Multispectral Image Data Research Articles

Parallel multispectral ghost imaging data acquisition with supercontinuum

A wavelength multiplexing with spectral supercontinuum source in computational ghost imaging is implemented and investigated. In our proof-of-concept experiment, we used two diffraction gratings and one spatial light modulator to form up to 10 independent spectral channels. Registration of data by a spectrometer in such a scheme makes it possible to reduce the time of data acquisition by a factor of ten. It was established that an increase the number of spectral channels from 1 to 10 leads to the increase of the contrast-to-noise ratio by the factor of 2.8. This makes the proposed technique attractive for high-speed demanding applications such as communications and remote sensing.

A Generic Framework for Enhancing Autonomous Driving Accuracy through Multimodal Data Fusion

Higher-level autonomous driving necessitates the best possible execution of important moves under all conditions. Most of the accidents in recent years caused by the AVs launched by leading automobile manufacturers are due to inadequate decision-making, which is a result of their poor perceivance of environmental information. In today’s technology-bound scenarios, versatile sensors are used by AVs to collect environmental information. Due to various technical and natural calamities, the environmental information acquired by the sensors may not be complete and clear, due to which the AVs may misinterpret the information in a different context, leading to inadequate decision-making, which may then lead to fatal accidents. To overcome this drawback, effective preprocessing of raw sensory data is a mandatory task. Pre-processing the sensory data involves two vital tasks, namely data cleaning and data fusion. Since the raw sensory data are complex and exhibit multimodal characteristics, more emphasis is given to data preprocessing. Since more innovative models have been proposed for data cleaning, this study focused on data fusion. In particular, this study proposed a generic data fusion engine, which classifies different formats of sensory data and fuses them accordingly to improve accuracy. This study proposed a generic framework to fuse the text, image, and audio data. In the first stage of this research, an innovative hybrid model was proposed to fuse multispectral image and video data. Simple and efficient models to extract the salient image features were also proposed. The hybrid image fusion model that was proposed did not yield satisfactory outcomes when combining 3D point cloud data, and its performance declined when evaluating large datasets. To address this issue, the study expanded by introducing an advanced generative adversarial network (GAN) to transform the hybrid image fusion model into a machine learning model capable of handling substantial datasets. Additionally, customized kernel functions were suggested to fuse 3D point cloud data effectively. The performance of the proposed models was assessed using standard metrics and datasets, comparing them with existing popular models. The results revealed that the proposed image fusion model outperformed the other models.

Water color from Sentinel-2 MSI data for monitoring large rivers: Yangtze and Danube

ABSTRACT Rivers provide key ecosystem services that are inherently engineered and optimized to meet the strategic and economic needs of countries around the world. However, limited water quality records of a full river continuum hindered the understanding of how river systems response to the multiple stressors acting on them. This study highlights the use of Sentinel-2 Multi-Spectral Imager (MSI) data to monitor changes in water color in two optically complex river systems: the Yangtze and Danube using the Forel-Ule Index (FUI). FUI divides water color into 21 classes from dark blue to yellowish brown stemming from the historical Forel-Ule water color scale and has been promoted as a useful indicator showing water turbidity variations in water bodies. The results revealed contrasting water color patterns in the two rivers on both spatial and seasonal scales. Spatially, the FUI of the Yangtze River gradually increased from the upper reaches to the lower reaches, while the FUI of the Danube River declined in the lower reaches, which is possibly due to the sediment sink effect of the Iron Gate Dams. The regional FUI peaks and valleys observed in the two river systems have also been shown to be related to the dams and hydropower stations along them. Seasonally, the variations of FUI in both systems can be attributed to climate seasonality, especially precipitation in the basin and the water level. Moreover, land cover within the river basin was possibly a significant determinant of water color, as higher levels of vegetation in the Danube basin were associated with lower FUI values, whereas higher FUI values and lower levels of vegetation were observed in the Yangtze system. This study furthers our knowledge of using Sentinel-2 MSI to monitor and understand the spatial-temporal variations of river systems and highlights the capabilities of the FUI in an optically complex environment.

