ALBERT

Description: Although remote sensors have been increasingly providing dense data and deriving reanalysis data for inversion of particulate matters, the use of these data is considerably limited by the ground monitoring samples and conventional machine learning models. As regional criteria air pollutants, particulate matters present a strong spatial correlation of long range. Conventional machine learning cannot or can only model such spatial pattern in a limited way. Here, we propose a method of a geographic graph hybrid network to encode a spatial neighborhood feature to make robust estimation of coarse and fine particulate matters (PM10 and PM2.5). Based on Tobler’s First Law of Geography and graph convolutions, we constructed the architecture of a geographic graph hybrid network, in which full residual deep layers were connected with graph convolutions to reduce over-smoothing, subject to the PM10–PM2.5 relationship constraint. In the site-based independent test in mainland China (2015–2018), our method achieved much better generalization than typical state-of-the-art methods (improvement in R2: 8–78%, decrease in RMSE: 14–48%). This study shows that the proposed method can encode the neighborhood information and can make an important contribution to improvement in generalization and extrapolation of geo-features with strong spatial correlation, such as PM2.5 and PM10.

Description: Water depth estimation in seaports is essential for effective port management. This paper presents an empirical approach for water depth determination from satellite imagery through the integration of multiple datasets and machine learning algorithms. The implementation details of the proposed approach are provided and compared against different existing machine learning algorithms with a single training set. For a single training set and a single machine learning method, our analysis shows that the proposed depth estimation method provides a better root-mean-square error (RMSE) and a higher coefficient of determination (R2) under turbid water conditions, with overall RMSE and R2 improvements of 1 cm and 0.7, respectively. The developed method may be employed in monitoring dredging activities, especially in areas with polluted water, mud and/or a high sediment content.

Description: An ionospheric anomaly is the irregular change of the ionosphere. It may result in potential threats for the ground-based augmentation system (GBAS) supporting the high-level precision approach. To counter the hazardous anomalies caused by the steep gradient in ionospheric delays, customized monitors are equipped in GBAS architectures. A major challenge is to rapidly detect the ionospheric gradient anomaly from environmental noise to meet the safety-critical requirements. A one-class support vector machine (OCSVM)-based monitor is developed to clearly detect ionospheric anomalies and to improve the robust detection speed. An offline-online framework based on the OCSVM is proposed to extract useful information related to anomalous characteristics in the presence of noise. To validate the effectiveness of the proposed framework, the influence of noise is fully considered and analyzed based on synthetic, semi-simulated, and real data from a typical ionospheric anomaly event. Synthetic results show that the OCSVM-based monitor can identify the anomaly that cannot be detected by other commonly-used monitors, such as the CCD-1OF, CCD-2OF and KLD-1OF. Semi-simulation results show that compared with other monitors, the newly proposed monitor can improve the average detection speed by more than 40% and decrease the minimum detectable gradient change rate to 0.002 m/s. Furthermore, in the real ionospheric anomaly event experiment, compared with other monitors, the OCSVM-based monitor can improve the detection speed by 16%. The result indicates that the proposed monitor has encouraging potential to ensure integrity of the GBAS.

Description: Monitoring of land use, land-use changes, and forestry (LULUCF) plays a crucial role in biodiversity and global environmental challenges. In 2015, the Food and Agriculture Organization of the United Nations (FAO) launched the Global Forest Survey (GFS) integrating medium- (MR) and very-high-resolution (VHR) images through the FAO’s Collect Earth platform. More than 11,150 plots were inventoried in the Temperate FAO ecozone in Europe to monitor LULUCF from 2000 to 2015. As a result, 2.19% (VHR) to 2.77% (MR/VHR) of the study area underwent LULUCF, including a 0.37% (VHR) to 0.43% (MR/VHR) net increase in forest lands. Collect Earth and VHR images have also (i) allowed for shaping a preliminary structure of the land-use network, showing that cropland was the land type that changed most and that cropland and grassland were the more frequent land uses that generated new forest land, (ii) shown that, in 2015, mixed and monospecific forests represented 44.3% and 46.5% of the forest land, respectively, unlike other forest sources, and (iii) shown that 14.9% of the area had been affected by disturbances, particularly wood harvesting (67.47% of the disturbed forests). According to other authors, the area showed a strong correlation between canopy mortality and reported wood removals due to the transition from past clear-cut systems to “close-to-nature” silviculture.

