There are many different wavelengths available for usage among today's SAR sensors. The SAR sensors of different satellite systems include a broad range of wavelength configurations, designated X ( λ = 1 cm), C (λ = 6 cm), S (λ = 10 cm) and L-band (λ = 24 cm). There are comparative studies made that indicate that longer wavelengths will result in a higher contrast between forest and non-forest such as clear-cuts (Watanabe et al., 2007). This makes L-band satellite SAR systems highly interesting for large-area mapping of forest changes as it is at present the longest wavelengths available from satellite SAR systems. On the other hand, today’s L-band satellite SAR images do not have as high spatial resolution as for instance TerraSAR-X (X-band) or various optical instruments, but further developments are being made in this field to improve the resolution of L-band satelliteimages (Börner et al., 2007; Rosenqvist et al., 2007). There have been several airborne and spaceborne missions conducted with L-band SAR systems (e.g. AIRSAR, E- SAR, Seasat, SIR-A/B/C, JERS-1, ALOSPALSAR), which have indicated the potential usefulness for forest applications (Rignot et al., 1997; Salas et al., 2002). Therefore, this study suggests that imagesacquired from the ALOSPALSAR L-band SAR sensor are suitable for detection of changes in the boreal forest landscape.
and L-band backscatter is utilized. This time however, due to the very large areal extent, the classification procedure needs to be fully automated. This will be achieved by means of the implementation of multitemporal data. Multi-temporal metrics contain additional information such as the backscatter variability and allow for threshold based classification strategies. First results based on C-band were characterized a classification accuracy above 95% . Detection of forest cover changes, i.e. monitoring the dynamics of forest cover, requires images that not only present certain sensitivity of the observable to forest cover but are also characterized by a limited effect of environmental conditions on the observable. In this respect L-band data are most appealing. In  it was shown that clear-cut detection in European and Siberian forests is possible with HH-backscatter acquired by the JERS-1 SAR sensor. The JERS winter coherence was instead proved to be effective to detect clear-cuts under frozen conditions in Siberia . The polarimetric feature of ALOSPALSAR enables an improvement with respect to JERS because of the strong sensitivity of the cross-polarized backscatter to forest density . Clear-cut detection in Swedishforest with a simple thresholding algorithm was found to perform well, with a detection accuracy of about 90% .
to levels typical of relatively undisturbed or remnant formations. However, maps of forests are different stages of regeneration are needed to facilitate restoration planning, including prevention of further re-clearing. Focusing on the Tara Downs subregion of the BBB and on forests with brigalow (Acacia harpophylla) as a component, this research establishes a method for differentiating and mapping early, intermediate and remnant growth stages from Japan Aerospace Exploration Agency (JAXA) Advanced Land Observing Satellite (ALOS) Phased-Array L-band Synthetic Aperture Radar (PALSAR) Fine Beam Dual (FBD) L-band HH- and HV-polarisation backscatter and Landsat-derived Foliage Projective Cover (FPC). Using inventory data collected from 74 plots, located in the Tara Downs subregion, forests were assigned to one of three regrowth stages based on their height and cover relative to that of undisturbed stands. The image data were then segmented into objects with each assigned to a growth stage by comparing the distributions of L-band HV and HH polarisation backscatter and FPC to that of reference distributions using a z-test. Comparison with independent assessments of growth stage, based on time-series analysis of aerial photography and SPOT images, established an overall accuracy of >70%, with this increasing to 90% when intermediate regrowth was excluded and only early-stage regrowth and remnant classes were considered. The proposed method can be adapted to respond to amendments to user-definitions of growth stage and, as regional mosaics of ALOSPALSAR and Landsat FPC are available for Queensland, has application across the state.
GAMMA Remote Sensing, Worbstrasse 225, CH-3073 Gümligen, Switzerland
The study focuses on investigation and evaluation of wind- thrown forestmappingusingsatellite remotely sensed data from three synthetic aperture radar (SAR) sensors. The study is carried out at Remningstorp, a test site in the south of Sweden dominated by coniferous forest, where trees were manual felled to simulate wind-thrown forest. The satellite data consisted of time series of HH polarized SAR imagesacquired by the Advanced Land Observing Satellite (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR), Radarsat-2 (C-band) and TerraSAR-X (X- band). The results from visual interpretation of SAR imagesacquired before and after the simulated wind-throw together with corresponding ratio images show that ALOSPALSAR HH polarized intensity images are not able to detect wind- thrown forest, probably due to too coarse spatial resolution. In contrast, the wind-thrown forest is clearly visible in the Radarsat-2 and TerraSAR-X HH polarized images, implying that it may be possible to develop a new application using these SAR data for mapping of wind-thrown forests.