Detection of aerosol and cloud features for the EarthCARE atmospheric lidar (ATLID): the ATLID FeatureMask (A-FM) product

Abstract. The EarthCARE satellite mission's objective is to retrieve profiles of aerosol and cloud physical and optical properties using the combination of cloud-profiling radar (CPR), high-spectral-resolution UV lidar (ATLID) and passive multi-spectral imager (MSI) data. Based on synergistic retrievals using data from these instruments, the 3D atmospheric cloud–aerosol state is estimated and then used to model the top-of-atmosphere (TOA) broadband radiances, which may then be compared to co-incident EarthCARE broadband radiometer (BBR) measurements. A high-spectral-resolution lidar enables the independent retrieval of extinction and backscatter but, being space based, suffers from relatively low signal-to-noise ratio (SNR) levels. The ATLID FeatureMask (A-FM) product provides a feature detection mask for the existence of atmospheric features within the lidar profiles based on a number of (statistical) image reconstruction techniques. Next to this, it also identifies those regions where the lidar beam has been fully attenuated and where the surface backscatter has impacted the measured lidar backscatter signals directly above the surface. From the pixels assigned as clear sky (with no features present above), the clear-sky-averaged profiles for the three ATLID channels, the co-polar Mie channel, the total cross channel and the co-polar Rayleigh channel are created. These feature-free or clear-sky profiles are useful for e.g., assessing the quality of the ATLID Level-1 (L1) attenuated backscatters. An important goal of the A-FM product is to guide smoothing strategies within downstream processors e.g., the ATLID profile retrieval (A-PRO) algorithm which directly follows A-FM within the EarthCARE Level-2 (L2) processing chain. Within the A-PRO algorithm, profiles of extinction, backscatter and linear depolarization ratio are retrieved. However, smoothing of the ATLID L1 attenuated backscatter is necessary since the SNR levels present at the ATLID native resolution are generally not sufficient for meaningful retrievals to be conducted. At the same time, to prevent biased retrievals, any smoothing procedure must respect the cloud–aerosol structure and avoid mixing strong features, e.g., clouds, and weak features, e.g., aerosol regions, together. The A-FM product provides the A-PRO algorithm with important information that is used to guide various smoothing procedures. To enable the processing of the large datasets from observation up to L2 retrievals, each EarthCARE orbit is separated into eight frames, divided at latitudes of 22.5∘ N and 22.5∘ S and 62.5∘ N and 62.5∘ S. As a secondary product, A-FM outputs can be used to conduct a frame-by-frame evaluation of the ATLID L1 cross-talk calibration, where an EarthCARE frame is one-eighth of a full orbit. This evaluation can be performed by comparing the retrieved clear-sky profiles to the expected channel profiles. The A-FM product has been applied to both synthetic data from the EarthCARE end-to-end simulator (ECSIM) and the L1 data from the Aeolus wind lidar mission. Comparisons against the ECSIM model truth indicate that A-FM has a percentage correctness > 90 % and is capable of reliably detecting aerosol and cloud regions within extinctions (> 10−5 m−1).

Mapping river turbidity at a large basin-scale with a spatially transferable ensemble model using Landsat 8 multispectral imagery

ABSTRACT Mapping river water quality at a basin scale with satellite image data represents a tremendous technical challenge, since traditional empirical remote sensing models calibrated for specific reaches of a river cannot be applied to all the rivers and streams in the basin with an adequate prediction accuracy. In this study, we developed a multi-predictor ensemble model to map and assess water turbidity for all rivers and streams in the Tombigbee River Basin with Landsat 8 multispectral images. Coincident with Landsat 8 satellite overpass on 2 May 2019, we conducted boat-based water quality data collection in the confluence area of Tombigbee River and Black Warrior River near the city of Demopolis, Alabama. Based on the coincident satellite image and field water samples, we calibrated and evaluated the multi-predictor ensemble model, which consists four component empirical models. Our evaluations suggest that the multi-predictor ensemble model can improve turbidity estimation accuracy by 79% in comparison with the best traditional empirical model. More importantly, our multi-predictor ensemble model can be applied to other river and stream reaches across three major sub-basins with a high prediction accuracy. With the multi-predictor ensemble model, we derived a turbidity map for all rivers and streams with a width greater than 90 m in the basin, by processing three Landsat 8 image scenes. This basin-wide turbidity map indicates that the turbidity of rivers and streams exhibits longitudinal gradients and distinct spatial trends within the basin, which were related to the basin land cover/land use, hydrometeorological condition, and the spatial distribution of dams. Owing to its strong spatial transferability and high prediction accuracy, the ensemble model, in conjunction with synoptic multispectral satellite image data, is recommended as an effective and efficient tool for mapping and monitoring river and stream water quality at a large basin or regional scale.

Satellite-derived sediment distribution mapping using ICESat-2 and SuperDove

Nearshore sediment distribution is a significant component of fundamental geographic data, but the inherent optical properties of the water column make the large-scale observation of sediment particularly difficult. The development of satellite-based imaging sensors has enabled earth observation to capture more types and finer granularities of ground information, making it possible to obtain the target sediment distribution on a large scale. We use the proposed wavelet clustering algorithm to separate noise points and signal photons from ICESat-2 and then combine multispectral images and a quasi-analytical algorithm to calculate substrate reflectance to reduce the influence of the inherent properties of the water body. Furthermore, a new deep learning model, TSDANet, is being built to combine substrate reflectance and multispectral image data from SuperDove for nearshore sediment identification. We use field data to validate the method through a large number of experiments. The experiments demonstrate that the wavelet clustering algorithm has a coefficient of determination (R2) of 0.90 and 0.95 for signal point extraction over a large scale of high and low photon densities compared with a digital elevation model, respectively. Compared to single-source data, the TSDANet sediment classification kappa improves by 6.44% on average and 8.28% at its maximum. The results indicate that the identification and classification of sediment using ICESat-2 and SuperDove data is a new and highly feasible method.

Spatio-temporal reconstruction of substance dynamics using compressed sensing in multi-spectral magnetic resonance spectroscopic imaging

The objective of our study is to observe dynamics of multiple substances in vivo with high temporal resolution from multi-spectral magnetic resonance spectroscopic imaging (MRSI) data. The multi-spectral MRSI can effectively separate spectral peaks of multiple substances and is useful to measure spatial distributions of substances. However it is difficult to measure time-varying substance distributions directly by ordinary full sampling because the measurement requires a significantly long time. In this study, we propose a novel method to reconstruct the spatio-temporal distributions of substances from randomly undersampled multi-spectral MRSI data on the basis of compressed sensing (CS) and the partially separable function model with base spectra of substances. In our method, we have employed spatio-temporal sparsity and temporal smoothness of the substance distributions as prior knowledge to perform CS. By directly reconstructing the spatio-temporal distributions of the substances themselves without reconstructing the spectra, this method significantly reduces the amount of MRSI data required per single time frame. We have formulated a regularized minimization problem for reconstruction and solved it by the alternating direction method of multipliers (ADMM). The effectiveness of our method has been evaluated using phantom data sets of glass tubes filled with glucose or lactate solution in increasing amounts over time and animal data sets of a tumor-bearing mouse to observe the metabolic dynamics involved in the Warburg effect in vivo. The reconstructed results are consistent with the expected behaviors, showing that our method can reconstruct the spatio-temporal distribution of substances with a temporal resolution of four seconds which is extremely short time scale compared with that of full sampling. Since this method utilizes only prior knowledge naturally assumed for the spatio-temporal distributions of substances and is independent of the number of the spectral and spatial dimensions or the acquisition sequence of MRSI, it is expected to contribute to revealing the underlying substance dynamics in MRSI data already acquired or to be acquired in the future.

Machine learning algorithm inversion experiment and pollution analysis of water quality parameters in urban small and medium-sized rivers based on UAV multispectral data.

