A refined intersensor comparison study is carried out to evaluate the radiometric stability of the 2257 nm channel (M11) of the first Visible Infrared Imaging Radiometer Suite (VIIRS) aboard the Suomi National Polar-orbiting Partnership (SNPP) satellite. This study is initiated as part of the examination into the performance of key shortwave infrared (SWIR) bands for SNPP VIIRS ocean color data processing and applications, with Band M11 playing key role over turbid and inland waters. The evaluation utilizes simultaneous nadir overpasses (SNOs) to compare SNPP VIIRS Band M11 against Band 7 of the MODerate-resolution Imaging Spectroradiometer (MODIS) in the Aqua satellite over concurrently observed scenes. The standard result of the radiance comparison is a seemingly uncontrolled and inconsistent time series unsuitable for further analyses, in great contrast to other matching band-pairs whose radiometric comparisons are typically stable around 1.0 within 1% variation. The mismatching relative spectral response (RSR) between the two respective bands, with SNPP VIIRS M11 at 2225 to 2275 nm and Aqua MODIS B7 at 2125 to 2175 nm, is demonstrated to be the cause of the large variation because of the different dependence of the spectral responses of the two bands over identical scenes. A consistent radiometric comparison time series, however, can be extracted from SNO events that occur over snowy surfaces. A customized selection and analysis procedure successfully identifies the snowy scenes within the SNO events and builds a stable comparison time series. Particularly instrumental for the success of the comparison is the use of the half-kilometer spatial resolution data of Aqua MODIS B7 that significantly enhances the statistics. The final refined time series of Aqua MODIS B7 radiance over the SNPP VIIRS M11 radiance is stable at around 0.39 within 2.5% showing no evidence of drift. The radiometric ratio near 0.39 suggests the strong presence of medium-grained snow of a mixed-snow condition in those SNO scenes leading to successful comparison. The multi-year stability indicates the correctness of the on-orbit RSB calibration of SNPP VIIRS M11 whose result does not suffer from long-term drift. Full article

The size of phytoplankton not only influences its physiology, metabolic rates and marine food web, but also serves as an indicator of phytoplankton functional roles in ecological and biogeochemical processes. Therefore, some algorithms have been developed to infer the synoptic distribution of phytoplankton cell size, denoted as phytoplankton size classes (PSCs), in surface ocean waters, by the means of remotely sensed variables. This study, using the NASA bio-Optical Marine Algorithm Data set (NOMAD) high performance liquid chromatography (HPLC) database, and satellite match-ups, aimed to compare the effectiveness of modeling techniques, including partial least square (PLS), artificial neural networks (ANN), support vector machine (SVM) and random forests (RF), and feature selection techniques, including genetic algorithm (GA), successive projection algorithm (SPA) and recursive feature elimination based on support vector machine (SVM-RFE), for inferring PSCs from remote sensing data. Results showed that: (1) SVM-RFE worked better in selecting sensitive features; (2) RF performed better than PLS, ANN and SVM in calibrating PSCs retrieval models; (3) machine learning techniques produced better performance than the chlorophyll-a based three-component method; (4) sea surface temperature, wind stress, and spectral curvature derived from the remote sensing reflectance at 490, 510, and 555 nm were among the most sensitive features to PSCs; and (5) the combination of SVM-RFE feature selection techniques and random forests regression was recommended for inferring PSCs. This study demonstrated the effectiveness of machine learning techniques in selecting sensitive features and calibrating models for PSCs estimations with remote sensing. Full article

During August 2016, a quasi-stationary high-pressure system spreading over Central and North-Eastern Europe, caused weather conditions that allowed for 24/7 observations of aerosol optical properties by using a complex multi-wavelength PollyXT lidar system with Raman, polarization and water vapour capabilities, based at the European Aerosol Research Lidar Network (EARLINET network) urban site in Warsaw, Poland. During 24–30 August 2016, the lidar-derived products (boundary layer height, aerosol optical depth, Ångström exponent, lidar ratio, depolarization ratio) were analysed in terms of air mass transport (HYSPLIT model), aerosol load (CAMS data) and type (NAAPS model) and confronted with active and passive remote sensing at the ground level (PolandAOD, AERONET, WIOS-AQ networks) and aboard satellites (SEVIRI, MODIS, CATS sensors). Optical properties for less than a day-old fresh biomass burning aerosol, advected into Warsaw’s boundary layer from over Ukraine, were compared with the properties of long-range transported 3–5 day-old aged biomass burning aerosol detected in the free troposphere over Warsaw. Analyses of temporal changes of aerosol properties within the boundary layer, revealed an increase of aerosol optical depth and Ångström exponent accompanied by an increase of surface PM₁₀ and PM_2.5. Intrusions of advected biomass burning particles into the urban boundary layer seem to affect not only the optical properties observed but also the top height of the boundary layer, by moderating its increase. Full article

