Sigma Naught Research Articles

NISAR S‐SAR Digital Beamforming Architecture and Implementation

AbstractISRO(Indian Space Research Organization) and JPL(Jet Propulsion Laboratory), NASA are jointly developing a state‐of‐the‐art L&S band SAR, planned to be launched through GSLV during 2023‐24. The conventional SAR systems has a limitation of providing higher spatial resolution as well as wider swath coverage simultaneously[6]. The NASA‐ISRO Synthetic Aperture Radar (NISAR) is based on a novel SAR concept, which will be utilized to image wide swath as high resolution of stripmap SAR, also called SweepSAR[1]. NISAR uses Digital Beam Forming on receive along with SweepSAR concept to get the advantage of lesser transmit power, immunity to range ambiguities with narrower receive beams, improved SNR and noise equivalent sigma naught as compared to conventional SAR system and directly radiating arrays with digital beamforming[5].In this paper, digital beamforming (DBF)[2] technique being used in NISAR S‐SAR receive chain is presented. The paper focuses on the DBF architecture, methodology, mathematical equations, practical implementation challenges and strategies used for mitigation of such challenges in order to achieve desired performance. Subsequently, DBF Hardware implementation algorithm and improvement in S‐SAR performance as a result of DBF technique is discussed.

The effects of radiometric terrain flattening on SAR-based forest mapping and classification

ABSTRACT Terrain-induced variations of radar backscatter represent an important limiting factor of many Synthetic Aperture Radar (SAR)-based applications. Radiometric terrain flattening (RTF) is a well-established method that minimizes these variations in SAR imagery. To fully understand the implications of SAR RTF, validation of its impact on the derived products is needed. In this study, we quantified the influence of the RTF on a forest mapping and classification algorithm over Austria, and compared the classification results for the conventional sigma naught and radiometrically terrain-corrected gamma backscatter. The overall accuracy for forest/non-forest mapping and forest type classification improved by 2% and 4%, respectively, over the whole of Austria, with improvements of up to 16% and 20%, respectively, in regions with strong topography.

Removal of Ionospheric Effects from Sigma Naught Images of the ALOS/PALSAR-2 Satellite

The monitoring of forest degradation in the Amazon through radar remote sensing methodologies has increased intensely in recent years. Synthetic aperture radar (SAR) sensors that operate in L-band have an interesting response for land use and land cover (LULC) as well as for aboveground biomass (AGB). Depending on the magnetic and solar activities and seasonality, plasma bubbles in the ionosphere appear in the equatorial and tropical regions; these factors can cause stripes across SAR images, which disturb the interpretation and the classification. Our article shows a methodology to filter these stripes using Fourier fast transform (FFT), in which a stop-band filter removes this noise. In order to make this possible, we used Environment for Visualizing Images (ENVI), Sentinel Application Platform (SNAP), and Interactive Data Language (IDL). The final filtered scenes were classified by random forest (RF), and the results of this classification showed superior performance compared to the original scenes, showing this methodology can help to recover historic series of L-band images.

A Convolutional Neural Network Algorithm for Soil Moisture Prediction from Sentinel-1 SAR Images

