Increasingly, new satellites with dual resolutions capabilities are available, which include Landsat 7, SPOT 1-5, EO-1, IKONOS, QuickBird, WorldView-2, GeoEye-1, FormoSat and DubaiSat. Such sensors capture simultaneously a high (spatial) resolution panchromatic image (HRPI), which is good for identifying spatial details, and a low (spatial) resolution multispectral image (LRMI), which is pertinent for the spectral classification of features. Spatialresolution is the most significant factor that influences the accuracy of freshwater vegetation classifications because of their limited width and their heterogeneous nature (Booth et al., 2007; Goetz, 2006; Ozesmi and Bauer, 2002). Data fusion (DF) techniques aim to integrate HRPI and LRMI to produce a high (spatial) resolution multispectral image (HRMI) for further analysis. Many earlier techniques produce one possible HRMI; however, different applications, according to their purpose of data fusion, require a focus either on the spectral information from the LRMI or on the spatial details from the HRPI (Chen et al., 2006). Contemporary data fusion methods use sophisticated algorithms to balance these characteristics to ensure the best integration of spectral and spatial qualities of the input data (Boloorani et al., 2005). Different applications may require different balances between spectral characteristic preservation and high spatial detail retention. For classification purposes it is important to preserve the spectral information, whereas other applications (e.g. feature extraction and cartography) may only require a sharp and detailed display of the scene (Cetin and Musaoglu, 2009; Chen et al., 2006).
This research aims first at the estimation of water fractions within the mixed pixels (i.e. unmixing) and then at the spatial allocation of corresponding sub-pixels (i.e. SRM) in order to map river boundaries at the sub-pixel level. To this end, NDWIs are leveraged for the estimation of water fractions. As different combinations of spectral bands can be used in the structure of NDWI, a full search approach is proposed to identify the optimal pair of bands leading to the highest correlation of NDWI values with water fractions. The effectiveness of a regression model is explored for estimation of water fractions from NDWI values. The accuracy of the proposed method is compared against an advanced unmixing method, namely fully constrained simplex projection unmixing (SPU). A thorough investigation is carried out on the performance of SRM techniques in the context of river mapping. Several SRM techniques are focused including spatial optimization techniques such as pixel swapping (PS) as well as some interpolation-based algorithms. Furthermore, the PS algorithm is modified to speed up the binary water/non-water classification. Both semi-simulated and the fractions derived from real imagery are used for evaluation of SRM techniques. The first of these provides the possibility of accuracy assessment of the sole spatial allocation of sub-pixels task, while the latter considers also the uncertainties involved in estimation of water fractions. In addition, effectiveness of current thresholding methods on NDWIs is examined for hard water/non-water classification. Small rivers have been the interest of this study and, accordingly, HRSI including WorldView-2 (WV-2) and Geoeye imagery are used to exercise the implementations.
ABSTRACT: Life on Earth depends on water, yet water resources are severely stressed by the rapid growth of the human population and activities. In arid environments the exploration and monitoring of water resources is a prerequisite for water accessibility and rational use and management. To survey large arid areas for water, conventional land-based techniques must be complemented by using satellite and airborne remote sensors. Surface water systems can be mapped using image processing techniques; this study assessed the benefits of using higher spatialresolution images. This work attempts to provide a solution to detect or map water bodies in remotelysensed aerial images using image segmentation and image morphology techniques, after applying segmentation and morphology techniques to a input images it will highlights the water bodies. The whole work was done on actual satellite images of Tapi River, Bhusawal, Jalgaon, M.S. , India region.
The study of riparian buffer zones, the areas surrounding streams, is important in understanding water quality, nutrient cycles, and erosion and sedimentation deposition processes. In this work, we mapped and identified riparian zone vegetation along streams within the Neuse River Basin of Eastern North Carolina using Digital Orthophoto Quarter Quadrangles (DOQQs) that were created using National Aerial Photography Program (NAPP) 1:40,000-scale color infrared aerial (CIR) photography. The main objectives of this study were to create a comprehensive riparian buffer zone database and evaluate the use of high spatialresolutiondata for riparian buffer zone characterization. The database contains both image data and the attributes that were used for riparian vegetation analysis. These attributes include percent vegetation coverage, vegetation type, stream width, and shaded areas. The nature of the Neuse River Basin, which contains three main land cover categories; agricultural, urban, and forest in close proximity to streams, is used to demonstrate the predicting the effectiveness of land-use legislation on water quality.
Coarse and medium spatialresolution multispectral data, such as Landsat, SPOT and MODIS (Moderate Resolution Imaging Spectroradiometer), have been used widely for crop classification and mapping [3,6,7]. However, the accuracy of crop maps generated from these images is inevitably compromised by the spatial limitation , especially over the highly fragmented and heterogeneous agricultural areas. As stated by , a spatialresolution of less than 10 m is required for precision agriculture. More recently, remotelysensed imagery from fine spatialresolution (FSR) (<10 m) satellite systems (e.g., RapidEye, IKONOS, and WorldView) as well as airborne systems (e.g., Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR)) is now available commercially, providing new opportunities for crop classification and mapping in very fine detail [9,10]. However, high intra-class variance and low inter-class separability over croplands in FSR images may exist because of differences in climatic conditions, topographic properties, soil composition, farming practices and so on . Moreover, FSR imagery has fewer multispectral bands (around four) in comparison to medium resolutiondata (e.g., MODIS and Landsat), which leads to subtle differences in spectral/polarimetric properties amongst crop types (i.e., crop types are difficult to discriminate). Therefore, developing advanced classification methods for accurate crop mapping and monitoring is of prime concern, especially with a view to exploiting the deep hierarchical features presented in FSR imagery.
aggregated values produced clearly more fragmented patterns than actual sensor ones. Different aggregation algorithms were tested in the context of forest fragmentation estimates across various spatial scales (Garcia-Gigorro and Saura, 2005). Thirty-meter Landsat-TM forest data were transferred to 188 m IRS-WiFS (Indian Remote Sensing Satellite-Wide Field Sensor) and compared with actual WiFS data. Sensor point spread function was found to greatly improve comparability of forest fragmentation indices. However, a poor performance of power scaling laws was observed at finer spatial resolutions, and accordingly Garcia-Gigorro and Saura (2005) suggested that the true accuracy and practical utility of these scaling functions may have been overestimated in previous literature. In addition, the sensitivity of each of the indices varied with the gradient of spatialresolution. But Cain et al. (1997) conducted a multivariate analysis of pattern metrics, and pointed out that measures of land cover diversity, texture, and fractal dimension were more consistent than measures of average patch shape or compaction among the land cover maps. Wu et al. (2002) summarized the responses of the 19 landscape metrics that fell into three general categories when calculated at the landscape level: Type I metrics showed predictable responses with changing scale, and their scaling relations could be represented by simple scaling equations (linear, power-law, or logarithmic functions); Type II metrics exhibited staircase-like responses that were less predictable; and Type III metrics behaved erratically in response to changing scale, suggesting no consistent scaling relations. Therefore, if metrics fall within category Type I, they can be readily and accurately extrapolated or interpolated across spatial scales, whereas if they fall in Type II or Type III categories, more explicit consideration of idiosyncratic details are required for successful scaling.
