Retrieval Of Surface Physical Parameters With AVHRR And SMMR Over Africa

Abstract

I t is well established that optical remote sensing can bc uscd to retrieve surface vegctation information providcd that the atmosphere is sufficiently clear and cloud free. Passive microwave remote sensing has becn successfully uscd lor extracting information on surface soil moisture provided that information on vegetation characteristics is known. In this study, we merge the two sources of information to characterizc thc soil and vegetation surface / atmosphere intcrfacc. The technique utilizes the information containcd in two microwave frequcncics as well as visible and near-infrared wavebands to cxtract both soil moisture and vegetation characteristics. It rclates the microwave polarization ratios to Vegetation indiccs derived from optical remote sensors through a theoretical radiative transfer model. The method was applied in a rcgional scale context, over a diverse range of biomcs along a 60". African transect for the year 1986. Thc sensors used includcd the Nimbus-7 Scanning Multichannel Microwave Radiometcr (SMMR) and the NOAA Advanced Very High Rcsolution Radiometer (AVHRR). The synergistic use of the two scnsors allowed for the retrieval of soil moisture, vegetation cover. In this paper, we will describe the method and results, together with discussions on the potentials and limitations of this synergistic relationship involving the optical and microwave rcmote sensing. Introduction Optical remote sensing of terrestrial surfaces has been mainly focused on vegetation characterization and monimring(l,2), and is successful provided that the atmosphere and other external factors are corrected for. Microwave remote sensing of earth surfaces, on the other hand, has been mainly focused on soil moisture information extraction(3,4,5). The microwave remote sensing of surface soil moisture is much less affected by the atmosphere(6) at low frequency compared with that of optical remote sensing, but is affected by the presence of vegetation(7,8). Therefore, microwave sensing of vegetated earth surfaces is more difficult since microwave signal contains information about both soil moisture and vegetation. The 37 GHz microwave measurements from satellites have been studied for vegetation characterizations(9), and found LO be correlated with vegetation indices derived from optical remote sensing. For effective soil moisture information extraction, vegetation contribution must be decoupled. However, it is difficult LO decouple these two factors with microwave sensing alone. Optical remote sensing provides information about vegetation, and therefore, can be combined with the microwave remote sensing for decoupling soil moisture and vegetation. It is interesting, therefore, to merge optical and microwave remote sensing to simultaneously retrieve soil moisture and vegetation information of terrestrial surfaces. The objective of this paper is to merge optical and microwave data sources by developing synergistic relations so as to effectively retrieve surface parameters that otherwise are difficult to obtain. 96 Approaches For a randomly distributed vegetation layer over a bare soil substrate, the brightness tempcrature is given by (9,10,11): where Tbpo is brightness temperature (K) of the canopy close to ground at polarization p,T, and T, are soil and vegetation physical temperatures (K) respectively, e? is soil emissivity, a is single scattering albedo, and y is transmissivity of the vegetation layer over the soil, which is related to the optical thickness z at viewing angle of 8 by ~ 0 ) = exp (-~/p) and p = COS (e). The above model applies only to the brightness temperature close to the surfaces. For satellite radiometric measurements, the atmosphere effect has to be taken into account(6,ll). Therefore, the observed brightness temperature from satellites should be where z, is the atmosphere transmittance and T,, is the atmospheric contribution, and the third tcrm can bc neglected at low frequency. Assuming T, = T, and 1-0. = 1, we obtain Lhc microwave polari7ation difference ralio (PR = 2(TbV Tbl{)/(TbV + Tbll)) as: