small branches are still present as a randomly oriented scattering volume. At coarse resolution, the speckle noise is also more extended compared with higher resolution SARimages, limiting the possibilities to detect wind-thrownforests. On the other hand, longer wavelengths are sensitive to larger structures like stems and large branches that would improve the possibilities to detect wind-thrown forest. In particular, the significantly improved spatial resolution of satelliteSARimages is expected to improve the possibilities to detect wind-thrown forest. Thus, to further explore the use of radar remote sensing to detect wind-thrown forest it is of interest to analyze images from the new satelliteSAR systems, i.e. the Advanced Land Observing Satellite (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR) because of the longer wavelength and Radarsat-2 and TerraSAR-X because of the high spatial resolution.
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 satelliteSAR systems highly interesting for large-area mapping of forest changes as it is at present the longest wavelengths available from satelliteSAR systems. On the other hand, today’s L-band satelliteSARimages 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, ALOS PALSAR), which have indicated the potential usefulness for forest applications (Rignot et al., 1997; Salas et al., 2002). Therefore, this study suggests that images acquired from the ALOS PALSAR L-band SAR sensor are suitable for detection of changes in the boreal forest landscape.
ALS has established its position as a reliable data source in forest inventories in Finland. However, at the moment the repetition frequency of ALS in nation-wide surveys is rather poor – maybe 10 years or so. Thus, SARsatellite image could have potential in forestry, namely in updating the inventories or change detection. In publication VI, the Quad and Dual polarization ALOS PALSAR images were used in mapping of a forested area in Finland. The test area was located in the Nuuksio national park in Southern Finland, which mainly consists of mixed forests in their natural conditions. Therefore, the test area can be characterized as a challenging one with respect to the use of SARimages. Reference information was obtained from Optech ALS data with the point density of 3/m 2 . ALS data was used to create a Canopy Height Model (CHM) by subtracting the ground elevation values from the canopy elevation values. Then, CHM was used to estimate the Above Ground Volume (AGV) for 111 manually digitised forest stands, which were located at as flat areas as possible in order to cancel the effect of topography on the SAR backscattering values. Even though AGV is not directly related to the stem volume, it seems that it is possible to estimate stem volume relatively accurately (Hyyppä et al., 2008). In total, 2 Dual polarization and 3 Quad polarization ALOS PALSAR images were used. In addition to the amplitude information of SAR backscattering, polarimetric descriptors (alpha-entropy-anisotropy decomposition) were calculated. In the case of Dual polarimetric images, the interferometric coherence between the two images was calculated also. Then, SAR descriptors were used to estimate the LIDAR based AGV using multiple linear regression modelling. The results showed relatively good correlation between the AGV and SAR signal. In the cases of Dual and Quad polarization images the R 2 values of 0.53 and 0.72 for the AGV estimation were achieved respectively. The scatterplot of estimated and LIDAR based AGVs in the case of Quad polarization SAR descriptors is presented in Figure 10.
Lake outlines are produced in the requested projection as shape files from the SAR imagery at 10 m spatial resolution. Water surfaces are very well visible in SARimages because of the generally low backscattering intensity at all microwave wavelengths (Strozzi et al., 2000). We preferred to map the outlines manually because of the challenges and uncertain- ties using automatic methods as a consequence of the speckle of the radar images and of the wind and waves conditions of the lakes, which can increase locally the roughness of the wa- ter surface and thus the backscattering intensity. Supplemen- tary geotiff’s of the geocoded SARimages with a layover and shadow mask are also produced. The resulting informa- tion is thus in a form that can easily be integrated in the end- users’ geographic information system (GIS) and used within the available infrastructure. Consistency tests based, e.g. on multiple independent results from different time periods or sensors, or validation with reference information of indepen- dent origin, are essential to better characterize the error and to estimate the reliability of the product.
The inaccuracy of the absolute geo-localization of the optical satellite data in the geo-referencing process arises mainly from inaccurate measurements of the satellite attitude and thermally-affected mounting angles between the optical sensor and the attitude measurement unit. This insufficient pointing knowledge leads to local geometric distortions of orthorectified images caused by the height variations of the Earth’s surface. To achieve higher geometric accuracy of the optical data, ground control information is needed to adjust the parameters of the physical sensor model. We are following the approach described in  to estimate the unknown parameters of the sensor model from GCPs by iterative least squares adjustment. In order to get a reliable set of GCP, different levels of point filtering and blunder detection are included in the processing chain. In contrast to , where the GCPs are generated from an optical image, we are using the matching points generated by our network.
The study period covers the winter of 2017-18, which was marked by particularly high avalanche activity recorded in the French Alps. Microwave backscattering over snow surfaces is complex because it combines several phenomena including reflection on the snow surface, scattering within the snowpack (which depends on its layers properties) and reflection at the snow-soil boundary. To detect avalanche debris, change detec- tion methods are typically used to isolate avalanche debris-like features based on the backscatter contrast between avalanche debris and the surrounding undis- turbed snowpack . Debris detection is based on major changes in the backscatter coefficients due to changes in snow properties following the avalanche event (height, density, roughness), Figure 1 shows an example of an RGB composition map using 3 Sentinel-1 images at VH polarization. The large avalanche event near "Les Houches" can be seen in green.
Rice mapping products were derived from Sentinel-1A and Landsat-8 OLI multi-temporal imagery over Northern Italy at the early stages of the 2015 growing season. A rule-based algorithm was applied to synthetic statistical metrics (TSDs-Temporal Spectra Descriptors) computed from temporal datasets of optical spectral indices and SAR backscattering coefficient. Temporal series are available up to the tillering/full canopy cover stage which is identified as the optimum timing for delivering in-season information on rice area (i.e. mid July). The approach relies on a-priori knowledge on crop dynamics to adapt time horizons for TSD computation and thresholds to local conditions. Output products consist of maps of rice cultivated areas, rice seeding techniques (dry and flooded rice) and flooding practices. Validation showed rice mapping overall accuracy to be 87.8% with commission and omission errors of 3.5% and 24.7%, respectively. Mapping of rice seeding technique showed good agreement with farmer declarations aggregated at the municipality scale (dry rice r 2 = 0.71 and flooded rice r 2 = 0.91). Finally, flood maps have an overall accuracy above 70%. Geo- products on rice areas and flooding occurrence are relevant information for water management at regional scale especially during summer in presence of multiple crops and water shortage.
retrieved from multi-temporal statistics. Another study used X-band SAR with eleven COSMO-SkyMed HH and HV images for land cover mapping over an agricultural area in Southern Australia . The temporal information improved the classification results, with an overall accuracy of ca. 82% for 10 classes . Five land use/land cover types (forests, urban infrastructure, surface water and marsh wetland) were mapped from multi-temporal polarimetric RADARSAT-2 imagery in North-eastern Ontario, Canada . Wetlands showed a seasonal increase in HH and HV backscatter intensity due to the growth of emergent vegetation over the summer but other classes showed little temporal variation in backscatter. Multi-temporal RADARSAT-2 polarimetric SAR data were used to discriminate high- density residential areas, low-density residential areas, industrial and commercial areas, construction sites, parks, golf courses, forests, pasture, water, and two types of agricultural crops using an object- based support vector machine and a rule-based approach (κ = 0.91) . In the Brazilian Pantanal, multi- temporal L-band ALOS/PALSAR and C-band RADARSAT-2 data gave an accuracy of 81% for the land cover types of forest, savanna, grasslands/agriculture, aquatic vegetation and open water .
Synthetic Aperture Radar is a radar technology that is used from satellite or airplane. It produces high resolution images of earth‘s surface by using special signal processing techniques. Synthetic aperture radar has important role in gathering information about earth‘s surface because it can operate under all kinds of weather condition (whether it is cloudy, hazy or dark). Polarimetric SAR (PolSAR) image classification is arguably one of the most important applications in remote sensing. Classification is the process of assigning a set of given data elements to a given set of labels or classes such that various parameter of assigning the data element to a class is optimized. Radar polarimetry is a technique for classification of land use features. Various research work have reported the use of polarimetric data to map earth terrain types and land covers (, , , , ). Image classification can be mainly divided into supervised and unsupervised classification techniques. An unsupervised classification technique, classifies the image automatically by finding the clusters based on certain criterion. On the other hand in supervised classification technique the location and the identity of some cover type and terrain type , for example urban, forest, and water are known prior to us . The data is collected by a field work, maps,
In order to provide a wall-to-wall height map, GLAS-derived heights (using the best ANN model), spectral and textural indices extracted from optical images and topographic information were used to build height models using MLR and RF regressions. The best result was obtained from an RF model with an RMSE of 5.5 m and adjusted R 2 of 0.59. The mixed use of spectral and textural features generally showed better explanation of canopy height than either spec- tral or textural which is consistent with Maillard ( 2006 ) and Su, Sheng, Du, Ch and Liu ( 2015 ) findings for image classification. The resulting model was used to provide a height map at a spatial resolution of 30 m for the entire study area. Relatively good com- patibility was observed between generated map and field measurements (RMSE = 4.3 m and R 2 = 0.50). It is worth noting that only 32 plots located in a small part of the study site were used for validation which corresponds to 3.2 ha of the mapped area (15,000 ha). More sample representatives of the entire study area are required to confirm reliability of the outcomes. The resulting map showed higher accuracy rather Figure 11. Fitted semivariograms of Lorey ’s height residuals.
The study presented here was commissioned by a cooperation of the Integrated Fire Management Project and the Sustainable Forerst Management Project, both of GTZ (Deutsche Gesellschaft für technische Zusammenarbeit mbH) and operating in Samarinda, East Kalimantan during the exceptional fire event that struck this Indonesian province in late 1997 and early 1998. It aimed at producing a map of fire damage at a scale of 1:200,000 for almost the entire province discriminating the most severely affected areas from the less damaged. This is of great importance for regional planning endeavours in the fire’s aftermath. Prior studies mapping the fire damage (Liew et al. 1998, Fuller and Fulk 2000) were hampered by availability of optical data due to cloud and haze coverage, did no provide the required spatial resolution or were conducted before the fires came to an end. Since no optical data at the desired spatial resolution were available, it was decided to make use of the cloud-penetrating capabilities of satellite-born imaging radar. Results of a pilot study have already been published elsewhere (Siegert and Ruecker 1999, Siegert and Ruecker 2000, Siegert and Hoffmann 2000). In this article we shall describe results of an empirical study about the ability to discern degrees of vegetation damage by fire in tropical lowland ecosystems with the help of C-band imaging radar. We also present the fire damage map compiled from a multitemporal mosaic of 46 ERS-2 SARimages.
Abstract— Because of the all-weather and all-time data acquisition capability, high resolution space borne synthetic aperture radar (SAR) plays an important role in remote sensing applications like earth mapping. However, the visual interpretation of SARimages is usually difficult, especially for urban areas. This paper shows a method for visual interpreting SARimages by means of optical and SARimages simulated from digital elevation models (DEM), which are derived from LiDAR data. The simulated images are automatically geocoded and enable a direct comparison with the real SAR image. An application for the simulation concept is presented for the city center of Munich where the comparison to the TerraSAR-X data shows good similarity. The simulated optical image can be used for direct and quick identification of objects in the corresponding SAR image. Additionally, simulated SAR image can separate multiple reflections mixed in the real SAR image, thus enabling easier interpretation of an urban scene.