Achieving the rational, optimal, and sustainable use of resources (water and soil) is vital to drink and feed 9.725 billion by 2050. Agriculture is the first source of food production, the biggest consumer of freshwater, and the natural filter of air purification. Hence, smart agriculture is a “ray of hope” in regard to food, water, and environmental security. Satellites and artificial intelligence have the potential to help agriculture flourish. This research is an essential step towards achieving smart agriculture. Prediction of soil moisture is important for determining when to irrigate and how much water to apply, to avoid problems associated with over- and under-watering. This also contributes to an increase in the number of areas being cultivated and, hence, agricultural productivity and air purification. Soil moisture measurement techniques, in situ, are point measurements, tedious, time-consuming, expensive, and labor-intensive. Therefore, we aim to provide a new approach to detect moisture content in soil without actually being in contact with it. In this paper, we propose a convolutional neural network (CNN) architecture that can predict soil moisture content over agricultural areas from Sentinel-1 images. The dual-pol (VV–VH) Sentinel-1 SAR data have being utilized (V = vertical, H = horizontal). The CNN model is composed of six convolutional layers, one max-pooling layer, one flatten layer, and one fully connected layer. The total number of Sentinel-1 images used for running CNN is 17,325 images. The best values of the performance metrics (coefficient of determination (R2=0.8664), mean absolute error (MAE=0.0144), and root mean square error (RMSE=0.0274)) have been achieved due to the use of Sigma naught VH and Sigma naught VV as input data to the CNN architecture (C). Results show that VV polarization is better than VH polarization for soil moisture retrieval, and that Sigma naught, Gamma naught, and Beta naught have the same influence on soil moisture estimation.

Application of C-band sentinel-1A SAR data as proxies for detecting oil spills of Chennai, East Coast of India

This paper presents the utilization of Synthetic Aperture Radar (SAR) data for monitoring and detection of oil spills. In this work, a case study of an oil spill has been investigated using C-band Sentinel-1A SAR data to detect the oil spill that occurred on 28 January 2017, near Ennore port, Chennai, India. Oil spill damages marine ecosystems causing serious environmental effects. Quite often, oil spills on the sea/ocean surface are seen nowadays, mainly in major shipping routes. They are caused due to tanker collisions, illegal discharge from the ships, etc. An oil spill can be monitored and detected using various platforms such as vessel-based, airborne-based and satellite-based. Vessel based and airborne methods are expensive with less area coverage. This process also consumes more time. For ocean applications such as oil spill and Ship detection, optical sensors cannot image during bad weather. As SAR is an active sensor, weather independent, and has cloud penetrating capability, the images can be acquired during the day as well as at night. Radar Remote Sensing (RRS) has rapidly gained popularity for monitoring and detection of oil spills and ships for more than a decade. With the availability of the satellite images, detection of oil spill has improved due to its wide coverage and less revisit time. The present paper gives an overview of the methodologies used to detect oil spills on the SAR images using dual-pol Sentinel-1A Level 1 SLC data. This work clearly demonstrates the preprocessing steps of the Sentinel 1A data for oil spill detection. The oil spill was only visible in the VV channel, therefore, for ocean application VV channel image is preferred. SEASAT was the first space-borne SAR mission launched in 1978 by NASA to observe sea surface. The preprocessing was carried out at the European Space Agency (ESA), the Sentinel Application Platform (SNAP) toolbox and Envi 5.1 toolbox. Based on the Sigma naught values, oil spill can be discriminated with the ocean surface. The results obtained with the VV channel are satisfactory and one could map out the oil spill very well. Supervised classifiers SVM and NN were applied on the boxcar filtered 3 × 3 VV channel image to delineate the oil spill. The result of oil spill detection mapping is validated with Supervised SVM and Neural Network classifiers. The results show there is a good agreement between oil spill mapping and classified image using SVM and NN classified images. The Overall Accuracy (OA) obtained using SVM classifier is 98.13% with kappa coefficient as 0.95 and using NN classifier is 98.11% with kappa coefficients 0.95. This technique is considered to be a potential proxy for the detection and monitoring of Oil spills on water bodies. Application of SAR data for oil spill detection is considered to be first of its kind from Indian coasts. This study aims to detect the oil spill occurred due to collision of two LPG tankers with Sentinel-1A SLC data in Chennai coast area.

