Coherence Tomography for Boreal Forest: Comparison with HUTSCAT Scatterometer Measurements Jaan Praks and Martti Hallikainen Department of Radio Science and Engineering, Helsinki University of Technology FIN-02015, Espoo, Finland Email: jaan.praks@tkk.ﬁ Florian Kugler and Konstantinos P. Papathanassiou Microwaves and Radar Institute, German Aerospace Center: Deutsches Zentrum für Luft- und Raumfahrt (DLR) D-82234 Wessling, Germany Abstract In this paper we report on X- and L-band polarization coherence tomography in boreal forest structure mapping. We take advantage of HUTSCAT ranging scatterometer measurements and compare E-SAR derived vertical structure proﬁles with measured vertical scattering proﬁles. HUTSCAT measurements provide vertical scattering proﬁle of the canopy and also provide supplemental forest height and ground phase estimates, allowing calculation of coherency proﬁles with alternative initial values. We present single baseline tomography for X-band and multibaseline tomography for L-band calculated with various initial values and compare the results with actual measurements of scattering proﬁles. 1 Introduction SAR polarimetric interferometry (POLInSAR) has shown its potential in forest remote sensing. Polarimetric inter- ferometric coherence measurement allows estimation of canopy parameters. The methods are based on model- ing the coherence inside the forest canopy. The Random Volume over Ground (RVoG) model [1] inversion [2] for fully polarimetric coherence data has been used in several studies to successfully estimate forest height [1], [3], [4]. Our results from the FINSAR campaign showed that even X-band single polarization interferometric coherence can be successfully used to invert the RVoG model for for- est height [5]. Recently Cloude [6] has taken a step for- ward and suggested a method to estimate the shape of the coherence function inside canopy in more detail. This is done by approximating the vertical structure function with Legendre polynomial series. The method is called Polar- ization Coherence Tomography (PCT). The approximation accuracy is dependent on the available amount of measure- ments, in this case interferometric baselines. Cloude [6] has shown that even for one baseline the general shape of single layered volume can be calculated. The FIN- SAR campaign material provides a good basis for coher- ence tomography study as together with E-SAR polarimet- ric interferometric measurements also vertical scattering proﬁles of the forest were measured with the HUTSCAT scatterometer [8]. The scatterometer measurement pro- vides supplemental initial values (forest height and ground phase) for tomography and gives an opportunity to com- pare vertical structure proﬁles with actual scattering pro- ﬁles. 2 Material The material in our study was collected during Fthe IN- SAR campaign [5], carried out in autumn 2003 in Finland. The main instruments of the campaign were E-SAR [7] and HUTSCAT ranging scatterometer [8]. On 29 Septem- ber the E-SAR sensor collected from 3 km altitude four L- band (1.3 GHz) repeat pass fully polarimetric interferomet- ric images and an X-band (9.6 GHz) single-pass single-pol (VV) interferometric image pair. The images were pro- cessed to a 2 × 2 m (range and azimuth) resolution grid. The helicopter-borne HUTSCAT scatterometer measure- ment was carried out two days later, providing a vertical backscattering proﬁle at C-band (5.4 GHz) and X-band (9.8 GHz) along the 36 km ﬂight track. The incidence angle was vertical and the helicopter location was mea- sured by a GPS receiver. Most of the HUTSCAT measure- ments were concentrated on a 2 × 2 km area covering the E-SAR near and mid range. The HUTSCAT range resolu- tion is 0.65 m, antenna beam width is 3.8 0 and the system along-track sampling distance is 1.25 m when helicopter moves with the ideal speed 25 m/s. The HUTSCAT and E-SAR slant range images were co-registered according to the pixel coordinates. The test site in southern Finland (N 60 0 11 ′ ,E 24 0 29 ′ ) comprises forest, ﬁelds and lakes. The