To examine and analyze the applicability of UAV multispectral images to urban river monitoring, this paper, taking the Fuyang River in the urban area of Handan Municipality as the object, the orthogonal image data of the river in different seasons were acquired by unmanned aerial vehicles (UAVs) equipped with multispectral sensors, and at the same time, the water samples were collected for physical and chemical indexes detection. Based on the image data, a total of 51 modeling spectral indexes were obtained by constructing three forms of band combinations ranging from the difference index (DI), ratio index (RI), and normalization index (NDI) and combining six single-band spectral values. Through the partial least squares (PLS), random forest (RF), and lasso prediction models, six fitting models of water quality parameters were constructed: turbidity (Turb), suspended, substance (SS), chemical oxygen demand (COD), ammonia nitrogen (NH4-N), total nitrogen (TN), and total phosphorus (TP). After verifying the results and evaluating the accuracy, the following conclusions were drawn: (1) The inversion accuracy of the three types of models is generally the same-summer is better than spring, and winter is the worst. (2) Water quality parameter inversion model based on two kinds of machine learning algorithms has more prominent advantages than PLS. RF model has good performance in the inversion accuracy and generalization ability of water quality parameters in different seasons. (3) The prediction accuracy and stability of the model are positively correlated to a certain extent with the size of the standard deviation of sample values. To sum up, by using the multispectral image data acquired by UAV and adopting the prediction models built upon machine learning algorithms, water quality parameters in different seasons can be predicted in different degrees.

Utilization of ASTER data in lithological and lineament mapping of the southern flank of the Central High Atlas in Morocco

Geological mapping undoubtedly plays an important role in several studies and remote sensing data are of great significance in geological mapping, particularly in poorly mapped areas situated in inaccessible regions. In the present study, Principal Component Analysis (PCA), Band Rationing (BR) and Minimum Noise Fraction (MNF) algorithms are applied to map lithological units and extract lineaments in the Amezri-Amassine area, by using multispectral ASTER image and global digital elevation model (GDEM) data for the first time. Following preprocessing of ASTER images, advanced image algorithms such as PCA, BR and MNF analyses are applied to the 9ASTER bands. Validation of the resultant maps has relied on matching lithological boundaries and faults in the study area and on the basis of pre-existing geological maps. In addition to the PCA image, a new band-ratio image, 4/6–5/8–4/5, as adopted in the present work, provides high accuracy in discriminating lithological units. The MNF transformation reveals improvement over previous enhancement techniques, in detailing most rock units in the area. Hence, results derived from the enhancement techniques show a good correlation with the existing litho-structural map of the study area. In addition, the present results have allowed to update this map by identifying new lithological units and structural lineaments. Consequently, the methodology followed here has provided satisfactory results and has demonstrated the high potential of multispectral ASTER data for improving lithological discrimination and lineament extraction.

Study on Monitoring SPAD Values for Multispatial Spatial Vertical Scales of Summer Maize Based on UAV Multispectral Remote Sensing

Rapid acquisition of chlorophyll content in maize leaves is of great significance for timely monitoring of maize plant health and guiding field management. In order to accurately detect the relative chlorophyll content of summer maize and study the responsiveness of vegetation indices to SPAD (soil and plant analyzer development) values of summer maize at different spatial vertical scales, this paper established a prediction model for SPAD values of summer maize leaves at different spatial scales based on UAV multispectral images. The experiment collected multispectral image data from summer maize at the jointing stage and selected eight vegetation indices. By using the sparrow search optimized kernel limit learning machine (SSA-KELM), the prediction models for canopy leaf (CL) SPADCL and ear leaf (EL) SPADEL were established, and a linear fitting analysis was conducted combining the measured SPADCL values and SPADEL values on the ground. The results showed that for SPADCL, the R2 of the linear fitting between the predicted values and measured values was 0.899, and the RMSE was 1.068. For SPADEL, the R2 of linear fitting between the predicted values and the measured values was 0.837, and the RMSE was 0.89. Compared with the model established by the partial least squares method (PLSR), it is found that the sparrow search optimized kernel limit learning machine (SSA-KELM) has more precise prediction results with better stability and adaptability for small sample prediction. The research results can provide technical support for remote sensing monitoring of the chlorophyll content of summer maize at different spatial scales.