Throughout its lifetime (2006–2011), the Phased Array L-type Synthetic Aperture Radar (PALSAR) instrument onboard the Advanced Land Observing Satellite (ALOS) acquired multiple images in several operating modes according to a predefined observation scenario . Given the repeatedly acknowledged sensitivity of L-band data to forest variables in particular in the cross-polarized backscatter [2–4] and under unfrozen conditions [5,6], the image acquisition of ALOSPALSAR were tailored to provide repeated dual-polarized (Horizontal–Horizontal, HH, and Horizontal–Vertical, HV) data in the Fine Beam Dual (FBD) mode during the summer and fall of each year. In addition, HH-polarized images were acquired during the winter season in Fine Beam Single (FBS) mode. During each spring and late fall, a single dataset was acquired in the polarimetric (PLR) mode to obtain a full scattering matrix. These modes acquiredimages with a resolution of approximately 20–30 m and were operated along ascending orbits, i.e., at nighttime. Along descending orbits during daytime, PALSAR operated in the Wide Beam (WB) mode with a spatial resolution of approximately 70 m, to allow sharing of resources with two optical instruments . The acquisition strategy was refined towards the end of 2006 by changing the look angle of the Fine Beam mode from 41.5° to 34.3° to reduce range ambiguities . For the PLR mode, images were acquired with a look angle of 21.5°: since 2009, images were also acquired at 23.1°. In the remainder of this paper, we will refer to a specific acquisition configuration in terms of mode and integer of the look angle (e.g., FBD34 stands for Fine Beam Dual mode with a look angle of 34.3°).
L-band SAR data from the ALOS-2 PALSAR-2 sensor used in this study were acquired during the period 2014-09-18 to 2015-10-06. The operating sensor frequency is 1.270 GHz, which corresponds to a wavelength of 23.6 cm. SAR images from the Fine Beam Dual (FBD) polarization mode and the Quad-polarimetric mode were available. The FBD polarization dataset consisted of 20 pairs of HH- and HV- polarized images (off-nadir angle between 28.2° and 36.2°). The Quad-polarimetric dataset consisted of 2 quadruplets acquired at HH-, HV-, VH- and VV-polarization (off-nadir angle between 25.0° and 34.9°). In particular here we concentrate on the HH and HV images given the strong similarity between cross-polarized data and the poorer sensitivity of VV-polarized backscatter data in forests . Images were available from both ascending and descending orbit passes. The data were provided in SLC format, from which images of the radar backscattered intensity were
conditions. They obtained a cover-state map for large regions with an accuracy of 78–91 % for open water, nonflooded for- est, and flooded forest cover-state classes. Lower accuracies were reported for aquatic macrophytes and for flooded wood- land. Hostache et al. (2009) proposed a SAR image analysis method for spatiotemporal characterization of flood events that includes extracting flood extent limit, estimating wa- ter level, and constraining water level estimates using hy- draulic coherence concepts. The method was applied to an ENVISAT SAR image during the January 2003 flood of the Alzette river. They showed that SAR imagery offers the pos- sibilities to obtain distributed remote-sensing-derived water levels over a large area with sufficient accuracy for calibra- tion of a hydraulic model. Matgen et al. (2011) proposed a hybrid approach to automatically extract flood extent from SAR images by estimating the statistical distribution of open water backscatter values from SAR images of floods, radio- metric thresholding to extract the core of the water bodies, and region growing to extract all water bodies. They pro- posed a change detection procedure using pre- or post-flood SAR reference images to remove over-detection of inundated areas. The methods were evaluated through the 2007 flood of the Severn river (using ENVISAT SAR images) and the 1997 flood of the Red river (using RADARSAT-1 images). Their study showed that the automated method that includes a change detection procedure yields the same performance as optimized manual approaches.
Unmanned Aerial Vehicles (UASs) have become an interest for a variety of jobs and entertainment in the past 10 years especially for current technology limits UAS use to an assistive tool for the inspector to perform a bridge inspection faster, safer, and without traffic closure . İn case of using the Synthetic Aperture Radar (SAR) data, the deflection of the bridge has been monitored . Over years the application of SAR satellite data has been widely performed, ranging from monitoring ground deformation in small and big areas to monitoring single infrastructures, such as building, dam, and bridge  – . The Previous works show that the measurement of bridge deflection has been conducted and completely discussed to find the solving. However, it was not developed a complete data usage by integrating processing UAV data, satellite data, and measurement data in the field. Therefore, this research aims to maximize results which are obtained for deflection measurements that occur on bridges, and to determine the effect the construction of the bridge had on the land.