Development of an Automated Tool for Delineation of Flood Footprints from SAR Imagery for Rapid Disaster Response: A Case Study

Floods are one of the most common natural disasters. In recent times, microwave synthetic aperture radar (SAR) satellite images have been used widely for mapping flood affected areas due to its all-weather capability and acquisition during day and night. Here, in this paper, an automated algorithm is proposed to delineate flood extent from SAR images without any human intervention. The algorithm consists of pre-processing steps like applying orbit file, calibrate to sigma naught, speckle filtering, terrain correction and linear to decibel conversion. The water layer is delineated using multi-segmentation and Otsu’s thresholding technique. Further, flood layer is extracted by postprocessing steps using majority filter, applying permanent water body mask and eliminating hill shadows. The algorithm is tested on a significant number of satellite images covering floods in India, which are having diverse terrain and flooding patterns. In this paper, the steps involved in delineating flood from SAR image of VH polarization from SENTINEL-1 satellite covering chronic flood-prone stretch of part of Ganga River in Bihar state is presented. Accuracy assessment is carried out with the flood layer derived from RADARSAT SAR HV polarized data acquired on the same day and an accuracy of about 96% is obtained. The total processing time taken for the extraction of the flood layer is 9 min. This automation process is beneficial for the generation of rapid flood inundation maps, with high accuracy, which is helpful for flood monitoring and relief management during a disaster.

An integrated approach for identification of waterlogged areas using RS and GIS technique and groundwater modelling

An integrated approach using remote sensing (RS) and geographical information system (GIS) and groundwater modelling (GM) has been used for waterlogged areas identification in the Rohtak district of Haryana State, India. Surface waterlogged areas were delineated using optical remote sensing satellite data-based normalized difference water index (NDWI) technique. Sentinel 2 MSS (optical data) images pertaining to pre-monsoon and post-monsoon seasons were acquired and processed to extract water pixels. To overcome the limitation of false positives and cloud penetration associated with optical images, water pixels were also extracted using synthetic aperture radar (SAR) images of Sentinel 1A. Thresholding of NDWI for optical images and sigma naught for SAR images was done using the respective histograms to distinguish water and terrestrial features. Surface waterlogged areas were delineated from the generated combined water body raster based on the visual interpretation technique in a GIS environment. Sub-surface waterlogging conditions were simulated for the same year using process-based groundwater model MODFLOW. A conceptual model was created to simulate groundwater system based on flow processes, hydrogeological characteristics, aquifer hydraulic properties, source and sinks and boundary conditions. Sub-surface waterlogged areas were delineated based on depth to water table classification criteria. It was found that surface waterlogged area varies from 0.55 to 1.0% of the district area from pre-monsoon to post-monsoon season, while the percentage of sub-surface waterlogged areas was higher, varying from 9.1 to 21.6%, respectively. Statistics also shows that most of the surface waterlogged areas do not fall under the areas plagued by sub-surface waterlogging.

Mass Balance of the Himalayan Glaciers and Their Regional Variations

Aim: Popular methods of glacial mass balance studies are time consuming, laborious and therefore, not repetitive at yearly manner. A compact, user-friendly classification algorithm has been applied to estimate Equilibrium Line Altitude (ELA) and mass balance of mountain or valley glaciers at yearly basis using Synthetic Aperture Radar (SAR) data for a period of 2012 to 2015. Cloud cover over the glacier catchment areas was studied to identify major seasons of precipitation over the study areas. The results were also compared to study the regional characteristics of the glaciers.
 Study Area: The proposed module was used to analyze data over three Himalayan glaciers, namely, Durung Drung, Gangorti and Zemu. Chhota Shigri glacier was also studied with these three glaciers to understand the regional variations of the Himalayan glaciers.
 Methodology: The classification algorithm was fixed for all glaciers and also independent from season, therefore, required least user intervention. The conditional loop based logics, consist of linear equations, classified glaciated region in different physical zones. Purely backscatter based classification result produced error by mixing zones due to overlapped signatures. Altitude thresholds of accumulation zones were employed to segregate the mixing in next level. The method required calibrated ortho-rectified sigma naught dual-pol SAR imagery as primary input. The glaciated area should be provided as Area of Interest (AOI). A Digital Elevation Model (DEM) file is required for altitude threshold. The output classes are saved in separate files with Boolean values. Optical data were used to estimate cloud coverage over the catchments of the glaciers.
 Results: Average mass balance of the selected glaciers is estimated as - 0.44 m w.e.a-1 during the study period. The mass balance of the glaciers are comparatively studied with variation in melting seasons, duration of melting periods, on set and cease of melting for each glacier to understand the regional pattern of mass loss.
 Conclusion: Co and cross polarized SAR data are employed to derived debris size; however, cross polarized SAR backscattering has better correlation with debris size. The accuracy of the result derived from the developed method is ± 50 mm. Collection of field data on the surface topography is difficult for Mountain glaciers, especially over Himalaya. Use of satellite data can generate detailed information of glacier surface which will be further help to understand role of debris in glacier mass balance.