Improved U-Net3+ With Spatial–Spectral Transformer for Multispectral Image Reconstruction

Multispectral image reconstruction, which aims to recover a three-dimensional (3D) spatial–spectral signal from a two-dimensional measurement in a spectral camera based on ghost imaging via sparsity constraint (GISC), has been attracting much attention recently. However, faced with abundant 3D spectral data, the reconstruction quality cannot meet the visual requirements. Based on the robust data processing capability of deep learning, a novel network called SSTU-Net3+ is constructed by improving U-Net3+ with a spatial–spectral transformer (SST). To enhance the feature representation of images during reconstruction, mixed pooling modules and new convolution processes are proposed to improve the performance of the encoder and decoder, with U-Net3+ as the backbone. To boost the quality of reconstructed images, with split and concatenate (Concat) operations, we construct SST modules by exploiting both spatial and spectral correlations of multispectral images to refine the spatial and spectral features. Furthermore, we employ the SST in the decoder to reconstruct the desired 3D cube. Given similar network parameters, experiments on GISC spectral imaging data show that, compared to convolutional neural network–based methods, the average peak signal-to-noise ratio of images reconstructed using SSTU-Net3+ is improved by 3%, the structural similarity is enhanced by 3%, and the spectral angle mapping is cut by 12%. Particularly, compared to differential ghost imaging and compressed sensing, the reconstruction quality of SSTU-Net3+ has been significantly improved. SSTU-Net3+ can process a large amount of 3D multispectral image data more efficiently and construct the target image more accurately than the abovementioned methods.

Evaluation of the spectral misalignment on the Earth Clouds, Aerosols and Radiation Explorer/multi-spectral imager cloud product

Abstract. A cloud identification and profiling algorithm is being developed for the multi-spectral imager (MSI), which is one of the four instruments that the Earth Clouds, Aerosols, and Radiation Explorer (EarthCARE) spacecraft will feature. During recent work, we noticed that the MSI response function could shift substantially among some wavelengths (0.67 and 1.65 µm bands) owing to the spectral misalignment (SMILE), in which a shift in the center wavelength appears as a distortion in the spectral image. We evaluated how SMILE affects the cloud retrieval product qualitatively and quantitatively. We chose four detector pixels from bands 1 and 3 with the nadir pixel as the reference to elucidate how the SMILE error affects the cloud optical thickness (τ) and effective cloud droplet radius (re) by simulating the MSI forward radiation with Comprehensive Analysis Program for Cloud Optical Measurement (CAPCOM). We also evaluated the error in simulated scenes from a global cloud system-resolving model and a satellite simulator to measure the effect on actual observation scenes. For typical shallow warm clouds (τ = 8, re = 8 µm), the SMILE error on the cloud retrieval was not significant in most cases (up to 6 % error). For typical deep convective clouds (τ = 8, re = 40 µm), the SMILE error on the cloud retrieval was even less significant in most cases (up to 4 % error). Moreover, our results from two oceanic scenes using the synthetic MSI data agreed well with the forward radiation simulation, indicating that the SMILE error was generally within 10 %. Generally, this negligible impact of the SMILE is true for water surfaces, but it still needs to be investigated further for land surfaces in future works.

