ABSTRACT Recent growth in unmanned aerial vehicle (UAV) technology have promoted the detailed mapping of individual tree species. However, the in-depth mining and comprehending of the significance of features derived from high-resolution UAV data for tree species discrimination remains a difficult task. In this study, a state-of-the-art approach combining UAV-borne light detection and ranging (LiDAR) and hyperspectral was used to classify 11 common tree species in a typical natural secondary forest in Northeast China. First, comprehensive relevant structural and spectral features were extracted. Then, the most valuable feature sets were selected by using a hybrid approach combining correlation-based feature selection with the optimized recursive feature elimination algorithm. The random forest algorithm was used to assess feature importance and perform the classification. Finally, the robustness of features derived from point clouds with different structures and hyperspectral images with different spatial resolutions was tested. Our results showed that the best classification accuracy was obtained by combining LiDAR and hyperspectral data (75.7%) compared to that based on LiDAR (60.0%) and hyperspectral (64.8%) data alone. The mean intensity of single returns and the visible atmospherically resistant index for red-edge band were the most influential LiDAR and hyperspectral derived features, respectively. The selected features were robust in point clouds with a density not lower than 5% (~5 pts/m2) and a resolution not lower than 0.3 m in hyperspectral data. Although canopy surface features were slightly different from original LiDAR features, canopy surface information was also important for tree species classification. This study proved the capabilities of UAV-borne LiDAR and hyperspectral data in natural secondary forest tree species discrimination and the potential for this approach to be transferable to other study areas.

ABSTRACT Forty percent of global food production relies upon irrigation, which accounts for 70% of total global freshwater use. Thus, the mapping of cropland irrigation plays a significant role in agricultural water management and estimating food production. However, current spaceborne irrigated cropland mapping is highly reliant upon its spectral behavior, which often has high uncertainty and lacks information about the method of irrigation. Deep learning (DL) allows for the classification of irrigated cropland according to unique spatial patterns, such as the central pivot irrigation system (CPIS). But convolutional neural networks (CNNs) are usually biased toward color and texture features, a spatial transferable and accurate CPIS identification model is lacking owing to previous model seldom involves the round shapes of CPIS, which is usually key to distinguishing CPIS. To address this lack, we proposed a shape attention neural network by integrating spatial-attention gate, residual block, and multi-task learning, Pivot-Net, to incorporate shape information identify CPIS in satellite imagery. Specifically, we employed CPIS in Kansas to train our model using Sentinel-2 and Landsat-8 optical images. We found that Pivot-Net is superior to seven state-of-the-art semantic segmentation models on second-stage validation. We also evaluated the performance of Pivot-Net at three validation sites, which had an average F1-score and mean IOU of 90.68% and 90.45%, respectively, which further demonstrated the high accuracy of the proposed model. Moreover, to show that the proposed Pivot-Net can map CPIS at the country scale, we generated the first CPIS map at 30 m for the contiguous US using a cloud computing platform and our Pivot-Net model. The total CPIS area for the contiguous US was 61,094 km2 in 2018, which comprised 26.22% of all irrigated lands. Our results can be accessed at https://tianfyou.users.earthengine.app/view/cpisus. Therefore, the proposed shape-attention Pivot-Net demonstrates the ability to classify CPIS at large spatial scales and are feasible to map CPIS at national scales.