Verification of Kompsat-5 Sigma Naught Equation

Verification of Kompsat-5 Sigma Naught Equation

Retrieval and validation of soil moisture from FRS-1 data set of radar imaging satellite (RISAT-1)

In this work, we estimated soil moisture (SM) by using dielectric constant properties of soil with radar backscattering coefficient (bc) from simulated annealing techniques of RISAT-1 (radar imaging satellite, based on synthetic aperture radar (SAR) technique) data. We examined the performance of simulated annealing in retrieving SM where the vegetation cover is not very high (NDVI ≈ 0.35 for wheat-dominated area on January 19, 2013).To overcome the land surface model limits on SM estimation accuracy, point measurement spatial coverage limits, and microwave remote sensing spatial-temporal sampling limits, we reduced uncertainties through a combination of these approaches. Near-surface SM measurements from the 5.35-GHz (C-band) channels of RISAT-1 were collocated against ground-truth data (collected during the flight time of RISAT-1 over the study area), to establish SAR-SM relationships for FRS-1(circular horizontal (RH) and circular vertical (RV)) data set of RISAT-1. Comparison with the limited ground-based point (total 24 points) measurements of SM content exhibited a net improvement when near-surface SM observations were assimilated. Comparison of the SM derived from the sigma naught (σ 0) (of RISAT-1 data set FRS-1) using the inversion algorithm with the observed measurements (using time domain reflectometry) of SM showed root mean square error of 0.24, nRMSE = 0.03, R-RMSE = 0.38, MAE = 0.63, NRMSE = 1.02, NSE = 1, d = 0.87, r 2 = 0.65, and RMSE% = 12.79 for RH polarized image while RV polarized image failed each statistical test for predicting good SM with respect to the observed SM. The ability to extract additional information comes at the expense of including more measurements, especially at frequencies lower than the L-band. This approach is therefore intended for future space-borne systems.

Karakterisktik Backscatter Citra Alos Palsar Polarisasi Hh dan Hv terhadap Parameter Biofisik Hutan di Sebagian Taman Nasional Kerinci Seblat

Synthetic Aperture Radar (SAR) is one kind of an active remote sensing system, utilizing microwave to records the earth surface. One of the SAR’s satellites is ALOS PALSAR which is capable of penetrating clouds so that the underlying objects can be identified. ALOS PALSAR has a polarization system operating with single beam polarization (HH or HV), dual polarization (HH + HV or VV + VH) and full polarization (HH + HV + VH + VV). This study aims to use the backscatter value of ALOS PALSAR's image for monitoring and mapping the forest. The Utilization of ALOS PALSAR imagery to measure forest biophysical parameters and saturation levels identified based on backscatter sensitivity. Image extraction is done through image calibration where the digital value is converted to sigma naught backscatter value then the result of calibration unit is changed to decibels. Backscatter sensitivity depends on forest structure which can affect scattering mechanism. The backscatter value is also influenced by humidity and weather dynamics which can also affect the dielectric constant. The result of accurate classification of land use polarization fall in HH accuracy of 62,4%. Whereas for HV polarization accuracy obtained equal to 74,88%.