Forest Land Resource Information Acquisition with Sentinel-2 Image Utilizing Support Vector Machine, K-Nearest Neighbor, Random Forest, Decision Trees and Multi-Layer Perceptron

Forestry work involves scientific management and the effective utilization of forest land resources, and finding economical, efficient and accurate acquisition methods for forest land resource information. In previous land-use classification research, machine learning algorithms have achieved good results, and Sentinel public data have been used in various remote sensing applications. However, there is a paucity of research using these data to evaluate the performance of machine learning algorithms in the extracting of complex forest land resource information. Using the Sentinel-2 satellite multispectral image data, the spectral reflectance, vegetation index characteristics and image texture characteristics of different forest land resources in the study area were calculated and compared. Then, based on three groups of features, the performances of the Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Random Forest (RF), decision trees (DT) and Multi-layer Perceptron (MLP) were examined and compared to identify and classify forest land resource types. The research indicates the following: (1) The SVM algorithm achieved the highest OA (95.8%). The average accuracy of the SVM algorithm was much higher than other algorithms (SVM 88.3%, KNN 87.5, RF 85.3%, MLP 85.00% and DT 77.5%). (2) The classification accuracies of each algorithm for coniferous forests were relatively high, and the recognition accuracy was above 95%, whereas the classification accuracies of the other categories varied greatly. (3) Adding texture features can improve the accuracy of the five algorithms. This study reports new references for the qualitative methods of forest land resource distribution. It has also produced more efficient and accurate acquisitions of forest land resource information, scientific management and effective use of forest land resources.

UAV imagery data and machine learning: A driving merger for predictive analysis of qualitative yield in sugarcane.

Predicting sugarcane yield by quality allows stakeholders from research centers to industries to decide on the precise time and place to harvest a product on the field; hence, it can streamline workflow while leveling up the cost-effectiveness of full-scale production. °Brix and Purity can offer significant and reliable indicators of high-quality raw material for industrial processing for food and fuel. However, their analysis in a relevant laboratory can be costly, time-consuming, and not scalable. We, therefore, analyzed whether merging multispectral images and machine learning (ML) algorithms can develop a non-invasive, predictive framework to map canopy reflectance to °Brix and Purity. We acquired multispectral images data of a sugarcane-producing area via unmanned aerial vehicle (UAV) while determining °Brix and analytical Purity from juice in a routine laboratory. We then tested a suite of ML algorithms, namely multiple linear regression (MLR), random forest (RF), decision tree (DT), and support vector machine (SVM) for adequacy and complexity in predicting °Brix and Purity upon single spectral bands, vegetation indices (VIs), and growing degree days (GDD). We obtained evidence for biophysical functions accurately predicting °Brix and Purity. Those can bring at least 80% of adequacy to the modeling. Therefore, our study represents progress in assessing and monitoring sugarcane on an industrial scale. Our insights can offer stakeholders possibilities to develop prescriptive harvesting and resource-effective, high-performance manufacturing lines for by-products.

Karst vegetation coverage detection using UAV multispectral vegetation indices and machine learning algorithm

BackgroundKarst vegetation is of great significance for ecological restoration in karst areas. Vegetation Indices (VIs) are mainly related to plant yield which is helpful to understand the status of ecological restoration in karst areas. Recently, karst vegetation surveys have gradually shifted from field surveys to remote sensing-based methods. Coupled with the machine learning methods, the Unmanned Aerial Vehicle (UAV) multispectral remote sensing data can effectively improve the detection accuracy of vegetation and extract the important spectrum features.ResultsIn this study, UAV multispectral image data at flight altitudes of 100 m, 200 m, and 400 m were collected to be applied for vegetation detection in a karst area. The resulting ground resolutions of the 100 m, 200 m, and 400 m data are 5.29, 10.58, and 21.16 cm/pixel, respectively. Four machine learning models, including Random Forest (RF), Support Vector Machine (SVM), Gradient Boosting Machine (GBM), and Deep Learning (DL), were compared to test the performance of vegetation coverage detection. 5 spectral values (Red, Green, Blue, NIR, Red edge) and 16 VIs were selected to perform variable importance analysis on the best detection models. The results show that the best model for each flight altitude has the highest accuracy in detecting its training data (over 90%), and the GBM model constructed based on all data at all flight altitudes yields the best detection performance covering all data, with an overall accuracy of 95.66%. The variables that were significantly correlated and not correlated with the best model were the Modified Soil Adjusted Vegetation Index (MSAVI) and the Modified Anthocyanin Content Index (MACI), respectively. Finally, the best model was used to invert the complete UAV images at different flight altitudes.ConclusionsIn general, the GBM_all model constructed based on UAV imaging with all flight altitudes was feasible to accurately detect karst vegetation coverage. The prediction models constructed based on data from different flight altitudes had a certain similarity in the distribution of vegetation index importance. Combined with the method of visual interpretation, the karst green vegetation predicted by the best model was in good agreement with the ground truth, and other land types including hay, rock, and soil were well predicted. This study provided a methodological reference for the detection of karst vegetation coverage in eastern China.