ABSTRACT Leaf functional traits are key indicators of plant functions useful for inferring complex plant processes, including their responses to environmental changes. Vegetation indices (VIs) composed of a few reflectance wavelengths hold the advantages of being relatively simple and effective and have been widely used within remote sensing to estimate leaf traits. However, the difference between the reflectance from the upper and lower part of the leaf suggests that leaf reflectance mainly provides one-sided information, constraining its ability to estimate leaf functional traits. Leaf transmittance, on the other hand, gives information about the whole leaf and has more potential to be sensitive to changes in leaf biochemistry. As transmittance-based VI is rare, this study aims to propose new transmittance-based VIs for accurate estimations of leaf traits. Three forms, i.e. the normalized difference VI, the simple ratio VI, and the difference VI were employed, and wavelength selection for transmittance-based and reflectance-based VIs were conducted, respectively. The applicability of these VIs for estimating four leaf functional traits (leaf chlorophyll (Cab), leaf carotenoids (Car), equivalent water thickness (EWT), and leaf mass per area (LMA)) were evaluated. Cross-validation using three datasets of field observations and sensitivity analysis showed that the VIs constructed using transmittance were relatively less affected by interferences from other leaf parameters, improving the estimation accuracy of Car, EWT, and LMA compared to their optimal reflectance counterparts (RMSE reduced by 2% to 15%, and MAE reduced by 7% to 20% for the pooled dataset). Our study revealed that the normalized difference VI based on transmittance showed considerable sensitivity to Car, EWT, and LMA, whereas the difference VI based on reflectance was effective in indicating Cab. The proposed transmittance-based VIs will aid remote monitoring of leaf traits and thereby plant adaptations and acclimation to changes in environmental conditions.

ABSTRACT Rapeseed mapping is important for national food security and government regulation of land use. Various methods, including empirical index-based and machine learning-based methods, have been developed to identify rapeseed using remote sensing. Empirical index-based methods commonly employed empirically designed indices to enhance rapeseed’s bright yellow spectral feature during the flowering period, which is straightforward to implement. Unfortunately, the heavy cloud cover in the flowering period of China would lead to serious omission errors; and the required flowering period varies spatially and yearly, which often cannot be acquired accurately. Machine learning-based methods mitigate the reliance on clear observations during the flowering period by inputting all-season imagery to adaptively learn features. However, it is difficult to collect sufficient samples across all of China considering the large intraclass variation in both land cover types of rapeseed and non-rapeseed. This study proposed an automated rapeseed mapping approach integrating rule-based sample generation and a one-class classifier (RSG-OC) to overcome the shortcomings of the two types of methods. First, a set of sample selection rules based on empirical indices of rapeseed were developed to automatically generate samples in cloud-free pixels during the predicted flowering period throughout China. Second, all available features composited based on the rapeseed phenological calendar were used for classification to eliminate the phenology differences in different regions. Third, a specific sample augmentation that removes the observation in the flowering period was employed to improve the generalization to the pixels without cloud-free observation in the flowering period. Finally, to avoid the need for diverse samples of nonrapeseed classes, a typical one-class classifier, positive unlabeled learning implemented by random forest (PUL-RF) trained by the generated samples, was applied to map rapeseed. With the proposed method, China rapeseed was mapped at 20 m resolution during 2017–2021 based on the Google Earth Engine (GEE). Validation on six typical rapeseed planting areas demonstrates that RSG-OC achieves an average accuracy of 94.90%. In comparison, the average accuracy of the other methods ranged from 83.33% to 88.25%, all of which were poorer than the proposed method. Additional experiments show that the performance of RSG-OC was not sensitive to cloud contamination, inaccurate predicted flowering time and the threshold of sample selection rule. These results indicate that the rapeseed maps produced in China are overall reliable and that the proposed method is an effective and robust method for annual rapeseed mapping across China.

ABSTRACT Understory vegetation contributes considerably to biodiversity and total aboveground biomass of forest ecosystems. Whereas field inventories and LiDAR data are generally used to estimate understory vegetation density, methods for large-scale and spatially continuous estimation of understory vegetation density are still lacking. For an evergreen coniferous forest area in southern China, we developed and tested an effective and practical remote sensing-driven approach for mapping understory vegetation, based on phenological differences between over and understory vegetation. Specifically, we used plant area volume density (PAVD) calculations based on GEDI data to train a support vector regression model and subsequently estimated the understory vegetation density from Sentinel-2 derived metrics. We produced maps of PAVD for the growing and non-growing season respectively, both performing well compared against independent GEDI samples (R2 = 0.89 and 0.93, p < 0.01). Understory vegetation density was derived from the differences in PAVD between the growing and non-growing season. The understory vegetation density map was validated against field samples from 86 plots showing an overall R2 of 0.52 (p < 0.01), rRMSE = 21%. Our study developed a tangible approach to map spatially continuous understory vegetation density with the combination of GEDI LiDAR data and Sentinel-2 imagery, showing the potential to improve the estimation of terrestrial carbon storage and better understand forest ecosystem processes across larger areas.