Crop Classification by Machine Learning Algorithm Using C-band SAR Data

This paper presents crop classification using satellite data to establish a mapping method to replace the existing ground survey. We used five scenes of C-band fully polarimetric SAR satellite Radarsat-2 data. Datasets of sigma naught and four polarimetric parameters, Freeman－Durden (FD), Van Zyl (VZ), Yamaguchi (YG), and Cloude－Pottier (CP), were calculated from each image data. We assessed the accuracy of the classification obtained by the random forest machine learning algorithm. Three results are shown. First, the highest accuracy using only one of the five datasets (0.918) was obtained by the VZ parameter dataset. Second, using three datasets, the combination of the sigma naught, VZ parameter, and CP parameter datasets obtained the highest accuracy (0.922). Third, when we used all five datasets, the accuracy (0.918) was not improved. These results confirm that crop classification using Radarsat-2 C-band data is very effective and the use of a combination of sigma naught, VZ parameters, and CP parameters obtained the highest accuracy.

Examining High-Resolution PiSAR-L2 Textures for Estimating Tropical Forest Carbon Stocks

This study examines the potential of airborne PiSAR-L2 data for estimating forest carbon stocks in central Sumatra. Polarimetric interferometric synthetic aperture radar L-band-2 (PiSAR-L2) is a second-generation airborne sensor developed by JAXA. We acquired full-polarimetric data at a fine spatial resolution of 2.5 m during the PiSAR-L2 flight campaign in August 2012. A total of 59 field measurement plots for aboveground forest carbon stocks (AFCSs) were established in same year where AFCS ranged between 4.8 and $253.5\; \text{Mg}\; \text{C} \;\text{ha}^{-1}$ . The plots comprised natural and plantation forests. These plot-level field data were used for calibrating and validating AFCS estimation models with the SAR data. Various possibilities including direct sigma naught backscatters and their ratios and various types of textures obtained from HH, HV, and VV polarizations were examined by applying regression modeling. The main indicators used for the selection of best potential models in the calibration phase were $\text{R}^{2}$ , variable inflation factor (VIF), p-value, and root-mean-squared errors (RMSEs). The potential models were validated using the leave-one-out (LOO) method. The results indicated that a simple combination of backscatters and their ratios provides an AFCS estimate with an RMSE of $42.37\; \text{Mg}\; \text{C}\; \text{ha}^{-1}$ and an $\text{R}^{2}$ of 0.65. Inclusion of SAR textural parameters improved the AFCS estimates with an RMSE of $30.93\;\text{Mg}\; \text{C}\; \text{ha}^{-1}$ and an $\text{R}^{2}$ of 0.80. This indicates that the airborne PiSAR-L2 full-polarimetric data have the potential to estimate forest carbon stocks with an improved accuracy in the tropical region.

EXTRACTION OF FLOODED AREAS DUE THE 2015 KANTO-TOHOKU HEAVY RAINFALL IN JAPAN USING PALSAR-2 IMAGES

Triggered by two typhoons, heavy rainfall hit Kanto and Tohoku regions of Japan from September 9 to 11, 2015. Increased river water by the continuous rainfall overflowed and destroyed several river banks and caused damaging floods in wide areas. PALSAR-2 onboard ALOS-2 satellite carried out emergency observation for the impacted areas during and after the heavy rainfall. In this study, two pre-event and four co- and post-event PALSAR-2 images were used to extract the inundation area in Joso city, Ibaraki prefecture. First, using the pre-event SAR intensity image and a detailed topographic map, the backscattering coefficient of river water was investigated. Then the flooded areas were extracted by a common threshold value of backscatter for water bodies in the six temporal images. The colour composite of the sigma naught values was also made to visualize pixels that had been converted from ground to water. Finally, the extracted results were compared with those from the visual interpretation of aerial photographs and field survey reports. This comparison revealed that the accuracy of the flood extraction was fairly good for agricultural lands and non-urban land uses. But for built-up urban areas, it was not easy to extract water body since radar illumination did to reach the ground (water) surface.