dramatically increases the precision of identification of minerals through unique spectral signatures. This has also been enhanced by hyperspectral signature library created in lab conditions containing hundreds of signatures for different minerals, land cover and earth materials. However, hyperspectral data has several limitations. The most significant is the engineering challenge to acquisition at sufficient quality especially at space borne level, this might explain why most of the studies noted in literature employ airborne hyperspectral data and even in cases where space borne data are used, it is mostly in combination with airborne data e.g. (Calvin, et al., 2015; Kratt, et al., 2010; Kruse, 2002; Vaughan, et al., 2005). Other limitations include complex calibration, pre and post processing of data, data redundancy due to acquisitions of many spectral regions of less user interest, environmental conditions and natural variations in materials, which makes them different from standard library materials, are still some of the many challenges in hyperspectral applications. Satellite imaging usually covers a larger area at a lower cost than airborne hyperspectral surveys and is a good remote sensing option when surveying larger regions. However, airborne hyperspectral surveys despite being costly, typically produce greater spatial and spectral fidelity for imaging of surface indicators and are usually suitable in more site specific remote sensing operations. Consequently, majority of previous studies on GT exploration of surrogate minerals have employed airborne sensors e.g. SEBASS, Hymap due to higher resolution and better precision in mineral identification as compared to space borne sensors. The Hyperion, a hyperspectral space borne imager which has been operational since the year 2000 is not without its limitations, as a result of data quality, spatial coverage and the complexity of the pre-processing of the data.
Monitoring irrigation water demands and consumption requires mapping irrigated areas either through agricultural census or using remotely sensing data. The current spatiotemporal extent of irrigated lands and inter-annual change at regional scales in India is still relatively uncertain and available maps are often outdated or prepared with spatially-coarse resolutiondata. The primary sources of irrigation data in India are the Directorate of Economics and Statistics of the Ministry of Statistics (DES), Ministry of Water Resources (MoWR), and Food and Agricultural Organization of United Nations (FAO). During the past few years, several spatialdata sets of irrigated area at global scale have been developed. For instance, the USGS Global Land Cover Map 8 was developed using 1 km monthly composite of NDVI obtained from Advanced Very High Resolution Radiometer (AVHRR). The Global Map of Irrigation Areas (GMIA) published by the FAO was developed by Siebert, et al. 9 using approximate information of total irrigated area from national information and other data sources (irrigated area per national statistical unit, irrigated area from point, polygon, and raster maps of land cover and other satellite data) at a spatialresolution of 5-arc minutes. Recently, International Water Management Institute (IWMI) released global irrigated area map for a 10-km grid resolution using methods described in Thenkabail, et al. 10 . Moreover, Zhao and Siebert 11 developed crop class based irrigated area maps for India using net sown area and extent of irrigated crops from the census and land use land cover data at 500 m spatialresolution for year 2005. For the Indian region, a high resolution (250–1000 m) irrigation map based on remote sensing data was completed 10,12 for the Ganga, Indus, and Krishna River basins. Siddiqui, et al. 13 developed irrigated area map for Asia and Africa regions using canonical correlation analysis and time lagged regression, which is available at 250 m resolution for the year 2000 and 2010 and can be obtained from the International Water Management Institute (IWMI, http://waterdata.iwmi.org/applications/ irri_area/) portal. However, high resolution (250 m) irrigated area maps that cover the period of 2000–2015 and all the agroecological zones of India are unavailable, which are required for estimation of irrigation water use and hydrologic modelling. Here, we develop annual irrigated area maps at a spatialresolution of 250 m for the period of 2000–2015 using data from the MODIS and high resolution land use/land cover (LULC) information in India.