Description: In hyperspectral image (HSI) classification, convolutional neural networks (CNN) have been attracting increasing attention because of their ability to represent spectral-spatial features. Nevertheless, the conventional CNN models perform convolution operation on regular-grid image regions with a fixed kernel size and as a result, they neglect the inherent relation between HSI data. In recent years, graph convolutional networks (GCN) used for data representation in a non-Euclidean space, have been successfully applied to HSI classification. However, conventional GCN methods suffer from a huge computational cost since they construct the adjacency matrix between all HSI pixels, and they ignore the local spatial context information of hyperspectral images. To alleviate these shortcomings, we propose a novel method termed spectral-spatial offset graph convolutional networks (SSOGCN). Different from the usually used GCN models that compute the adjacency matrix between all pixels, we construct an adjacency matrix only using pixels within a patch, which contains rich local spatial context information, while reducing the computation cost and memory consumption of the adjacency matrix. Moreover, to emphasize important local spatial information, an offset graph convolution module is proposed to extract more robust features and improve the classification performance. Comprehensive experiments are carried out on three representative benchmark data sets, and the experimental results effectively certify that the proposed SSOGCN method has more advantages than the recent state-of-the-art (SOTA) methods.

Description: Moored upward-looking Acoustic Doppler Current Profilers (ADCPs) can be used to observe sea ice draft. While previous studies relied on the availability of auxiliary pressure sensors to measure the instrument depth of the ADCP, we present an adaptive approach that infers instrument depth from ADCP bottom track (BT) mode measurements of error velocity and range. The ADCP-derived ice draft time series are validated with data from adjacent Upward-Looking Sonar (ULS) moorings. We demonstrate that this method can be used to obtain daily mean sea ice draft time series that, on average, are within 20% of ULS-derived draft time series. ULS and ADCP ice draft time series were observed by four moorings in the Laptev Sea and show correlations between 0.7 and 0.9. This new approach is not a substitute for high-frequency, high-precision ULS measurements of ice draft but it provides a low-cost opportunity to derive daily mean ice draft time series accessing existing ADCP data that have not been not used for that purpose to date. This method has the potential to close data gaps and extend existing ice draft time series in all ice-covered regions and supports the validation of sea ice thickness products from satellite missions such as CryoSat-2, SMOS or ENVISAT.

Description: Yilan Bay is in the northeast corner of Taiwan at the junction of the East China Sea (ECS) and the Pacific Ocean. This study clarified the composition of water masses adjacent to Yilan Bay. The upper seawater in the bay is characterized by Kuroshio surface water, Taiwan warm current water, and shelf mixed water masses. The flow field in this area is mainly determined by the inter-actions among the northeastern Taiwan countercurrent, Kuroshio Current (KC), and tidal currents. The fall season is the main rainfall period in Yilan Bay, which causes a large amount of river runoff and a further increase in chlorophyll concentration, and the salinity of the upper water layer is observed much lower than other seasons. Water with a high chlorophyll concentration can flow into the ECS with ebb currents and the KC with ebb and flood currents. Combining hourly geosynchronous ocean color imager data and numerical simulation flow field helps us understand short-term changes of chlorophyll concentration. The trajectories of the drifters and virtual particle simulations help us understand the sources and movement of ocean currents in Yilan Bay. The seasonal swing of the KC path outside the bay is an important factor affecting the flow field and hydrological characteristics.

Description: The Fabry–Pérot interferometer (FPI) and meteor radar are two important techniques for measuring the horizontal wind field in mesopause region, the observations of which still lack comprehensive comparison. Kunming Observatory (25.6°N,103.8°E) has deployed both instruments in recent years and provides collocated meteor radar and FPI observations. The meteor radar measures the horizontal wind fields over 24 hours every day continuously, whereas the FPI can only work during the night with clear air condition. FPI horizontal wind data from the 892.0-nm airglow emission (with a peak height at ~87 km) from 26 January to 8 February 2019 were comparatively analyzed with simultaneous meteor radar observations, which cover the range between 80 and 90 km with a vertical resolution of 1.8 km. It was found that the temporal variations in the horizontal wind fields observed by the FPI and meteor radar were generally consistent with one another, with the highest 2-D correlation coefficients of 0.91 (0.88) at 88 (87) km for the meridional (zonal) wind, which agreed with the peak height of OH airglow emission observed by the TIMED/SABER instrument. In addition, the correlation coefficient for the weighted meteor radar horizontal wind by OH concentration between 86 and 88 km and 85 and 89 km increased slightly from 0.91 (0.89) to 0.92 (0.89) for the meridional (zonal) wind, which indicated the contribution of OH concentration beyond the peak height to the FPI wind observations. We also found that the absolute horizontal wind values detected by two instruments were linearly correlated with a slope of ~1.3 for both wind components, and meteor radar wind observations were usually larger than the FPI observations.