EXTRACTION OF FLOODED AREAS DUE THE 2015 KANTO-TOHOKU HEAVY RAINFALL IN JAPAN USING PALSAR-2 IMAGES

Abstract. Triggered by two typhoons, heavy rainfall hit Kanto and Tohoku regions of Japan from September 9 to 11, 2015. Increased river water by the continuous rainfall overflowed and destroyed several river banks and caused damaging floods in wide areas. PALSAR-2 onboard ALOS-2 satellite carried out emergency observation for the impacted areas during and after the heavy rainfall. In this study, two pre-event and four co- and post-event PALSAR-2 images were used to extract the inundation area in Joso city, Ibaraki prefecture. First, using the pre-event SAR intensity image and a detailed topographic map, the backscattering coefficient of river water was investigated. Then the flooded areas were extracted by a common threshold value of backscatter for water bodies in the six temporal images. The colour composite of the sigma naught values was also made to visualize pixels that had been converted from ground to water. Finally, the extracted results were compared with those from the visual interpretation of aerial photographs and field survey reports. This comparison revealed that the accuracy of the flood extraction was fairly good for agricultural lands and non-urban land uses. But for built-up urban areas, it was not easy to extract water body since radar illumination did to reach the ground (water) surface.

A Preliminary Study to Forecast Japanese Encephalitis Vector Abundance in Paddy Growing Area, with the Aid of Radar Satellite Images.

Vector mosquitoes of Japanese encephalitis (JE) breed mostly in rice fields, and human cases occur scattered over extended rural rice-growing areas. From this, one may surmise an ecological connection with the irrigation facilities and paddy cultivation. Furthermore, it has been hypothesized that a particular stage of paddy growth is a premonitory sign that can lead to a markedly increased population of the vector mosquitoes. The present study aimed to forecast the vector abundance by monitoring the paddy growth using remote sensing and geographical information systems. The abundance of the JE vector Culex tritaeniorhynchus peaked when the paddy crop was at its heading stage and dipped when the crop reached the maturing stage. A significant positive correlation was observed between paddy growth and adult density (r = 0.73, p < 0.008). The sigma naught values (σ0) derived from satellite images of paddy fields ranged from -18.3 (during transplantation stage) to approximately -10 (during the noncultivation period). A significant positive correlation was observed between σ0 and paddy growth stages (r = 0.87, p < 0.05) and adult vector density (r = 0.74, p = 0.04). The σ0 value observed during the vegetative and flowering stages of paddy growth ranged from -17.6 to -17.16, at which period the vector density started building up. This could be the spectral signature that denotes the "risk," following which a high vector abundance is expected during heading stage of the paddy.

Estimation of Above Ground Biomass for Central Indian Deciduous Forests Using ALOS PALSAR L-Band Data

Reliable and accurate estimates of tropical forest above ground biomass (AGB) are important to reduce uncertainties in carbon budgeting. In the present study we estimated AGB of central Indian deciduous forests of Madhya Pradesh (M.P.) state, India, using Advanced Land Observing Satellite – Phased Array type L-band Synthetic Aperture Radar (ALOS-PALSAR) L-band data of year 2010 in conjunction with field based AGB estimates using empirical models. Digital numbers of gridded 1 × 1° dual polarization (HH & HV) PALSAR mosaics for the study area were converted to normalized radar cross section (sigma naught - σ0). A total of 415 sampling plots (0.1 ha) data collected over the study area during 2009–10 was used in the present study. Plot-level AGB estimates using volume equations representative to the study area were computed using field inventory data. The plot-level AGB estimates were empirically modeled with the PALSAR backscatter information in HH, HV and their ratios from different forest types of the study area. The HV backscatter information showed better relation with field based AGB estimates with a coefficient of determination (R2) of 0.509 which was used to estimate spatial AGB of the study area. Results suggested a total AGB of 367.4 Mt for forests of M.P. state. Further, validation of the model was carried out using observed vs. predicted AGB estimates, which suggested a root mean square error (RMSE) of ±19.32 t/ha. The model reported robust and defensible relation for observed vs. predicted AGB values of the study area.