It can be seen that the increases in the accuracies achieved in some of the datasets by the addition of cer- tain variables are not large. RF produces a measure of the variable importance by analyzing the deterioration of the predictive ability of the model when each predictor vari- able is replaced in turn by random noise (Vincenzi et al. 2011). In general, the texture measures and radar data have low importance scores. The class-specific contribu- tions of different variables to the models are shown in Figure 3. Due to their negligible influence, the texture measures (optical and radar) have been omitted. In all three study areas, all models strongly relied on distinct spectral bands and band ratios. The influence of the ancil- lary data is variable between classes and study sites. The RF models were applied across the entire study areas to obtain vegetation cover for the whole regions (see Fig. 4), while the upland subsets in these study areas are shown in Figure 5. These maps were created using the (vii) data- set, without the inclusion of the soils and elevation ancil- lary data. A 3 9 3 pixel majority filter was applied to the thematic outputs to improve the homogeneity of the final product. As can be seen from Figure 4, the dominant veg- etation cover in all areas is grasslands, and this is relatable to most areas in Ireland. There is very little forest cover on the Dingle Peninsula, while both the Galtee and Comeragh study areas have considerably larger forest areas, especially along the lower slopes of the upland areas. These areas usually represent lands that are mar- ginal for agriculture and since the 1950s, large extents have been afforested, supported through various govern- ment and EU incentive programmes. Concentrating on the upland subsets in Figure 5, the true value of upland areas in terms of habitat diversity is apparent. Mount Brandon (Fig. 5A) has extensive areas of wet heath, semi- improved (dry-humid acid) grasslands, blanket bog and dry siliceous heath. Large areas of montane heath are observed, especially along the western edge of the area making it quite distinctive when compared to the Galtee and Comeragh Mountains. From Figure 5(B), the domi- nant classes for the Galtee Mountains are dry-humid acid grassland along the north-west of the area, dry siliceous heath and blanket bog. Wet heath occurs less frequently, compared to the Mount Brandon area, though there are increased areas of wet grassland. Similar to the Galtees, the dominant classes in the Comeragh Mountains area (Fig. 5C) are blanket bog, dry siliceous heath and dry- humid acid grassland. Small areas of wet heath are scat- tered throughout the area and areas of dense bracken are prevalent along the eastern edges of the upland area.
Table 1 shows the forest AGB, tree density, mean tree height, and mean DBH for the 29 forests, as estimated by the Bitterlich method. The forest with the smallest AGB, and also the smallest density, was site #5 (2.2 Mg ha 1 , 266 trees ha 1 ). A forest with such small AGB was generally characterized by very sparse black or white spruce trees with most trees being less than 10 m tall. In contrast, the largest forest AGB was 116.2 Mg ha 1 at site #16. Here the forest was composed of white spruce trees with the highest mean tree height (16.6 m) of all 29 forests (excluding #23 for which no tree height measurement was made); the tree density was relatively small (814 trees ha 1 ). Although site #17 had the densest forest (11,263 trees ha 1 ), the forest AGB was not very large (81.7 Mg ha 1 ) because the mean tree height was low (5.9 m). Generally, for- ests with low biomass were young and appeared to have experienced recent forest fire activity.
In the MLR, multiple scattering mechanisms and other parameters of SAR decomposition as the phase and orientation angles in one resolution cell can be used as prediction variables instead of only using the backscattering coefficient. However, MLR can have some limitations, namely, in terms of parameter interpretation and over fitting (many prediction variables). In the selected MLR model (Equation (13)) it was observed the importance of polarimetric attributes not commonly used for AGB estimation as the off-diagonal terms of the coherency and covariance matrices. The hypothesis that forested areas show reflection symmetry can be denied by the increasing value of the T23_imag and T12_realB terms, since according to Lee and Pottier  the terms off-diagonal are null if they satisfy the condition of reflection symmetry (monostatic case). This observation motivated us to apply unusual terms as angular responses from decompositions as well as the off-diagonal element of the [T] and [C] as prediction variables, despite the fact that their physical relationship with forest targets is not fully understood.
iii The recti ﬁed SAR image is obtained by resampling the slave SAR images to the master SAR image geometry using the improved transformation relationship. The offset estimation is conducted again to remove some residual offsets to obtain the accurately recti ﬁed SAR images.
As for the recti ﬁed SAR images, the pixel misalignments caused by topographic relief and incidence angle difference are effectively eliminated, which can be employed for offset estimation. Amplitude cross ‐ correlation method was used to estimate the offsets; at each pixel location, a varying window size of 64 to 128 pixels rather than a ﬁxed window was employed to avoid the discontinuous offsets to some extent (Zhao et al., 2013). At each location, the offset estimation with the highest signal ‐to‐noise ratio was retained as the ﬁnal results. As the topographic relief and incidence angle difference effects have been rectiﬁed before, the 2 ‐D offsets can be directly transferred into the 2‐D deformation of the landslide.