Incoming surface solar irradiance (SSI) is essential for calculating Earth’s surface radiation budget and is a key parameter for terrestrial ecological modeling and climate change research. Remote sensing images from geostationary and polar-orbiting satellites provide an opportunity for SSI estimation through directly retrieving atmospheric and land-surface parameters. This paper presents a new scheme for estimating SSI from the visible and infrared channels of geostationary meteorological and polar-orbiting satellite data. Aerosol optical thickness and cloud microphysical parameters were retrieved from Geostationary Operational Environmental Satellite (GOES) system images by interpolating lookup tables of clear and cloudy skies, respectively. SSI was estimated using pre-calculated offline lookup tables with different atmospheric input data of clear and cloudy skies. The lookup tables were created via the comprehensive radiative transfer model, Santa Barbara Discrete Ordinate Radiative Transfer (SBDART), to balance computational efficiency and accuracy. The atmospheric attenuation effects considered in our approach were water vapor absorption and aerosol extinction for clear skies, while cloud parameters were the only atmospheric input for cloudy-sky SSI estimation. The approach was validated using one-year pyranometer measurements from seven stations in the SURFRAD (SURFace RADiation budget network). The results of the comparison for 2012 showed that the estimated SSI agreed with ground measurements with correlation coefficients of 0.94, 0.69, and 0.89 with a bias of 26.4 W/m², −5.9 W/m², and 14.9 W/m² for clear-sky, cloudy-sky, and all-sky conditions, respectively. The overall root mean square error (RMSE) of instantaneous SSI was 80.0 W/m² (16.8%), 127.6 W/m² (55.1%), and 99.5 W/m² (25.5%) for clear-sky, cloudy-sky (overcast sky and partly cloudy sky), and all-sky (clear-sky and cloudy-sky) conditions, respectively. A comparison with other state-of-the-art studies suggests that our proposed method can successfully estimate SSI with a maximum improvement of an RMSE of 24 W/m². The clear-sky SSI retrieval was sensitive to aerosol optical thickness, which was largely dependent on the diurnal surface reflectance accuracy. Uncertainty in the pre-defined horizontal visibility for ‘clearest sky’ will eventually lead to considerable SSI retrieval error. Compared to cloud effective radius, the retrieval error of cloud optical thickness was a primary factor that determined the SSI estimation accuracy for cloudy skies. Our proposed method can be used to estimate SSI for clear and one-layer cloud sky, but is not suitable for multi-layer clouds overlap conditions as a lower-level cloud cannot be detected by the optical sensor when a higher-level cloud has a higher optical thickness. Full article

Remote sensing (RS) scene classification is important for RS imagery semantic interpretation. Although tremendous strides have been made in RS scene classification, one of the remaining open challenges is recognizing RS scenes in low quality variance (e.g., various scales and noises). This paper proposes a deep salient feature based anti-noise transfer network (DSFATN) method that effectively enhances and explores the high-level features for RS scene classification in different scales and noise conditions. In DSFATN, a novel discriminative deep salient feature (DSF) is introduced by saliency-guided DSF extraction, which conducts a patch-based visual saliency (PBVS) algorithm using “visual attention” mechanisms to guide pre-trained CNNs for producing the discriminative high-level features. Then, an anti-noise network is proposed to learn and enhance the robust and anti-noise structure information of RS scene by directly propagating the label information to fully-connected layers. A joint loss is used to minimize the anti-noise network by integrating anti-noise constraint and a softmax classification loss. The proposed network architecture can be easily trained with a limited amount of training data. The experiments conducted on three different scale RS scene datasets show that the DSFATN method has achieved excellent performance and great robustness in different scales and noise conditions. It obtains classification accuracy of 98.25%, 98.46%, and 98.80%, respectively, on the UC Merced Land Use Dataset (UCM), the Google image dataset of SIRI-WHU, and the SAT-6 dataset, advancing the state-of-the-art substantially. Full article