Utilization of Pisar L-2 Data for Land Cover Classification in Forest Area Using Pixel-Based and Object-Based Methods

Abstract. Polarimetric and Interferometric Airborne SAR in L-band 2 (PiSAR-L2) program is an experimental program of PALSAR-2 sensor in ALOS-2 satellite. Japan Aerospace Exploration Agency (JAXA) and Indonesian National Institute of Aeronautics and Space (LAPAN) have a research collaboration to explore the utilization of PiSAR-L2 data for forestry, agriculture, and disaster applications in Indonesia. The research explored the utilization of PiSAR-L2 data for land cover classification in forest area using the pixel-based and object-based methods. The PiSAR-L2 data in the 2.1 level with full polarization bands were selected over part of forest area in Riau Province. Field data collected by JAXA team was used for both training samples and verification data. Preprocessing data was carried out by backscatter (Sigma naught) conversion and Lee filtering. Beside full polarization images (HH, HV, VV), texture imagess (HH deviation, HV deviation, and VV deviation) were also added as the input bands for the classification processes. These processes were conducted for 2.5 meter and 10 meter spatial resolution data applying two methods of the maximum likelihood classifier for pixel-based classification and the support vector machine classifier for the object-based classification. Moreover, the average overall accuracy was calculated for each classification result. The results show that the use of texture images could improve the accuracy of land cover classification, particularly to differentiate between forest and acacia plantation. The pixelbased method showed a more detail information of the objects, but has “salt and pepper”. In the other hand, the object-based method showed a good accuracy and clearer border line among objects, but has often some misinterpretations in object identification.

TOWARDS AN EASIER ORIENTATION FOR SPHERICAL PHOTOGRAMMETRY

Abstract. For architectural metric documentation, Spherical Photogrammetry (SP) has demonstrated its validity and efficiency in many projects already. The speed of surveying is high, the accuracy and completeness of the plotting are satisfactory. However, there are still many problems to be solved. The weakest point is the orientation procedure, which is rather difficult to perform, in the sense that only very experienced people can run it, and few people only make use of it. The old orientation steps are 1) model formation (limited to binocular panoramas couples); 2) link of all the models in a block adjustment with independent model triangulation; 3) block bundle adjustment with 4 parameters/pano (3 coord.+1 orientation bearing); 4) block bundle adjustment with 6 parameters/pano, say the previous 4 + 2 correction angles around the horizontal axes. The panoramas must be spherical and quasi-horizontal. In order to make easier the orientation, enabling more people to use SP, an improved approach has been set up. It consists in the combination of any possible model formed either by three and two panoramas. The trinocular vision, say the combination of three different panoramas to form a unique model, has the advantage to be much more robust in comparison to binocular vision in the sense that the trinocular model is likely to be more error-free than any of the three composing binocular models. It contains less model deformation, the model coordinates are validated by the mutual comparison of the three intersecting binocular models. In addition, the number of possible trinocular models is normally much larger than the one of binocular models. The steps for a semi-automatic orientation of a block of panoramas proceed as follows: - Form any possible trinocular models by combination of the panoramas; - in case that no trinocular model has been formed, form any possible binocular model; - run a block adjustment with the algorithm of independent model, to link together the models in a unique reference system; - run a block bundle adjustment with collinearity equations with 4 unknown parameters per panorama; - run a block bundle adjustment with collinearity equations with 6 unknown parameters per panorama. The control requirements are limited to one fixed control point and two perpendicular constraints. Given a project with n panoramas, the trinocular possible models are the combination of n, 3 by 3, say (n*(n-1)*(n-2))/3!. As an example, in a project with 10 panoramas the trinocular possible models are 120, while the possible binocular models are (n*(n-1))/2= 45. Obviously not all the panoramas will be formed, because a minimum of four common tie points is required. It is possible to select and filter out the best models by fixing a critical value on sigma naught for testing. The advantages are as follows: - no need of approximate value for the unknown coordinates and parameters; - easiness to find gross errors in the gradual block formation. The epipolar geometry has been investigated, the expression of the curve described and assessed: it can be useful in the trinocular model formation. For any try of the model the operator, thus causing troubles to him, must approximately choose formation, up to now the starting value of the orientation bearing for any panorama. However, a new procedure, not requiring any approximate value of the orientation bearing, has been set-up and tested: the orientation bearing can vary stepwise by a given selected interval, until the model is successfully formed. The computation time increases significantly, but it is still negligible. Up to now the control and tie points selection and observation is manual, making the orientation and the plotting to be slow and time consuming. Nevertheless, on the other end, the procedure is fully controlled and supervised by the operator, enabling the observation of the residuals and allowing the improvement of the orientation, and to perform some statistical quality analysis on the adjustment, which are not possible with automatic procedures based on Structure from Motion (SfM). Another weak point of the SP is the speed of plotting. Commercial programs are now available, based on SFM algorithm, producing point clouds of good quality in an unknown scale and orientation. The correct scale and orientation can be therefore supplied by SP orientation with high accuracy. Examples are shown taken from two real projects.

