Effect of surface soil moisture gradients on modelling radar backscattering from bare fields

• Published in: International journal of remote sensing Vol. 18; no. 1; pp. 153 - 170
• Main Authors: Boisvert, J. B., Gwyn, Q. H. J., Chanzy, A., Major, D. J., Brisco, B., Brown, R. J.
• Format: Journal Article
• Language: English
• Published: Abingdon Taylor & Francis Group 01.01.1997; Taylor and Francis; Taylor & Francis
• Subjects: Earth sciences; Earth, ocean, space; Engineering Sciences; Exact sciences and technology; Other; Soils; Surficial geology; Alberta; Canada; Western Canada; models; diffusion; interfaces; time variations; soil moisture; reflectance; penetration; stratification; air; North America; backscattering; dielectric constant; agriculture; irrigation; depth; soils; correlation coefficient; evaporation; radar methods
• ISSN: 0143-1161; 1366-5901
• DOI: 10.1080/014311697219330

In agriculture, soil moisture gradients near the surface are important. The objective of this study was to determine their impact on the estimation of the radar backscatter when using surface diffusion models. Two surface radar backscatter models were evaluated in a bare soil with soil moisture stratification induced by irrigation and evaporation. The models were the semi-empirical model (OM) and the Integral Equation Model (IEM). These models were coupled with a penetration depth model. Three approaches were used to calculate the reflectivity at the air-soil interface required by the models. The first one was based on the incoherent reflectivity contribution of each stratum. The second approach calculated the Fresnel reflectivity based on the mean complex dielectric constant over a fixed depth, and the third one computed the Fresnel reflectivity using the mean complex dielectric over the penetration depth of the incident signal. The estimations from the OM model were highly correlated with the observations for all combinations of incidence angles, frequencies and polarizations. However, the backscatter was systematically underestimated and bias increased with wavelength. The IEM provided the most accurate estimate but was sensitive to radar configuration and roughness. The two Fresnel reflectivity based approach tended to overestimate the backscatter in the drying sequence after intensive irrigation following field work. The incoherent reflectivity on the other end underestimated the backscatter in freshly dry disturbed soil. The sampling depth had to be chosen carefully when estimating the backscatter from the second approach, especially in Ku and L-bands. Resume. Les gradients hydriques pres de la surface jouent un role important en agriculture. L'objectif de cette etude etait de determiner leur impact sur l'estimation du signal radar par un modele de diffusion de surface. Deux modeles furent evalues dans un sol nu ou une stratification hydrique avait ete induite par irrigation et evaporation. Il s'agissait du modele semi-empirique (OM) et du modele de l'equation integrale (IEM). Ces modeles furent couples a un modele de penetration du signal. Trois approches furent retenus pour le calcul de la reflectivite a la surface requise par les modeles. La premiere etait basee sur la reflectivite incoherente qui tient compte de la contribution de chaque strate de sol. La second calculait la reflectivite de Fresnel a l'interface air-sol en utilisant la constante dielectrique complexe moyenne sur une profondeur de penetration fixe; la troisieme calculait la reflectivite de Fresnel a partir de la constante dielectrique moyenne sur la profondeur de penetration du signal. Les estimes du modele OM etaient fortement correles aux observations pour toutes les combinaisons d'angles d'incidence, de frequences et de polarizations etudiees. Cependant, les coefficients de retrodiffusion furent systematiquement sous-estimes et l'ecart augmenta avec la longueur d'onde. L'IEM fournit des resultats precis mais s'est revele sensible a la configuration du radar et a la rugosite. Les deux approches basees sur la reflectivite de Fresnel ont surestime le signal retrodiffuse dans la phase de dessechement qui a suivi une forte irrigation effectuee apres le travail du sol. L'approche basee sur la reflectivite incoherente a par contre sous-estime le signal retrodiffuse dans les sols secs fraichement laboures. La profondeur d'echantillonnage devrait etre choisi esoigneusement quand le coefficient de retrodiffusion est estime par la seconde approche, particulierement en bande Ku et L.