Aligning Small Datasets Using Domain Adversarial Learning: Applications in Automated in Vivo Oral Cancer Diagnosis.

Deep learning approaches for medical image analysis are limited by small data set size due to factors such as patient privacy and difficulties in obtaining expert labelling for each image. In medical imaging system development pipelines, phases for system development and classification algorithms often overlap with data collection, creating small disjoint data sets collected at numerous locations with differing protocols. In this setting, merging data from different data collection centers increases the amount of training data. However, a direct combination of datasets will likely fail due to domain shifts between imaging centers. In contrast to previous approaches that focus on a single data set, we add a domain adaptation module to a neural network and train using multiple data sets. Our approach encourages domain invariance between two multispectral autofluorescence imaging (maFLIM) data sets of in vivo oral lesions collected with an imaging system currently in development. The two data sets have differences in the sub-populations imaged and in the calibration procedures used during data collection. We mitigate these differences using a gradient reversal layer and domain classifier. Our final model trained with two data sets substantially increases performance, including a significant increase in specificity. We also achieve a significant increase in average performance over the best baseline model train with two domains (p = 0.0341). Our approach lays the foundation for faster development of computer-aided diagnostic systems and presents a feasible approach for creating a robust classifier that aligns images from multiple data centers in the presence of domain shifts.

Characteristics of Tidal Zone in Kalubhar Island, Gulf of Kachchh, Gujarat and Its Geomorphological Implications

Tidal networks, tidal flats and mangroves are some of the important features of tidal areas in the Gulf of Kachchh. Insight on the patterns of tidal networks, vegetation and sedimentation is useful in understanding geomorphological processes active on the region. In the present study features of tidal zone of Kalubhar island, located in the south-central part of the Gulf have been studied with emphasis on tidal networks using Sentinel-2 Multispectral Imager (MSI) data having spatial resolution of 10m. Typology of tidal network is described for the first time. The typology of tidal network pattern, their density, presence of mangroves and beaches indicate low energy depositional setting in eastern, south-eastern and southern parts. This coupled with continued sedimentation and long sub-aerial exposure during low tides may not be ideal environmentin future for live corals present in the intertidal zones. Keywords: Tidal Network Pattern, Sentinel-2 MSI, Kalubhar Island, Gulf of Kachchh

Fusion level of satellite and UAV image data for soil salinity inversion in the coastal area of the Yellow River Delta

ABSTRACT Rapid and accurate determination of soil salt content (SSC) and its spatial distribution are of great significance for the prevention and improvement of soil salinization. Satellite and unmanned aerial vehicle (UAV) remote sensing data have complementary advantages. The fusion of satellite and UAV multisource remote sensing data to improve the accuracy of SSC based on inversion methods has become a hot topic, and the appropriate fusion level of multisource remote sensing data needs to be explored and determined. The objective of this study was to determine the appropriate fusion level of Sentinel-2A Multispectral Instrument (Sentinel-MSI) and UAV image data for SSC inversion by comparing the fusion effect of three levels (spectral data, spectral index, and spectral model). A numerical regression method was employed to analyse the relationship between Sentinel-MSI and UAV image data (MSI-UAV), and MSI-UAV data were fused at different levels. Then, the appropriate fusion level and best inversion model were optimized to realize regional SSC inversion. The results indicate that spectral data fusion was better than spectral index fusion for enhancing the SSC spectral response, with the correlation between spectral indices and SSC increasing by 0.139–0.167 after fusion. After spectral data fusion, the model improved the SSC inversion accuracy most obviously, with a calibration R2 of 0.623, validation R2 of 0.571, and ratio of performance to deviation (RPD) of 1.821. Therefore, spectral data fusion was found to be superior in enhancing the spectral response of soil salinity and in improving the accuracy of the estimation model. This research optimized spectral data fusion as the appropriate fusion level of MSI-UAV for SSC inversion and formed a set of high-precision MSI-UAV multisource remote sensing fusion inversion approaches for SSC.