Extreme Learning Machine-based Crop Classification using ALOS/PALSAR Images

Classification maps are required for agricultural management and the estimation of agricultural disaster compensation. The extreme learning machine (ELM), a newly developed single hidden layer neural network is used as a supervised classifier for remote sensing classifications. In this study, the ELM was evaluated to examine its potential for multi-temporal ALOS/PALSAR images for the classification of crop type. In addition, the k-nearest neighbor algorithm (k-NN), one of the traditional classification methods, was also applied for comparison with the ELM. In the study area, beans, beets, grasses, maize, potato, and winter wheat were cultivated; and these crop types in each field were identified using a data set acquired in 2010. The result of ELM classification was superior to that of k-NN; and overall accuracy was 79.3%. This study highlights the advantages of ALOS/PALSAR images for agricultural field monitoring and indicates the usefulness of regular monitoring using the ALOS-2/PALSAR-2 system. Discipline: Agricultural engineering Additional key words: Hokkaido, machine learning, sigma naught *Corresponding author: reysnb@gmail.com Received 12 May 2014; accepted 4 February 2015. Introduction Land-cover classification is one of the most common applications of remote sensing. Crop-type classification maps are useful for yield estimation and agricultural disaster compensation, in addition to the management of agricultural fields. Optical remote sensing remains one of the most attractive options for the accumulation of biomass information and forest monitoring (Samreth et al. 2012, Sarker and Nichol 2011). In addition, while optical satellites such as ALOS/AVNIR-2 (Sonobe et al. 2014a), Landsat (Hartfield et al. 2013), MODIS (Sakamoto et al. 2009), and NOAA (Hirano and Batbileg 2013) have been employed in the identification of species and conditions of vegetation, cloud cover significantly limits the number of available optical images, radar is unaffected by cloud cover or low solar zenith angles (Bindlish and Barros 2001). And significant information about soil and vegetation parameters can also be obtained through microwave remote sensing (Sonobe et al. 2014b). These techniques are employed increasingly to manage land and water resources for agricultural applications (Sonobe et al. 2014c). The number of studies on rice and wheat monitoring and mapping using SAR data has increased, and some studies utilizing multi-temporal SAR data have reported high correlations between backscattering coefficients, and plant height and age (Chakraborty et al. 2005, Sonobe et al. 2014d, Waisurasingha et al. 2008). These examples highlight possible uses in the area of agricultural management, specifically for the identification of paddy fields. They indicate the potential use of SAR data for the discrimination of crop types. Furthermore, a number of examples have shown the practical usefulness of supervised learning for classification (Sonobe et al. 2014b). This paper reports a comparison of crop classification using PALSAR data performed by extreme learning machine (ELM) and k-nearest neighbor algorithm (k-NN).