The validation results show some false positives in our mapping result. For example, among the 19 low density honeysuckle sites identified from the aerial photo, 4 of them show as medium/high density honeysuckle cells on the image. One cause of a false positives can be the presence of vegetation such as grasses, sedges or vines that are still green in the late fall when the satellite image was acquired. If an image cell covers both bare tree branches and other green vegetation, it is possible for the computer to mistakenly recognize the other green vegetation as honeysuckle’s green leaves and classify such cells as medium/high density honeysuckle areas. In the future, we need to improve the classification process. For example, the texture of the other vegetation may look different from the honeysuckle bush texture on the image. We may use this texture difference in addition to the greenness of the cell to distinguish the regular tree from the honeysuckle bush.
Surveying the condition of individual trees and the plants growing densely in detail is important for the environmental conservation and the maintenance of vegetation and forest area. The authors have verified a possibility of vegetation investigation by high-resolution satellite image as IKONOS and Quickbird. Two-dimensional distribution of small-scale vegetation in Tokyo can be known from these satelliteimages. Digital Roof Model (DRM) to display the height of building’s roof is developed from the image observed by the IKONOS satellite. This model enables us to give the position of tree canopy shape three dimensionally in a case of using the IKONOS image. A tree canopy map of the forest in a city park in the outline was studied from DRM used the IKONOS image three dimensionally. Capability of depiction of canopy pattern influenced by the difference of performance of two satelliteimages in the overcrowded tree region where the tree crown parts overlapped was investigated. The array configuration of the trees was extracted accurately for City Wildlife Park
task. Consequently, the use of effective unsupervised change detection methods is fundamental to many applications for which ground truth is not available. The unsupervised approach is attractive for change detection tasks due to its self organizing, generalizable and fault tolerant characteristics [ 6 ]. In some cases, change detection can be viewed as a particular case of the multi-temporal image classification problem. Post and pre classification comparisons are the two main approaches in this view. In the first case, the images of two dates are independently classified and co-registered, and an algorithm is used to identify those pixels whose predicted labels change between dates. In the second case, a single classification is performed on the combined image data sets for the two dates [ 7 ].
image, non-calibrated bands with zero value were eliminated through preprocessing steps. Each image was geometrically corrected in order to define images under a common coordinate system and correct pixel relative location distortions. In this study, Tasseled Cap Transformation (TCT) coefficients were produced for SPOT 5 data by using ―Gram-Schmidth method‖. Images of brightness, greenness and wetness were obtained by using these coefficients. TCT images that were produced via 2003 dated SPOT4 and 2007 dated SPOT5 satellite data were used in ―Change Vector Analysis‖ (CVA) to detect emerging land cover changes in Terkos Basin. For the change vector analysis, three TCT difference images, one change vector magnitude image, three vector direction images; and one final landscape dynamic image depicting the most changed landscapes were produced. Threshold value was determined for the change detection determination by using statistical calculations and the analyst‘s expertize. In this phase as a secondary method, ―Principle Component Analysis (PCA) based change detection method‖ was applied. Results of the above stated methods were compared to examine the performance of the methods for change detection. In PCA based change detection method, 2003 and 2007 dated satelliteimages were stacked hence a new eight-band image was formed. PCA was applied on this new image and a hybrid classification was obtained by using the first three components with info of great importance. In this hybrid method, firstly unsupervised classification was applied and two categories were prepared as ―an image change is present‖ and ―an image change is not present‖. Masking was applied to select ―an image change is present‖ category and supervised classification method was followed. Through both methods, it has been shown that detection of change is possible for heterogeneous natural lands. The accuracy assessment results showed that better change detection results can be obtained by ―PCA based change detection‖ method. Semivariogram and spatial profile analyses were applied alongside the test area near Lake Terkos. Accuracy of change detection results was supported by obtained results. At the end of change detection analyses it was found that changes in study area are rather limited because a certain part of the region is under protection by international criteria. It was detected that present changes have an orientation from agricultural lands to settlements and/or open lands, a decrease in wetlands, an increase in the area of roads and associated land, a transformation from forestry lands to sparsely vegetated lands and forestation of some open lands.