Abstract: The Dabus Wetland complex in the highlands of Ethiopia is within the headwaters of the Nile Basin and is home to significant ecological communities and rare or endangered species. Its many interrelated wetland types undergo seasonal and longer-term changes due to weather and climate variations as well as anthropogenic land use such as grazing and burning. Mapping and monitoring of these wetlands has not been previously undertaken due primarily to their relative isolation and lack of resources. This study investigated the potential of remote sensing based classification for mapping the primary vegetation groups in the Dabus Wetlands using a combination of dry and wet season data, including optical (Landsat spectral bands and derived vegetation and wetness indices), radar (ALOSPALSAR L-band backscatter), and elevation (SRTM derived DEM and other terrain metrics) as inputs to the non-parametric Random Forest (RF) classifier. Eight wetland types and three terrestrial/upland classes were mapped using field samples of observed plant community composition and structure groupings as reference information. Various tests to compare results using different RF input parameters and data types were conducted. A combination of multispectral optical, radar and topographic variables provided the best overall classification accuracy, 94.4% and 92.9% for the dry and wet season, respectively. Spectral and topographic data (radar data excluded) performed nearly as well, while accuracies using only radar and topographic data were 82–89%. Relatively homogeneous classes such as Papyrus Swamps, Forested Wetland, and Wet Meadow yielded the highest accuracies while spatially complex classes such as Emergent Marsh were more difficult to accurately classify. The methods and results presented in this paper can serve as a basis for development of long-term mapping and monitoring of these and other non-forested wetlands in Ethiopia and other similar environmental settings.
The objective of this research was to estimate and map Above Ground Biomass (AGB) in plantation and Natural forestusingAlosPalsar data. To achieve this, correlation analysis was used to assess the relation of AGB and other stand pa- rameter measured from the field with Palsar backscatter extracted from L-band HH and HV polarization. After that, a linear regression model was established with the chosen inputs from previous correlation analysis to estimate AGB. The AGB in the natural forest could be estimated and mapped accurately using the linear regression of L-band HH and HV. The plantation forest had a weaker cor- relation in HV but strong in HH backscatter. The strong correlation of HV of 0.753 developed the biomass estimation equation y = 0.0116x − 12.904. y de- notes the biomass value while x is the HV value. The equation was used for bio- mass maps generation for the Kericho and Aberdares study areas.
5.3. Forest biomass information in SAR data
This study shows that no single channel of SAR data provides enough information for biomass retrieval, so multiple channels from multiple polarizations, bands and operation modes, and temporal acquisitions were needed. The height of the scattering phase center at C-band from SRTM DEM –LVIS surface elevation was an important variable in this study. Polarimetric SAR interferometry (Pol-InSAR) data has been used to estimate forest height ( Cloude & Papathanassiou, 2003 ) and then were subsequently converted to forest biomass through forest height – biomass relation ( Caicoya et al., 2010; Mette et al., 2004 ). The DESDynI mission will provide temporal L-band InSAR data. The height of scattering phase center at L band derived from these data if surface DEM is available, and the temporal coherence data may play the similar rule in the biomass estimation. The preference of polarizations of SAR data is not very clear in this study probably because of the ﬂat terrain of the study area. When terrain slope exists, the effects of terrain on co-
content is responsive to a range of stresses on vegetation because of
its direct role in the photosynthetic processes of light harvesting and initiation of electron transport ( Zarco-Tejada et al., 2000 ). The higher values of water content (Cw) observed at the polluted site may be linked to the adaptation process of plants to close the stomata under stress conditions as strategy to reduce transpiration, which in turns reduce photosynthetic rate linked to the lower chlorophyll and thus total tree metabolism ( Larcher, 2003; Zweifel et al., 2009 ). Other foliar properties related to water, those expressed on mass basis (% LWC and % LDMC) also differed and is due to the fact that as these parameters are not normalised by the leaf area, these differences can be explained by the high species diversity of the sample sites where leaves vary greatly in morphology, anatomy and physiology in response to their growing conditions ( Tedersoo et al., 2010 ). Of these leaf variables, it is chlorophyll content that lends itself to be measured from space using a hyperspectral sensor, and since it is this that showed dif- ferences between the polluted and unpolluted sites, this suggests that by measuring this biochemical in vegetation compartments, detection of petroleum contamination across vast expanse of tropical forests is indeed possible.
west, south-east from the center. The dbh for every tree with a dbh > 3cm, and the height, crown length and width of 8 157
trees in each plot were measured. These sites represent a range of forest structures and biomass levels. The locations of 158
This is the final accepted version of the above article and has been deposited according to guidelines: http://www.sherpa.ac.uk/romeo/issn/0269-7491/. The final published version (Arellano, P., Tansey, K., Balzter, H. and Boyd, D.S. (2015) Detecting the effects of hydrocarbon pollution in the Amazon forestusing hyperspectral satelliteimages.