ABSTRACT Satellite remote sensing is a highly valuable data source useful in the monitoring of surface water dynamics and an essential tool in flood risk management although several factors can interfere with the detection of water features. This study explores an approach that integrates satellite optical images and DEM-based hydrogeomorphic features to enhance real‐time identification of river flooding. Spectral indices, namely MNDWI and the combination of NDWI/NDTI, were taken from Sentinel-2 images and used to detect flooded areas, while DEMs were used to identify areas more exposed to river flooding using the Geomorphic Flood Index. The integration/overlapping of these layers allowed us to refine the final map with a significant detection improvement compared to those based exclusively on the segmentation of a spectral index. The proposed integrated approach was tested in five areas of the Piave river (Northern Italy) where a flooding event caused by the Vaia storm took place in October 2018. Results have been evaluated against reference maps produced manually in five selected areas, demonstrating good capabilities of the proposed procedure (mean accuracy 0.90), significantly reducing false alarms (Matthews Correlation Coefficient improvements among 0.02–0.43 in 4 out of 5 areas). The key advantages of such a procedure are the rapid application and the free and easy availability of the data that, in turn, allow for the investigation of large areas in the aftermath of an event.

ABSTRACT Building extraction is significant in urban planning, economic evaluation, and driverless technology development. However, automatic building extraction from high spatial resolution remote sensing images has been a challenging task due to the various building shapes and colors, imaging conditions, and complex background objects. Current methods in building extraction are generally based on deep convolution networks, and they mostly use an encoder-decoder architecture, wherein detailed building features and small buildings are easily omitted in continuous convolution operations. Moreover, buildings with blurred boundaries are only completely extracted with difficulty. To meet these challenges, we propose a multi-task architecture of frequency-spatial learning Transformer to extract buildings from high spatial resolution remote sensing images. Different from current architecture, we designed a frequency-spatial learning module in the framework of multi-task to synthesize the multi-scale spatial features and frequency decomposition features of high-resolution image. Spiking convolution is proposed in this study to enhance the frequency features of buildings by mimicking the neural transmission in human brains. In this way, multi-scale building features can be better preserved and distinguished from background objects. Moreover, a masked-attention Transformer is adopted to improve multi-scale building mask prediction accuracy by synthesizing successive pixel-wise up-sampled feature maps. We also propose a strategy to evaluate the practical transferability of the proposed method by mimicking practical application cases through training and evaluating images with different spatial resolutions from different study areas and datasets. Experiments using five public building datasets (WHU-Building Satellite Dataset I, WHU-Building Satellite Dataset II, Massachusetts Buildings Dataset, Inria Aerial Image Dataset, xBD Building Dataset) demonstrate the strong potential applicability of our proposed method for practical application cases. Our method outperforms five recently proposed state-of-the-art semantic segmentation methods with 36.60% accuracy improvement on extracted buildings and approximately 53.55% recall progress in extracting small building instances. The implementation code will be released after the paper is published.