In this paper, the analytical scattering model of SSA1 is employed for the sea surface wind speed estimating from Sentinel-1A V V -polarized SARimages. The accuracy of sea spectrum has undeniable eﬀects on the accuracy of the analytical scattering coeﬃcients. Among the various spectra proposed in the literature, we choose Elfouhaily spectrum as the most common directional spectrum and Hwang spectrum as the newest omnidirectional spectrum to evaluate their accuracy for our sea scattering problem. In addition to Elfouhaily spreading function, we use the McDaniel spreading function to see the eﬀect of directionality in the accuracy of the results. The comparisons of the Elfouhaily omnidirectional spectrum with the Hwang spectrum show more power of Hwang spectrum at high winds, and comparisons of the Elfouhaily spreading function with the McDaniel function show greater directionality of McDaniel function at high wavenumbers. Then, the SSA1 simulations are performed with the E-E, E-M, H-E, and H-M composite spectra. The comparison of the backscattering NRCSs of sea surface simulated by the SSA1 model using four spectra with the empirical function CMOD6 shows that the various spectra have diﬀerent performances in diﬀerent situations depending on wind speed, wind direction, and incident angle. From the viewpoint of the omnidirectional spectrum, Hwang spectrum has better performance in most cases. From the viewpoint of angular function, in some conditions, the Elfouhaily function has better performance than the McDaniel function, and in some other conditions, better performance belongs to the McDaniel function. Among the composite spectra, the H-E spectrum is more accurate than others in more situations. At very high wind speeds ( > 20 m/s), diﬀerences between the SSA1 model with various spectra and CMOD6 are high in most situations and lead to a large error in wind speed estimation.
2. For disasters of this magnitude, multi-lateral agencies who are conducting their own damage assessment, researches and studies in the affected state using earth observation data may consider sharing their raw data with the national geospatial information agency (NGIA) of the said state to be used as additional input in the production of base maps, since these are the primary planning tool for rehabilitation, recovery and multi-hazard mapping activities, as well as in the formulation of updated land use plans by the local government units. Going even farther, in providing humanitarian and relief assistance, satellite imageries or other earth observation data should likewise be included.
Synthetic aperture radar (SAR) oil spill images are corrupted by speckle noise due to random interference of electromagnetic waves. The speckle degrades the quality of the oil spill images and makes interpretations, analysis and classifications of SARimages harder. Therefore, some speckle mitigation is necessary prior to the processing of SAR oil spill images. In this paper, some basic speckle reduction filters like Kuan, Lee, Frost, antistrophic diffusion and SRAD filters are used. A New method is proposed for despeckling of SAR oil spill images which combines the frost filter with relaxed median filter. The proposed method gives better results when compared to other methods in terms of statistical parameters like PSNR, MSE, energy and entropy value. The approach can also reduce the computing time compared with other approaches.
Cloud motion wind map reflects the strength and direction of wind very well. It can be generalized as the important reference for the long-term weather forecasts in meteor- ology. It is also playing an important role on mastery atmospheric circulation and indicating disastrous weather. This article has completely described the whole process of calculating cloud motion wind by the use of the the- ory of analytic geometry and MATLAB computing soft- ware. The method is simple, fast, and easy to operate and promote. Using relevant meteorological data, we present intuitive and basically consistent with the actual cloud motion wind map. However, the window size of the clouds this algorithm selected is limited to 16 × 16 pixels. And the search scope is limited to 64 × 64 pixels. But in practical application, fixed window size and search scope do not completely extract the feature information of im- ages. Missing information reduce the quality of the pixel block matching. Therefore, it has limitations. In order to improve the quality of the pixel block matching, we can consider abolishing the limited window size and search scope. Design the effective method which can adaptively determine the window size and the search range. Because errors exist in both the cloud map matching process and the cloud motion wind calculating process, how to reduce these errors and make the accuracy of the results to a fur- ther improvement and enhancement is our work that we need to continue research.