This paper presents an analysis of the integration between interferometric and intensity-offset tracking-based SAR remote sensing for landslide hazard mitigation in the Italian Alps. Despite the advantages of Synthetic Aperture Radar Interferometry (InSAR) methods for quantifying landslide deformation, some limitations remain. The temporal decorrelation, the 1-D Line Of Sight (LOS) observation restriction, the high velocity rate and the multi-directional movement properties make it difficult to monitor accurately complex landslides in areas covered by vegetation. Therefore, complementary and integrated approaches, such as offset tracking-based techniques, are needed to overcome these InSAR limitations for monitoring ground surface deformations. As sub-pixel offset tracking is highly sensitive to data spatial resolution, the latest generations of SAR sensors, such as TerraSAR-X and COSMO-SkyMed, open interesting perspective for a more accurate hazard assessment. In this paper, we consider high-resolution X-band data acquired by the COSMO-SkyMed (CSK) constellation for Permanent Scatterers Interferometry (PSI), Multi-Aperture Interferometry (MAI) and offset tracking processing. We analyze the offset tracking techniques considering area and feature-based matching algorithms to evaluate their applicability to CSK data by improving sub-pixel offset estimations. To this end, PSI and MAI are used for extracting LOS and azimuthal displacement components. Then, four well-known area-based and five feature-based matching algorithms (taken from computer vision) are applied to 16 X-band corner reflectors. Results show that offset estimation accuracy can be considerably improved up to less than 3% of the pixel size using the combination of the different feature-based detectors and descriptors. A sensitivity analysis of these techniques applied to CSK data to monitor complex landslides in the Italian Alps provides indications on advantages and disadvantages of each of them. Full article

The identification and measurement of ground deformations in urban areas is of great importance for determining the vulnerable parts of the cities that are prone to geohazards, which is a crucial element of both sustainable urban planning and hazard mitigation. Interferometric synthetic aperture radar (InSAR) time series analysis is a very powerful tool for the operational mapping of ground deformation related to urban subsidence and landslide phenomena. With an analysis spanning almost 25 years of satellite radar observations, we compute an InSAR time series of data from multiple satellites (European Remote Sensing satellites ERS-1 and ERS-2, Envisat, Sentinel-1A, and its twin sensor Sentinel-1B) in order to investigate the spatial extent and rate of ground deformation in the megacity of Istanbul. By combining the various multi-track InSAR datasets (291 images in total) and analysing persistent scatterers (PS-InSAR), we present mean velocity maps of ground surface displacement in selected areas of Istanbul. We identify several sites along the terrestrial and coastal regions of Istanbul that underwent vertical ground subsidence at varying rates, from 5 ± 1.2 mm/yr to 15 ± 2.1 mm/yr. The results reveal that the most distinctive subsidence patterns are associated with both anthropogenic factors and relatively weak lithologies along the Haramirede valley in particular, where the observed subsidence is up to 10 ± 2 mm/yr. We show that subsidence has been occurring along the Ayamama river stream at a rate of up to 10 ± 1.8 mm/yr since 1992, and has also been slowing down over time following the restoration of the river and stream system. We also identify subsidence at a rate of 8 ± 1.2 mm/yr along the coastal region of Istanbul, which we associate with land reclamation, as well as a very localised subsidence at a rate of 15 ± 2.3 mm/yr starting in 2016 around one of the highest skyscrapers of Istanbul, which was built in 2010. Full article