ABSTRACT Accurate and frequent mapping of transient wetland inundation in the boreal region is critical for monitoring the ecological and societal functions of wetlands. Satellite Synthetic Aperture Radar (SAR) has long been used to map wetlands due to its sensitivity to surface inundation and ability to penetrate clouds, darkness, and certain vegetation canopies. Here, we track boreal wetland inundation by developing a two-step modified decision-tree algorithm implemented in Google Earth Engine using Sentinel-1 C-band SAR and Sentinel-2 Multispectral Instrument (MSI) time-series data as inputs. This approach incorporates temporal as well as spatial characteristics of SAR backscatter and is evaluated for the Peace-Athabasca Delta, Alberta (PAD), and Yukon Flats, Alaska (YF) from May 2017 to October 2019. Within these two boreal study areas, we map spatiotemporal patterns in wetland inundation classes of Open Water (OW), Floating Plants (FP), Emergent Plants (EP), and Flooded Vegetation (FV). Temporal variability, frequency, and maximum extents of transient wetland inundation are quantified. Retrieved inundation estimates are compared with in-situ field mapping obtained during the NASA Arctic-Boreal Vulnerability Experiment (ABoVE), and a multi-temporal Landsat-derived surface water map. Over the 2017–2019 study period, we find that fractional inundation area ranged from 18.0% to 19.0% in the PAD, and from 10.7% to 12.1% in the YF. Transient wetland inundation covered ~595 km2 of the PAD, comprising ~9.1% of its landscape, and ~102 km2 of the YF, comprising ~3.6%. The implications of these findings for wetland function monitoring, and estimating landscape-scale methane emissions are discussed, together with limitations and uncertainties of our approach. We conclude that time series of Sentinel-1 C-band SAR backscatter, screened with Sentinel-2 MSI optical imagery and validated by field measurements, offer a valuable tool for tracking transient boreal wetland inundation.

ABSTRACT Accurately monitoring the variation of snow cover from remote sensing is vital since it assists in various fields including prediction of floods, control of runoff values, and the ice regime of rivers. Spectral indices methods are traditional ways to realize snow segmentation, including the most common one – the Normalized Difference Snow Index (NDSI), which utilizes the combination of green and short-wave infrared (SWIR) bands. In addition, spectral indices methods heavily depend on the optimal threshold to determine the accuracy, making it time-consuming to find optimal values for different places. Convolutional neural networks ensemble model with DeepLabV3+ was employed as sub-models for snow segmentation using (Sentinel-2), which aims to distinguish clouds and water body from snow. The imagery dataset generated in this article contains sites in global alpine regions such as Tibetan Plateau in China, the Alps in Switzerland, Alaska in the United States, Southern Patagonian Icefield in Chile, Tsylos Provincial Park, Tatsamenie Peak, and Dalton Peak in Canada. To overcome the limitation of DeepLabV3+, which only accepts three channels as input features, and the need to use six features: green, red, blue, near-infraRed, SWIR, and NDSI, 20 three-channel DeepLabV3+ sub-models, were constructed with different combinations of three features and then ensembled together. The proposed ensemble model showed superior performance than benchmark spectral indices method, with mIoU values ranging from 0.8075 to 0.9538 in different test sites. The results of this project contribute to the development of automated snow segmentation tools to assist earth observation applications.

ABSTRACT Surface urban heat island (SUHI) can considerably influence the urban environment and the quality of life. It is vital to examine how underlying surface properties impact seasonal and diurnal SUHIs. However, the influence of three-dimensional (3D) urban morphological parameters (UMPs) on SUHIs has not been thoroughly studied under varying climatic settings. To fill this knowledge gap, the present study investigated seasonal and diurnal changes in SUHI intensities (ΔT) in 208 cities in China from 2014 to 2016 using moderate-resolution imaging spectroradiometer (MODIS) land surface temperature products. In addition, the influence of potential factors in urban surface energy balance, including two-dimensional (2D) and 3D UMPs, socio-economic indices, urban greening, and surface albedo, on seasonal and diurnal ΔT were assessed under different climatic settings and with different city sizes using the method of Geographically Weighted Regression (GWR). Results show that negative summer daytime ΔT was observed in some cities under dry climates. Generally, in summer, the ΔT during daytime was higher than at nighttime. The 3D UMPs (i.e. building height and volume) yielded more decisive influences on ΔT than 2D UMP (i.e. building coverage). This is particularly true for the summer diurnal cycle and under dry climatic settings. Building height was found to be negatively correlated with surface temperatures, while building volume was positively correlated. Additionally, the 3D UMPs yielded more influences on winter ΔT than summer ΔT. The capability of vegetation to regulate ΔT was more potent in dry climates than in wet climates and in small cities than in large cities. Varying climates and city sizes can modify the significance of the 2D and 3D UMPs on the urban surface energy balance, suggesting that urban thermal mitigation should consider climate background and population size.