Description: This study investigated monthly variations of surface urban heat island intensity (SUHII) and the applicability of the local climate zones (LCZ) scheme for land surface temperature (LST) differentiation within three spatial contexts, including urban, rural and their combination, in Shenyang, China, a city with a monsoon-influenced humid continental climate. The monthly SUHII and LST of Shenyang were obtained through 12 LST images, with one in each month (within the period between 2018 and 2020), retrieved from the Thermal InfraRed Sensor (TIRS) 10 in Landsat 8 based on a split window algorithm. Non-parametric analysis of Kruskal-Wallis H test and a multiple pairwise comparison were adopted to investigate the monthly LST differentiations with LCZs. Overall, the SUHII and the applicability of the LCZ scheme exhibited spatiotemporal variations. July and August were the two months when Shenyang underwent strong heat island effects. Shenyang underwent a longer period of cool than heat island effects, occurring from November to May. June and October were the transition months of cool–heat and heat–cool island phenomena, respectively. The SUHII analysis was dependent on the definition of urban and rural boundaries, where a smaller rural buffering zone resulted in a weaker SUHI or surface urban cool island (SUCI) phenomenon and a larger urban area corresponded to a weaker SUHI or SUCI phenomenon as well. The LST of LCZs did not follow a fixed order, where in July and August, the LCZ-10 (Heavy industry) had the highest mean LST, followed by LCZ-2 (Compact midrise) and then LCZ-7 (Lightweight low-rise). In comparison, LCZ-7, LCZ-8 (Large low-rise) and LCZ-9 (Sparsely built) had the highest LST from October to May. The LST of LCZs varied with urban and rural contexts, where LCZ-7, LCZ-8 and LCZ -10 were the three built LCZs that had the highest LST within urban context, while LCZ-2, LCZ-3 (Compact low-rise), LCZ-8, LCZ-9 and LCZ-10 were the five built LCZs that had the highest LST within rural context. The suitability of the LCZ scheme for temperature differentiation varied with the month, where from July to October, the LCZ scheme had the strongest capability and in May, it had the weakest capability. Urban context also made a difference to the suitability, where compared with the whole study area (the combination of urban and rural areas), the suitability of built LCZs in either urban or rural contexts weakened. Moreover, the built LCZs had a higher level of suitability in an urban context compared with a rural context, while the land-cover LCZs within rural had a higher level of suitability.

Description: Air temperature (Ta), defined as the temperature 2 m above the land’s surface, is one of the most important factors for environment and climate studies. Ta can be measured by obtaining the land surface temperature (LST) which can be retrieved with the 11- and 12-µm bands from satellite imagery over a large area, and LST is highly correlated with Ta. To measure the Ta in a broad area, we studied a Ta retrieval method through Deep Neural Network (DNN) using in-situ data and satellite data of South Korea from 2014 to 2017. To retrieve accurate Ta, we selected proper input variables and conditions of a DNN model. As a result, Normalized Difference Vegetation Index, Normalized Difference Water Index, and 11- and 12-µm band data were applied to the DNN model as input variables. And we also selected proper condition of the DNN model with test various conditions of the model. In validation result in the DNN model, the best accuracy of the retrieved Ta showed an correlation coefficient value of 0.98 and a root mean square error (RMSE) of 2.19 K. And then we additional 3 analysis to validate accuracy which are spatial representativeness, seasonal analysis and time series analysis. We tested the spatial representativeness of the retrieved Ta. Results for window sizes less than 132 × 132 showed high accuracy, with a correlation coefficient of over 0.97 and a RMSE of 1.96 K and a bias of −0.00856 K. And in seasonal analysis, the spring season showed the lowest accuracy, 2.82 K RMSE value, other seasons showed high accuracy under 2K RMSE value. We also analyzed a time series of six the Automated Synoptic Observing System (ASOS) points (i.e., locations) using data obtained from 2018 to 2019; all of the individual correlation coefficient values were over 0.97 and the RMSE values were under 2.41 K. With these analysis, we confirm accuracy of the DNN model was higher than previous studies. And we thought the retrieved Ta can be used in other studies or climate model to conduct urban problems like urban heat islands and to analyze effects of arctic oscillation.