Abstract: Accurate information on crop distribution is of great importance for a range of applications including crop yield estimation, greenhouse gas emission measurement and management policy formulation. Fine spatialresolution (FSR) remotelysensed imagery provides new opportunities for crop mapping at a detailed level. However, crop classification from FSR imagery is known to be challenging due to the great intra-class variability and low inter-class disparity in the data. In this research, a novel hybrid method (OSVM-OCNN) was proposed for crop classification from FSR imagery, which combines a shallow-structured object-based support vector machine (OSVM) with a deep-structured object-based convolutional neural network (OCNN). Unlike pixel-wise classification methods, the OSVM-OCNN method operates on objects as the basic units of analysis and, thus, classifies remotelysensed images at the object level. The proposed OSVM-OCNN harvests the complementary characteristics of the two sub-models, the OSVM with effective extraction of low-level within-object features and the OCNN with capture and utilization of high-level between-object information. By using a rule-based fusion strategy based primarily on the OCNN’s prediction probability, the two sub-models were fused in a concise and effective manner. We investigated the effectiveness of the proposed method over two test sites (i.e., S1 and S2) that have distinctive and heterogeneous patterns of different crops in the Sacramento Valley, California, using FSR Synthetic Aperture Radar (SAR) and FSR multispectral data, respectively. Experimental results illustrated that the new proposed OSVM-OCNN approach increased markedly the classification accuracy for most of crop types in S1 and all crop types in S2, and it consistently achieved the most accurate accuracy in comparison with its two object-based sub-models (OSVM and OCNN) as well as the pixel-wise SVM (PSVM) and CNN (PCNN) methods. Our findings, thus, suggest that the proposed method is as an effective and efficient approach to solve the challenging problem of crop classification using FSR imagery (including from different remotelysensed platforms). More importantly, the OSVM-OCNN method is readily generalisable to other landscape classes and, thus, should provide a general solution to solve the complex FSR image classification problem.
We developed a Sub-pixel Imperviousness Change Detection ( SICD ) approach to detect urban land-cover changes using Landsat and high-resolution imagery. The sub-pixel percent imperviousness was mapped for two dates (09 March 1993 and 11 March 2001) over western Georgia using a regression tree algorithm. The accuracy of the predicted imperviousness was reasonable based on a comparison using independent reference data. The average absolute error between predicted and reference data was 16.4 percent for 1993 and 15.3 percent for 2001. The correlation coefficient (r) was 0.73 for 1993 and 0.78 for 2001, respectively. Areas with a significant increase (greater than 20 percent) in impervious surface from 1993 to 2001 were mostly related to known land-cover/land-use changes that occurred in this area, suggesting that the spatial change of an impervious surface is a useful indicator for iden- tifying spatial extent, intensity, and, potentially, type of urban land-cover/land-use changes. Compared to other pixel-based change-detection methods (band differencing, rationing, change vector, post-classification), information on changes in sub-pixel percent imperviousness allow users to quantify and interpret urban land-cover/land-use changes based on their own definition. Such information is considered complemen- tary to products generated using other change-detection meth- ods. In addition, the procedure for mapping imperviousness is objective and repeatable, hence, can be used for monitoring urban land-cover/land-use change over a large geographic area. Potential applications and limitations of the products developed through this study in urban environmental studies are also discussed.
In an end-to-end computational infrastructure users should be able to evaluate data, develop science models, produce improved earthquake forecasts, and respond to disasters in intuitive map-based interfaces. Fault models can be constrained and improved not just by geology, but also by feature identification from InSAR (UAVSAR) and inversions of both GPS and InSAR crustal deformation data [2-3]. Forecasting is improved by development of better interacting fault models, pattern analysis, and fusion of both seismicity and crustal deformation data. Increasing the accessibility and utility of GPS, InSAR, and geologic data, addresses science challenges such as earthquake forecasting or fluid migration. Intuitive computational infrastructure also can enable new observations by providing tools to conduct simulation experiments and new information products for use in a wide variety of fields ranging from earthquake research to earthquake response. Timely and affordable delivery of information to users in the form of high-level products is necessary for earthquake forecasting and emergency response, but it also necessary for exploiting crustal deformation to enable new discoveries and uses. There are numerous practical issues to establishing an effective computational infrastructure. Of chief importance is that tools be intuitive and easily accessible. Some QuakeSim tools are public and reside outside of any required login. This mode of operation is often preferred by users as it avoids the need to remember another login and password combination and allows for greater privacy. However, there are also limitations. Chief of these is that
C. Deep Learning based classification methods[16-26]: In the last decade various Deep Learning based classification methods were then developed by researchers which were capable of learning the discriminative features on its own using deep learning neural network architectures. The unsupervised feature learning architectures has a shallow architecture whereas deep learning uses multi layered architecture, therefore it has a powerful feature learning capability. So it is capable of extracting the hidden information's and discriminative features of multi dimensional data's. The semantic features of the data are also observed in the top layers itself. All these factors led the successful implementation and state of art performance of deep neural networks architecture in semantic level scene classification.
