IMPROVING THE SPATIAL RESOLUTION OF SYNTHETIC APERTURE RADIOMETER IMAGERY USING AUXILIARY INFORMATION: APPLICATION TO THE SMOS MISSION A. Camps1,2, M. Vall·llossera1, M. Piles1, F. Torres1, I. Corbella1, N. Duffo1 1
Remote Sensing Laboratory, Universitat Politècnica de Catalunya, Campus Nord, D3; 08034 Barcelona, Spain Tel: +34 934054153; Fax: +34 934017232; E-mail:
[email protected] and SMOS-Barcelona Expert Center (SMOS-BEC) 2 Institut d’Estudis Espacials de Catalunya, IEEC-CRAE/UPC
1. INTRODUCTION Soil moisture can be measured using microwave radiometers. However, the resolutions than can be achieved from satellite-borne L-band radiometers using state-of-the art technology either synthetic aperture (SMOS: 30-60 km) or real aperture (SMAP: 40 km) are much coarser than the fine scales required for regional studies (1-10 km) [1,2]. Brute force solutions (increasing the size of the array or the antenna aperture) to improve the spatial resolution to meet this requirement does not seem a feasible goal in a mid-term future. Therefore, different approaches have been explored to disaggregate lowresolution passive microwave remote sensing data up to the higher resolution required. Cardot et al. [3] proposed the use of temporal interpolation to couple high and low spatial resolution images of mixed pixels. Crosson et al. [4] developed a neural network scheme to downscale low-resolution satellite microwave remote sensing using a coupled hydrologic/radiative transfer model as input for training the network. Kim and Barros [5] showed empirically the connections between the spatial and temporal variability of soil moisture and the behavior of auxiliary data such as topography, soil texture, vegetation water content and rainfall. Based on this idea, they developed an algorithm to downscale a coarse resolution soil moisture pixel using linear time-varying combinations of the spatial distributions of ancillary data [6]. 2. SOIL MOISTURE MISSIONS 2.1. SMAP: NASA’s Soil Moisture Active and Passive Mission The HYDROS mission was selected on July 2002 as an Earth System Science Pathfinder (ESSP) missions by NASA. Its payload consisted of a radar measuring VV, HH, and HV backscattering coefficients operating at 1.26 GHz (Hpolarization) and 1.29 GHz (V-polarization) and a radiometer measuring the brightness temperatures at H- and Vpolarizations, and U the Third Stokes parameter at 1.41 GHz. HYDROS was designed to provide the first global view of the Earth's changing soil moisture and surface freeze/thaw condition [7]. Although the mission was cancelled on December 2005, a number of studies were performed to merge data from different sensors, and particularly to combine the complementary HYDROS radiometric and radar data. For example, Narayan and Lakshmi [8] presented a simple approach to disaggregate the coarse resolution soil moisture estimates from radiometric data using the higher resolution radar backscatter and vegetation water content measurements. Zhan et al. [9] proposed the use of a Bayesian method to optimally combine 36km radiometer brightness temperatures and 3-km radar backscatter cross-section observations. The SMAP mission recently approved for development in the timeframe 2010-2013 uses the same instrument concept and builds on the NASA's ESSP Hydros mission. 2.2. SMOS: ESA’s Soil Moisture and Ocean Salinity Mission The SMOS mission was selected in May 1999 and will be launched in October 2008. SMOS’ single payload mission is MIRAS (Microwave Imaging Radiometer by Aperture Synthesis). MIRAS will be the first-ever space-borne 2D interferometric radiometer. It operates at 1.4 GHz and will provide multiangular (0q up to ~60q incidence angle) brightness temperature measurements with a temporal resolution of ~3 days. However, its spatial resolution will be on the order of 3060 km, which is not well-matched with the observed scale of variability of soil moisture, and hence not particularly useful for small scale applications. Specific disaggregation techniques for SMOS have been explored to address the limited spatial resolution: Merlin et al. [10] presented a disaggregation algorithm that combines 40-km resolution SMOS brightness temperatures with 1-km resolution auxiliary multispectral data and surface variables involved in a land-surface-atmosphere model. This is intended as a first approach, since it assumed that SMOS observations have all the same pixel size and orientation, and that the noise is the same for all pixels. It also relies on the assumption of a high correlation between the radiometric soil temperature inverted from the thermal infrared and the microwave soil moisture. Pellenq et al. [11] presented a methodology to interpret and assimilate the SMOS surface soil moisture data into a geophysical modeling framework
