covered by litter and/or weed could be unambiguously identified on the basis of this information. The separability of the
Acknowledgements
This study was funded by a Marie Curie research-training grant, provided by the European Commission DGXII. The authors wish to thank the technical and research personnel of the LISAH ENSA.M-INRA-IRD laboratory that greatly helped in acquiring the ground truth information during the flight campaigns.
References (42)
- et al.
A runoff capability classification system based on surface features criteria in semi-arid areas of West Africa
Journal of Hydrology
(1992) Testing a LiSK BRDF model with in situ bidirectional reflectance factor measurements over semiarid grasslands
Remote Sensing of Environment
(2000)- et al.
Surface albedos and angle-corrected NDVI from AVHRR observations of South America
Remote Sensing of Environment
(2000) - et al.
Infiltration characteristics of soil in Mediterranean vineyards in Southern France
Catena
(1998) - et al.
Small basin response in a Mediterranean mountainous abandoned farming area: research design and preliminary results
Catena
(1992) - et al.
Contribution of multi-temporal SPOT data to the mapping of a soil erosion index. The case of the loamy plateaux of northern France
Soil Technology
(1997) - et al.
Determining forest structural attributes using an inverted geometric-optical model in mixed eucalypt forests, Southeast Queensland, Australia
Remote Sensing of Environment
(2000) - et al.
Morphology, genesis and classification of surface crusts in loamy and sandy soils
Geoderma
(1992) - et al.
Partial area hydrology and remote sensing
Journal of Hydrology
(1985) Infiltration, runoff and erosion characteristics of agricultural land in extreme storm events, SE France
Catena
(1996)
Déterminisme du partage infiltration-ruissellement sur parcelles de vigne en climat méditerranéen
Documents du BRGM
Predicting infiltration rates by classifying soil surface features in a Mediterranean wine-growing area
Detection of soil crusting risks related to low soil organic carbon contents by using discriminant analysis on Thematic Mapper data
Geocarto International
Paramétrisation du potentiel de ruissellement des bassins versants au moyen de la télédétection et des systèmes d'information géographiques. Application à des bassins versants du Pays de Caux
Étude et Gestion des Sols
The art of winemaking
Imaging Notes
Predicting soil drainage class using remotely sensed and digital elevation data
Photogrammetric Engineering and Remote Sensing
Inferring soil surface roughness from soil bidirectional reflectance data
International Agrophysics
The quantitative evaluation of false colour photography with application of a red filter
International Journal of Remote Sensing
Potentiel des données radar ERS-1 pour la détection des surfaces contributives au ruissellement dans les vignobles méditerranéens du Roussillon (France)
Cited by (29)
Cover crop management and water conservation in vineyard and olive orchards
2021, Soil and Tillage ResearchManagement of service crops for the provision of ecosystem services in vineyards: A review
2018, Agriculture, Ecosystems and EnvironmentCitation Excerpt :This better infiltration is partly linked to the soil surface properties: service crops increase soil surface roughness, and the root system increases soil macroporosity (Leonard and Andrieux, 1998). As a consequence, soil surface hydraulic conductivity is improved (Wassenaar et al., 2005). During rainfall events, when soil is saturating, hydraulic conductivity of soil surface decreases, leading to surface water runoff.
Use of Landsat series data to analyse the spatial and temporal variations of land degradation in a dispersive soil environment: A case of King Sabata Dalindyebo local municipality in the Eastern Cape Province, South Africa
2017, Physics and Chemistry of the EarthCitation Excerpt :For instance, literature shows that in developing countries, agriculture remains the back-bone of the rural populace; if not for the whole country (Murdoch, 1980). Besides structural, technical and political constraints to improve and adequate food production in rural areas, as highlighted in Murdoch (1980), land degradation, is one of the major challenge that cannot be ignored or left to continue unabated if rural economies are to be improved (Wassenaar et al., 2005). For example, a study by Seutloali et al. (2015) has demonstrated that the economic costs linked with the undesirable impacts of land degradation are often too high, especially for developing economies.
Assessing the impact of soil surface characteristics on vineyard erosion from very high spatial resolution aerial images (CÔte de Beaune, Burgundy, France)
2014, CatenaCitation Excerpt :Remote sensing techniques have successfully been employed for soil and terrain mapping at a global or catchment scale (Irvin et al., 1997; Martin and Timmer, 2006) and are expected to contribute to Digital Soil Mapping as an auxiliary data source (Lagacherie et al., 2007; McBratney et al., 2003) (see Mulder et al., 2011 for a review). Aerial or satellite imagery has also demonstrated their usefulness when studying the relationships between hydrological runoff, sediment erosion and soil surface characteristics (King et al., 2005), evaluating surface roughness (Baghdadi et al., 2002), or assessing intra-plot soil surface variability by estimating soil surface reflectance (Corbane, 2006; Corbane et al., 2008; Touriño Soto, 2005; Wassenaar et al., 2005). Our approach combines sampling and remote sensing techniques at very high spatial resolution (VHSR), i.e. at a five centimetre spatial scale, to highlight fine spatial distribution of soil surface heterogeneities.
Multitemporal analysis of hydrological soil surface characteristics using aerial photos: A case study on a mediterranean vineyard
2012, International Journal of Applied Earth Observation and GeoinformationPredicting the spatio-temporal dynamic of soil surface characteristics after tillage
2011, Soil and Tillage Research