NASS is working with crop analysts to provide timely and useful imagery and data products. The primary purpose of this visualization is to provide near real-time capability using satellite data to monitor crop growth and progress in the major production areas of the United States. The satellite data provides an independent source of supplementary information to the survey data collected by our enumerators. Crop analysts use the satellite imagery integrated with a geographic information system to help in their assessment of current crop condition and vegetation vigor. NASS uses its GIS capability to combine various layers of information, to overlay image data with State and County boundaries, frost isoline data, and crop information. This visualization concentrates on the integration of GIS map products including AVHRR image data, crop progress of the specific stages of crop development, crop condition, frost isolines and survey data. The Intranet version allows for visualization of crop progress and condition data at the county level (too low a level of aggregation for publishing) and farmer reported survey data indications, which cannot be released.
In the present results, high Chl-a concentration were recorded during northeast monsoon season. Furthermore, intense Chl-a distribution were recorded along the northern part o f Sulu Sea. Meanwhile, the center and southern area o f Sulu Sea shows uniform and low Chl-a concentration. This was related with the influence o f the strong northeasterly wind which blew in December until February. These strong northeasterly winds induced the upwelling process which produced a strong vertical mixing process that brought the cold and nutrient rich water from the subsurface layer to the surface. Hence, this process enhancing the Chl-a concentration in the water surface near the northern part o f Sulu Sea. The occurrence o f this vertical mixing process was led to low SST during this monsoon season.
forecasting beyond a short time window resulted in uncertainty intervals far too wide for practical application. It is also difficult to know ahead of time if the peak has already passed, in which case the smoothing, rather than forecasting, distribution is more appropri- ate. The Bayesian hierarchical forecasts produced more realistic looking vegetation curves with less uncertainty. However, despite incorporating prior information from previous grow- ing seasons, the hierarchical forecasts often failed to capture the true timing of the peak until late in the growing season. This may simply reflect, however, the sensitivity of the crop growth process to changes in environmental conditions. For example, August 2017 was an unexpectedly cool month, which slowed down crop development. Correspondingly, the predictions from the hierarchical model failed to capture the timing of the peak till late September. Adapting prediction intervals to variation in temperature may require additional data sources such as climate model simulations.
Quantifying the amount of precipitation and its uncertainty is a challenging task all over the world, particularly over the African continent, where rain gauge (RG) networks are poorly distributed. In recent decades, several satellite remote sensing (SRS)-based precipitation prod- ucts have become available with reasonable spatial and temporal resolutions to be applied in hydrological and climate studies. However, uncertainties of these products over Africa are largely unknown. In this study, the generalized “three-cornered-hat” (TCH) method is applied to estimate uncertainties of gridded precipitation products over the entire African continent, without being dependent to the choice of a reference dataset. Six widely used SRS-based pre- cipitation products (at monthly scales) were evaluated over the entire continent during the period of 2003-2010. The TCH results are further compared to those of the classical evaluation using the Global Precipitation Climatology Center (GPCC) over entire Africa, as well as to the RG observations over the Greater Horn of Africa (GHA). Overall, for the study period (2003– 2010), the TCH results indicate that the RG-merged products contain smaller error amplitudes compared to the satellite-only products, consistent with the GPCC-based evaluation. A mul- tiple comparison procedure ranking, which was applied based on signal-to-noise ratios (SNR)s, indicated that PERSIANN contains the highest SNR and thus suitable over most of Africa, followed by ARCv2, TRMM, CMORPH, TAMSAT and GSMaP. To extract the main spatio- temporal variability of rainfall over Africa, complex empirical orthogonal function technique
One definition for data fusion is: “data fusion is a set of techniques used to generate fused images from a combination of primary images, by attempting to preserve the best characteristics of each primary image” (Mascarenhas et al., 1996). Similarly, a definition by Pohl and Genderen (1998) states that “image fusion is the combination of two or more different images to form a new image by using a certain algorithm.” Yocky (1996) refers to data fusion as “ image processing techniques that combine two image sets from two or more sensors, forming an enhanced final image.” Most of these definitions are image oriented. A more complex and broader definition is given by Wald (1999), stating that “data fusion is a formal framework in which are expressed means and tools for the alliance of data originating from different sources. It aims at obtaining information of greater quality; the exact definition of ‘greater quality’ will depend upon the application.” This definition is more suitable to the wider domain of data fusion (DFC, 2003).