coupling a Soil Vegetation Atmosphere Transfer model (SVAT) with a hydrological model able to redistribute spatially the soil water content as a function of the topographic and surface properties. Other approaches have also been analyzed, such as deconvolution techniques of SMOS-like imagery using improved Wiener and Constrained Least Squares (CLS) filters [12] or wavelet filters [13] that may include different levels of brightness temperature information such as a constant brightness temperature sea and land, a brightness temperature model for the sea and constant land, or even the exact brightness temperature model that was used as input of the simulator [14]. Despite the improvements of these methods, even in the case of using the exact brightness temperature model used as input (~30% reduction in the rms error over land, ~28-38% reduction in the rms error over the sea, and ~25-41% reduction of the coastline width), the results are still far for the 1-10 km goal. 2. INCLUDING AUXILIARY INFORMATION TO IMPROVE THE SPATIAL RESOLUTION OF SYNTHETIC APERTURE RADIOMETER PRODUCTS These approaches try to enhance the resolution of the retrieved brightness temperature images themselves or the derived products (soil moisture) by operating in the image domain. In addition, these methods must provide a high resolution image compatible with the low-resolution one derived from instruments’ observables, and an average soil moisture compatible with the soil moisture derived from the low resolution imagery. This condition is achieved in an iterative way by comparing the retrieved brightness temperature and soil moisture maps. In this work we will present a simple and very efficient technique to include the auxiliary information by adding it directly at visibility level (Fourier domain). High resolution brightness temperature maps computed from the auxiliary data (soil temperature, texture, roughness, vegetation fraction cover and vegetation opacity, topography, rainfall maps, …) already computed in other approaches are used to ingest the high resolution features of the image, and include polarization and incidence angle effects, as described for example in [10]. By construction, the new high-resolution brightness temperature images obtained are consistent with the instrument observables, and contain the high-frequency information of the auxiliary data. 3. RESULTS The results of the application of this technique to SMOS-like imagery derived with the SMOS End-to-end Performance Simulator ( [14] will be presented and discussed, as well as the way to extend this technique to real aperture radiometers. 4. REFERENCES [1] D. Entekhabi and P. S. Eagleson, “Land surface hydrology parameterization for atmospheric general circulation models including subgrid scale spatial variability,” J. Clim., vol. 2, pp. 816–831, 1989. [2] W. T. Crow et al., “Potential for downscaling soil moisture maps derived from spaceborne imaging radar data,” J. Geophys. Res., vol. 105, pp. 2203–2212, Jan. 2000. [3] H. Cardot et al., “Varying-Time Random Effects Models for Longitudinal Data: Spatial Disaggregation and Temporal Interpolation of Remote Sensing Data”, Journal of Applied Statistics, vol. 30, pp. 1185-1999. [4] W. L. Crosson et al., "Disaggregation of microwave remote sensing data for estimating near-surface soil moisture using a neural network" Preprint vol. of 16th Conf. on Hydrology (January 13-17, Orlando, FL), Amer. Meteor.Soc., Boston, MA, 85-90, 2002. [5] G. Kim and A. P. Barros, “Space–time characterization of soil moisture from passive microwave remotely sensed imagery and ancillary data,” Remote Sens. Environ., vol. 81, pp. 393–403, 2002. [6] H. Gwangseob, and A. P. Barros, “Downscaling of remotely sensed soil moisture with a modified fractal interpolation method using contraction mapping and ancillary data”, Remote Sens. Environ., vol. 83, pp. 400–413, 2002. [7] D. Entekhabi et al., “The Hydrosphere State (HYDROS) Satellite mission: an Earth system pathfinder for global mapping of soil moisture and land freeze/thaw”, IEEE Trans. on Geosci. and Rem. Sensing, vol. 42, pp. 2184 –2195, 2004. [8] U. Narayan and V. Lakshmi, “A simple method for spatial Disaggregation of radiometer derived soil moisture using higher resolution radar observations”, Proceedings of the IGARSS 2005 (Seoul, Korea), vol.1, pp. 387-391. [9] X. Zhan et al., “A method for retrieving high resolution surface soil moisture from Hydros L-band radiometer and radar observations”, IEEE Trans. Geosci. Remote Sensing, vol. 44,No 6, pp.1534 – 1544, June 2006. [10] O. Merlin et al., “A Combined Modeling and Multipectral/Multiresolution Remote Sensing approach for Disaggregation of Surface Soil Moisture: Application to SMOS Configuration”, IEEE Trans. on Geosci. and Rem. Sensing, vol. 43, No. 9, pp. 2036 – 2050, Sept 2005. [11] J. Pellenq, et al. “Scaling and Assimilation of SMOS Data for Hydrology”, in Proceedings of the IGARSS 2003 (Toulouse, France), vol. 5, pp. 3064 – 3066 [12] M. Piles et al., “Deconvolution algorithms in image reconstruction for aperture synthesis radiometers”, in Proceedings of the IGARSS 2007 (Barcelona, Spain), pp. 1460-1463, July 23-27, 2007. [13] M. Piles, et al., “Spatial Resolution Enhancement Of SMOS Data: A Combined Fourier Wiener Approach”, submitted to IGARSS 2008, July 7-11, 2008, Boston, Massachussetts, USA [15] A. Camps et al., “The SMOS End-to-end Performance Simulator: Description and Scientific Applications”, in Proceedings of the IGARSS 2003, pp. 13 - 15 vol. 1, 21-25 July 2003