Automatic building segmentation from aerial imagery is an important and challenging task because of the variety of backgrounds, building textures and imaging conditions. Currently, research using variant types of fully convolutional networks (FCNs) has largely improved the performance of this task. However, pursuing more accurate segmentation results is still critical for further applications such as automatic mapping. In this study, a multi-constraint fully convolutional network (MC–FCN) model is proposed to perform end-to-end building segmentation. Our MC–FCN model consists of a bottom-up/top-down fully convolutional architecture and multi-constraints that are computed between the binary cross entropy of prediction and the corresponding ground truth. Since more constraints are applied to optimize the parameters of the intermediate layers, the multi-scale feature representation of the model is further enhanced, and hence higher performance can be achieved. The experiments on a very-high-resolution aerial image dataset covering 18 km ${}^2$ and more than 17,000 buildings indicate that our method performs well in the building segmentation task. The proposed MC–FCN method significantly outperforms the classic FCN method and the adaptive boosting method using features extracted by the histogram of oriented gradients. Compared with the state-of-the-art U–Net model, MC–FCN gains 3.2% (0.833 vs. 0.807) and 2.2% (0.893 vs. 0.874) relative improvements of Jaccard index and kappa coefficient with the cost of only 1.8% increment of the model-training time. In addition, the sensitivity analysis demonstrates that constraints at different positions have inconsistent impact on the performance of the MC–FCN. Full article

IEA Wind Task 32 exists to identify and mitigate barriers to the adoption of lidar for wind energy applications. It leverages ongoing international research and development activities in academia and industry to investigate site assessment, power performance testing, controls and loads, and complex flows. Since its initiation in 2011, Task 32 has been responsible for several recommended practices and expert reports that have contributed to the adoption of ground-based, nacelle-based, and floating lidar by the wind industry. Future challenges include the development of lidar uncertainty models, best practices for data management, and developing community-based tools for data analysis, planning of lidar measurements and lidar configuration. This paper describes the barriers that Task 32 identified to the deployment of wind lidar in each of these application areas, and the steps that have been taken to confirm or mitigate the barriers. Task 32 will continue to be a meeting point for the international wind lidar community until at least 2020 and welcomes old and new participants. Full article

The discontinuous permafrost zone is one of the world’s most sensitive areas to climate change. Alaskan boreal forest is underlain by discontinuous permafrost, and wildfires are one of the most influential agents negatively impacting the condition of permafrost in the arctic region. Using interferometric synthetic aperture radar (InSAR) of Advanced Land Observation Satellite (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR) images, we mapped extensive permafrost degradation over interior Alaskan boreal forest in Yukon Flats, induced by the 2009 Big Creek wildfire. Our analyses showed that fire-induced permafrost degradation in the second post-fire thawing season contributed up to 20 cm of ground surface subsidence. We generated post-fire deformation time series and introduced a model that exploited the deformation time series to estimate fire-induced permafrost degradation and changes in active layer thickness. The model showed a wildfire-induced increase of up to 80 cm in active layer thickness in the second post-fire year due to pore-ice permafrost thawing. The model also showed up to 15 cm of permafrost degradation due to excess-ice thawing with little or no increase in active layer thickness. The uncertainties of the estimated change in active layer thickness and the thickness of thawed excess ice permafrost are 27.77 and 1.50 cm, respectively. Our results demonstrate that InSAR-derived deformation measurements along with physics models are capable of quantifying fire-induced permafrost degradation in Alaskan boreal forests underlain by discontinuous permafrost. Our results also have illustrated that fire-induced increase of active layer thickness and excess ice thawing contributed to ground surface subsidence. Full article

Pathogenic invasions are a major source of change in both agricultural and natural ecosystems. In forests, fungal pathogens can kill habitat-generating plant species such as canopy trees, but methods for remote detection, mapping and monitoring of such outbreaks are poorly developed. Two novel species of the fungal genus Ceratocystis have spread rapidly across humid and mesic forests of Hawaiʻi Island, causing widespread mortality of the keystone endemic canopy tree species, Metrosideros polymorpha (common name: ʻōhiʻa). The process, known as Rapid Ohia Death (ROD), causes browning of canopy leaves in weeks to months following infection by the pathogen. An operational mapping approach is needed to track the spread of the disease. We combined field studies of leaf spectroscopy with laboratory chemical studies and airborne remote sensing to develop a spectral signature for ROD. We found that close to 80% of ROD-infected plants undergo marked decreases in foliar concentrations of chlorophyll, water and non-structural carbohydrates, which collectively result in strong consistent changes in leaf spectral reflectance in the visible (400–700 nm) and shortwave-infrared (1300–2500 nm) wavelength regions. Leaf-level results were replicated at the canopy level using airborne laser-guided imaging spectroscopy, with quantitative spectral separability of normal green-leaf canopies from suspected ROD-infected brown-leaf canopies in the visible and shortwave-infrared spectrum. Our results provide the spectral–chemical basis for detection, mapping and monitoring of the spread of ROD in native Hawaiian forests. Full article