The regression intersection method of minimizing the effect of the atmosphere is attractive to many analysts as it provides absolute results from the image data without the use of ancillary data. The method generally involves calculation of regression lines for a number of surface materials of contrasting spectral properties. The regression line method (RLM) determines a 'best fit' line for multispectral plots of pixels within homogenous cover types. Ideally, the intersection of lines must represent a point of zero ground reflectance since this is the only point at which radiometric values of two spectrally different materials can be safe. If no atmospheric scattering has taken place, the intersection of the line would be expected to pass through the origin. The slope of the plot is proportional to the ratio of the reflective material. However, the lines will, in reality, intersect the x and y axis producing two offset values. These brightness values represent the amount of bias caused by atmospheric scattering Crippen (1987) recommends the collection of a series of training areas resulting in many regression lines intersecting in two dimensional spaces at the same point using training sets to represent homogeneous land cover types. The relative values generated by regression method tend to be more reliable.
content; a high chlorophyll concentration will give lower reflectance value, and thus it is difficult to discriminate the mangrove species. On the other hand, others reported an opposite view, stating that the use of remotelysensed imagery data is easy in mangrove forest mapping (Giri, 2016) since the mangrove forest possesses a very distinct spectral signature. The general consensus seems to be that mangrove forest mapping is not straightforward with the remote sensing application. It may be based on the precision of the image, resolution, processing algorithm, or expertise in observing the data; in addition, it might be affected by the different location, as a different location has different vegetation composition and structure (Hossain & Nuruddin, 2016; Ghosh et al., 2016; Matsui et al., 2015; Heumann, 2011b; Adam et al., 2010). A recent trend in processing the satellite image for mangrove forest is to perform it using science knowledge and engineering technology. Over the past few decades, innovations in remote sensing sensors and systems such as very High Resolution System (VHR) and Synthetic Aperture Radar (SAR) (i.e., Quickbird, IKONOS, GeoEye-1, Worldview-3, PRISM-ALOS PALSAR, ASAR ENVISAT) and airborne sensors (i.e., hyperspectral remote sensing) are a breakthrough due to their high resolution sensor and continuous spectral data that are helpful in discriminating features having similar spectra in the multispectral domain (Rhyma et al., 2016; Prasad et al., 2014). In parallel with the advances of sensors and systems that are extensively applied in mangrove forest, analysis techniques in order to improve the accuracy of mangrove forest classification have also been developed such as object-based classifications integrated with one of these methods: pixel-based classification (Walter, 2004), decision tree learning analysis of pixel-based classification (Liu et al., 2008a), receiver operating characteristics (ROC) curve analysis of spectral analysis (Alatorre et al., 2011), threshold and fuzzy rule classification approaches with that of the pixel-based (Hussain et al. 2013), and support vector machine (SVM) approach of object-based classification (Liu et al., 2008b, Heumann (2011a), Vidhya et al., 2014). The following section reviews details mapped from the mangrove forest with a number of examples of image/data analysis techniques.