Applications of SAR Polarimetry on Soil Moisture (1)
2015-01-27 12:10 - 2015-01-27 13:10
Chairs: Ridha Touzi, CCRS / Marco Lavalle, JPL/Caltec
-
12:10 Estimation of Soil Moisture under Vegetation Cover at Multiple Frequencies
Jagdhuber, Thomas (1); Hajnsek, Irena (2); Papathanassiou, Konstantinos P. (1) 1: German Aerospace Center, Germany; 2: ETH Zurich, Switzerland
Show abstract
Hydrological processes within the vadose zone, like wetting or drying of the upper soil layer, control the water availability for plants. Hence, the optimization of yield in precision farming benefits decisively by the knowledge of soil moisture distributions and their temporal and spatial change independent of varying vegetation cover on fields. For a catchment scale assessment of soil and plant characteristics in agricultural areas, an airborne SAR campaign, called CROPEX, was conducted by the German Aerospace Center (DLR), the ETH Zurich and the University of Munich (LMU) around the town of Wallerfing in Lower Bavaria, Germany. The campaign was conducted from April to August 2014 and covered the entire vegetation growth cycle for a variety of agricultural crops. The periodic acquisitions were carried out by DLR’s F-SAR sensor. The high-resolution sensor recorded fully polarimertric SAR data in different frequencies: X- (3cm), C- (5cm) and L-band (23cm). Concurrently to the airborne recordings, in situ measurements for soil and vegetation conditions were performed on selected agricultural test fields with varying crop and soil types. In order to extract soil moisture under agricultural vegetation from polarimetric SAR data, the different, occurring scattering mechanisms within the resolution cell have to be separated. Several polarimetric decomposition techniques were established in the last years to remove the perturbing vegetation component and to obtain the ground component for soil moisture inversion. In [1] to [3] soil moisture under vegetation cover was estimated from X-, C- and L-band data, acquired with different polarimetric configurations (full-pol./hybrid-pol./dual-pol.) and under differing vegetation cover. Due to the variation of polarimetric information content (from full-pol. to dual-pol.) and the differing vegetation scattering scenarios, a combined and joint analysis of all three frequencies, recorded simultaneously and on the same test site was not possible. Therefore a disentangling of the frequency-dependent vegetation effects and the impact of the limiting polarimetric observation space could not be accomplished for a rigorous performance analysis of soil moisture estimation at different frequencies. Therefore the moisture retrieval algorithms are adapted for a multi-frequency analysis. Afterwards, they are applied on fully-polarimetric X- C- and L-band data of the CROPEX 2014 campaign to investigate in detail the capability of each single frequency to provide estimates of soil moisture under a growing vegetation cover along the agricultural season. In a final step, the extracted SAR-based soil moisture values are validated with corresponding ground measurements for a first quality assessment and a discussion on potentials and limitations regarding space-borne SAR missions in X-, C- or L-band, like TanDEM-X, RADARSAT-2 or ALOS-2. [1] Jagdhuber, T., Hajnsek, I., Papathanassiou, K.P., Polarimetric Soil Moisture Retrieval Using an Iterative Generalized Hybrid Decomposition Technique, Proc. of IGARSS, July 13-18, Quebec, Canada, 2014, p.1-4. [2] Jagdhuber, T., Hajnsek, I., Caputo, M., Papathanassiou, K.P., Dual-Polarimetry for Soil Moisture Inversion at X-Band, Proc. of EUSAR, June 3-5, Berlin, Germany, 2014, p.1-4. [3] Ponnurangam, G.G., Jagdhuber, T., Hajnsek, I., Rao, Y.S.: Soil Moisture Inversion using Hybrid Polarimetric RISAT-1 Data. Proc. of EUSAR, June 3-5, Berlin, Germany, 2014.