High-frequency surface wave radar (HFSWR) has been widely applied in sea-state monitoring, and its performance is known to suffer from various unwanted interferences and clutters. Radio frequency interference (RFI) from other radiating sources and ionospheric clutter dominate the various types of unwanted signals because the HF band is congested with many users and the ionosphere propagates interference from distant sources. In this paper, various orthogonal projection schemes are summarized, and three new schemes are proposed for interference cancellation. Simulations and field data recorded by experimental multi-frequency HFSWR from Wuhan University are used to evaluate the cancellation performances of these schemes with respect to both RFI and ionospheric clutter. The processing results may provide a guideline for identifying the appropriate orthogonal projection cancellation schemes in various HFSWR applications. Full article

The Arctic freshwater budget is critical for understanding the climate in the northern regions. However, the hydrology of the Arctic circumpolar tundra region (ACTR) and the largest pan-Arctic rivers are still not well understood. In this paper, we analyze the spatiotemporal variations in the terrestrial water storage (TWS) of the ACTR and three of the largest pan-Arctic river basins (Lena, Mackenzie, Yukon). To do this, we utilize monthly Gravity Recovery and Climate Experiment (GRACE) data from 2002 to 2016. Together with global land reanalysis, and river runoff data, we identify declining TWS trends throughout the ACTR that we attribute largely to increasing evapotranspiration driven by increasing summer air temperatures. In terms of regional changes, large and significant negative trends in TWS are observed mainly over the North American continent. At basin scale, we show that, in the Lena River basin, the autumnal TWS signal persists until the spring of the following year, while in the Mackenzie River basin, the TWS level in the autumn and winter has no significant impact on the following year. As expected global warming is expected to be particularly significant in the northern regions, our results are important for understanding future TWS trends, with possible further decline. Full article

The instrument Spectral Response Function (ISRF) has a strong impact on spectral calibration and the atmospheric trace gases retrievals. An accurate knowledge or a fine characterization of the ISRF shape and its FWHM (Full width at half maximum) as well as its temporal behavior is therefore crucial. Designing a strategy for the characterization of the ISRF both on ground and in-flight is critical for future missions, such as the spectral imagers in the Copernicus program. We developed an algorithm to retrieve the instrument ISRF in-flight. Our method uses solar measurements taken in-flight by the instrument to fit a parameterized ISRF from on ground based calibration, and then retrieves the shape and FWHM of the actual in-flight ISRF. With such a strategy, one would be able to derive and monitor the ISRF during the commissioning and operation of spectrometer imager missions. We applied our method to retrieve the SCIAMACHY instrument ISRF in its different channels. We compared the retrieved ones with the on ground estimated ones. Besides some peculiarities found in SCIAMACHY channel 8, the ISRF results in other channels were relatively consistent and stable over time in most cases. Full article

Major challenges for automatic ship detection in optical remote sensing (ORS) images include cloud, wave, island, wake clutters, and even the high variability of targets. This paper presents a practical ship detection scheme to resolve these existing issues. The scheme contains two main coarse-to-fine stages: prescreening and discrimination. In the prescreening stage, we construct a novel visual saliency detection method according to the difference of statistical characteristics between highly non-uniform regions which allude to regions of interest (ROIs) and homogeneous backgrounds. It can serve as a guide for locating candidate regions. In this way, not only can the targets be precisely detected, but false alarms are also significantly reduced. In the discrimination stage, to get a better representation of the target, both shape and texture features characterizing the ship target are extracted and concatenated as a feature vector for subsequent classification. Moreover, the combined feature is invariant to the rotation. Finally, a trainable Gaussian support vector machine (SVM) classifier is performed to validate real ships out of ship candidates. We demonstrate the superior performance of the proposed hierarchical detection method with detailed comparisons to existing efforts. Full article