-
12:30 Soil Moisture Change Under Vegetation: PolInSAR modeling, simulation and observations
Lavalle, Marco; Pinto, Naiara JPL/Caltech, United States of America
Show abstract
In this contribution we present our latest developments in modeling the PolInSAR coherence of vegetated land surfaces. Our interest this time centers on understanding how changes of mean water content in the soil covered by vegetation affect the magnitude and phase of the coherence at different polarimetric channels. Coherence variations induced by soil moisture may be seen as either an undesired source of temporal decorrelation or as a useful signal for inferring soil moisture through interferometric measurements. The soil permittivity may change over time as a consequence of precipitation, temperature transition, evotranspiration and water uptake by vegetation. The permittivity controls the amount of backscatter from the soil and the penetration and propagation of microwaves in the soil. For bare soil, experiments have shown that an increment of moisture causes the coherence magnitude to decrease and the coherence phase to increase [1]–[6]. For vegetated soils, the scenario is more complex because an increment in soil moisture between two acquisitions changes also the ground-to-volume ratio, which affects the propagation in the vegetation layer and, consequently, the PolInSAR coherence. We developed a physical model that relates the PolInSAR coherence to the mean change of vegetated soil moisture. The model uses the structure function of the random-volume-over-ground (RVoG) model and the motion function of the random-motion-over-ground (RMoG) model. In addition, the new model accounts for dielectric changes through two distinct ground-to-volume ratios. The X-Bragg model of the soil coherency matrix [7, 8] and the Hallikanen’s empirical model [9] are employed to link the ground-to-volume ratios to the actual soil moisture change. This procedure allowed us to study how sensitive the coherence is to realistic changes in soil moisture. The model has been first validated using a discrete scattering simulator designed specifically for PolInSAR temporal decorrelation studies. The simulator generates a vertical distribution of scattering elements with second-order scattering contributions from an underlying surface. Structural and dynamic properties are defined as the input to the simulator through time-varying covariance matrices and other physical parameters. We are now in the process to analyze several L- and P-band PolInSAR data collected recently by the NASA/JPL UAVSAR instrument in the context of soil moisture and vegetation campaigns in the United States (uavsar.jpl.nasa.gov and airmoss.jpl.nasa.gov). At the workshop we will discuss the features of the new model and illustrate how the coherence locus - the RVoG/RMoG model line in the complex plane - modifies when dielectric changes occur. We will then describe our discrete simulator used for temporal decorrelation studies, and discuss the impact of soil moisture change on the estimation of tree height. Finally, we plan to present the results of the experiments with L-/P-band UAVSAR data analyzed in conjunction with in situ measurements obtained through the FluxNet sensors network over different sites in the United States. [1] A. K. Gabriel, R. M. Goldstein, and H. A. Zebker, “Mapping small elevation changes over large areas: differential radar interferometry,” Journal of Geophysical Research, vol. 94, no. B7, pp. 9183–9191, Jul.- 10 1989. [2] M. Nolan, D. R. Fatland, and L. Hinzman, “DInSAR measurements of soil moisture,” IEEE Transactions on Geoscience and Remote Sensing, vol. 41, no. 12, pp. 2802–2813, Dec. 2003. [3] M.Nolan and D. R. Fatland,“Penetration depth as a DInSAR observable and proxy for soil moisture,” IEEE Transactions on Geoscience and Remote Sensing, vol. 41, no. 3, pp. 532–537, Mar. 2003. [4] T. Zhang, Q. Zeng, Y. Li, and Y. Xiang, “Study on relation between insar coherence and soil moisture,” Proc. ISPRS Congr., vol. 37, pp. 131–134, Jun. 2008. [5] B. Barrett, P. Whelan, and N. Dwyer, “The Use of C- and L-Band Repeat-Pass Interferometric SAR Coherence for Soil Moisture Change Detection in Vegetated Areas,” The Open Remote Sensing Journal, vol. 5, pp. 37–53, Jun. 2012. [6] F. De Zan, A. Parizzi, P. Prats-Iraola, and P. Lopez-Dekker, “A sar in- terferometric model for soil moisture,” Geoscience and Remote Sensing, IEEE Transactions on, vol. 52, no. 1, pp. 418–425, Jan 2014. [7] Hajnsek, I, Pottier, E., Cloude, S.R., “Inversion of surface parameters from polarimetric SAR,” Geoscience and Remote Sensing, IEEE Trans. on , vol.41, no.4, pp.727,744, April 2003 [8] Jagdhuber, T., Hajnsek, I., “Model-based inversion of soil parameters under vegetation using ground-to-volume ratios” Synthetic Aperture Radar (EUSAR), 2010 8th European Conference. [9] M. Hallikainen, F. Ulaby, M. Dobson, M. El-Rayes, and L.-K. Wu, “Microwave dielectric behavior of wet soil-part 1: Empirical models and experimental observations,” IEEE Trans. Geosci. Remote Sens., vol. GE-23, no. 1, pp. 25–34, Jan. 1985.
-
12:50 Exploitation of Radarsat-2 dual and quad-pol images and modelled compact polarimetry parameters for surface soil moisture estimation in mountain areas.
Greifeneder, Felix (1); Notarnicola, Claudia (1); Stamenkovic, Jelena (2); Paloscia, Simonetta (3); Dabboor, Mohammed (4); Charbonneau, Francois (5) 1: Institute for Applied Remote Sensing, EURAC Research, Bolzano, Italy; 2: Signal Processing Laboratory, EPFL, Lausanne, Switzerland; 3: Institute for Applied Physics, CNR, Florence, Italy; 4: Science and Technology Branch, Environment Canada, Toronto, Canada; 5: Canadian Centre for Mapping and Earth Observation, Ottawa, Canada
Show abstract
In the last few years there has been an increasing interest towards the estimation of Soil Moisture Content (SMC) at finer scales (<1 km) using active microwave sensors like Synthetic Aperture Radar (SAR) [1]. Especially, when working in mountain areas high-resolution data is necessary to represent the strong spatial variability. Due to the terrain complexity these areas have been considered marginally in research and only pioneer studies can be found in the literature [2]–[4]. As shown in [2], the use of dual-pol (HH-HV) and possibly of quad-pol configurations (HH+ HV/VV) together with terrain and land-use data, as well as normalised vegetation index, is recommended to improve the estimation of SMC in areas with a complex terrain. Future SAR missions such as the Canadian RADARSAT Constellation Mission (RCM) will include Hybrid-polarity SAR architecture. These sensors will enable the use of compact polarimetry (CP) data in wide swath imagery. Compared to current quad-pol modes, a SAR system with hybrid-polarity architecture will be able to offer other significant advantages like increased swath width. This would mean shorter revisit times for a given ground location and therefore the possibility of improved SMC estimation, similar to that of fully polarimetric SAR-imagery. Hybrid-polarity architecture is included in the current Indian RISAT-1 satellite. The goal of this study is to assess the applicability and potential benefit of CP parameters, in combination with field measurements and machine learning, for the estimation of SMC in an alpine environment. The main focus is put on the analysis of the CP data, compared to standard dual-pol SAR data, and the possibility of improved SMC estimation capabilities. In this work we consider four quad-pol, single look complex RADARSAT-2 images (acquired between 21st of June and 1st of September, 2014), 23 simulated CP parameters [5] and in-situ TDR measurements of SMC, which were collected within field campaigns concurrently to the Radarsat-2 acquisitions. As ancillary data, necessary to describe the complex Alpine terrain, we used a high accuracy digital elevation model with the spatial resolution of 2.5 m, the MODIS NDVI product, as well as a land-use map. The test area is in Mazia Valley, which is located in northern Italy, province of South Tyrol. The terrain is very mountainous with an elevation ranging between 920 m a.s.l. and 3738 m a.s.l. As estimation techniques we consider two different kernel methods: Support Vector Regression (SVR), introduced in [2], and Gaussian Process Regression (GPR). The methods are able to model the relationship between several input features and a target variable. In this case, between in-situ SMC measurements, radar backscatter, modelled CP parameters and ancillary data. After a training phase and given the input features, the model can be used to estimate the SMC at unknown points. To assess the added value of the CP parameters, different input feature configurations are tested – e.g. Radarsat-2 in dual-pol, compared to CP and compared to Radarsat-2 in quad-pol. The performance is evaluated based on the comparison between estimated SMC and true SMC of an independent test-set. In a preliminary analysis, it is found that, compared to Radarsat-2 in dual-pol mode, the estimation accuracy can be improved by approximately 10% through adding the CP parameters. Furthermore, the results show that the use of standard quad-pol produces SMC estimates with the highest accuracy. We are able to demonstrate the benefit of CP compared to standard dual-pol SAR imagery, in terms of the estimation of SMC. Especially in Alpine areas, the added parameters can help to better describe the complex relations between radar backscatter, soil moisture, terrain and surface structure.
In May 2013, the Biomass mission concept was selected to become the next in the series of satellites developed to further our understanding of Earth.
REPLAY EARTH EXPLORER 7 USER CONSULTATION MEETING (PART 1)
POLinSAR 2013 Workshop