Short-range multispectral imaging is an inexpensive, fast, and accurate approach to estimate biodiversity in a temperate calcareous grassland.

Image sensing technologies are rapidly increasing the cost-effectiveness of biodiversity monitoring efforts. Species differences in the reflectance of electromagnetic radiation can be used as a surrogate estimate plant biodiversity using multispectral image data. However, these efforts are often hampered by logistical difficulties in broad-scale implementation. Here, we investigate the utility of multispectral imaging technology from commercially available unmanned aerial vehicles (UAVs, or drones) in estimating biodiversity metrics at a fine spatial resolution (0.1-0.5cm pixel resolution) in a temperate calcareous grassland in Oxfordshire, UK. We calculate a suite of moments (coefficient of variation, standard deviation, skewness, and kurtosis) for the distribution of radiance from multispectral images at five wavelength bands (Blue 450 ± 16 nm; Green 560 ± 16 nm; Red 650 ± 16 nm; Red Edge 730 ± 16 nm; Near Infrared 840 ± 16 nm) and test their effectiveness at estimating ground-truthed biodiversity metrics from in situ botanical surveys for 37-1 × 1m quadrats. We find positive associations between the average coefficient of variation in spectral radiance and both the Shannon-Weiner and Simpson's biodiversity indices. Furthermore, the average coefficient of variation in spectral radiance is consistent and highly repeatable across sampling days and recording heights. Positive associations with biodiversity indices hold irrespective of the image recording height (2-8m), but we report reductions in estimates of spectral diversity with increases to UAV recording height. UAV imaging reduced sampling time by a factor of 16 relative to in situ botanical surveys. We demonstrate the utility of multispectral radiance moments as an indicator of biodiversity in this temperate calcareous grassland at a fine spatial resolution using a widely available UAV monitoring system with a coarse spectral resolution. The use of UAV technology with multispectral sensors has far-reaching potential to provide cost-effective and high-resolution monitoring of biodiversity.

Determination of Wheat Heading Stage Using Convolutional Neural Networks on Multispectral UAV Imaging Data.

The heading and flowering stages are crucial for wheat growth and should be used for fusarium head blight (FHB) and other plant prevention operations. Rapid and accurate monitoring of wheat growth in hilly areas is critical for determining plant protection operations and strategies. Currently, the operation time for FHB prevention and plant protection is primarily determined by manual tour inspection of plant growth, which has the disadvantages of low information gathering and subjectivity. In this study, an unmanned aerial vehicle (UAV) equipped with a multispectral camera was used to collect wheat canopy multispectral images and heading rate information during the heading and flowering stages in order to develop a method for detecting the appropriate time for preventive control of FHB. A 1D convolutional neural network + decision tree model (1D CNN + DT) was designed. All the multispectral information was input into the model for feature extraction and result regression. The regression revealed that the coefficient of determination (R 2) between multispectral information in the wheat canopy and the heading rate was 0.95, and the root mean square error of prediction (RMSE) was 0.24. This result was superior to that obtained by directly inputting multispectral data into neural networks (NN) or by inputting multispectral data into NN via traditional VI calculation, support vector machines regression (SVR), or decision tree (DT). On the basis of FHB prevention and control production guidelines and field research results, a discrimination model for FHB prevention and plant protection operation time was developed. After the output values of the regression model were input into the discrimination model, a 97.50% precision was obtained. The method proposed in this study can efficiently monitor the growth status of wheat during the heading and flowering stages and provide crop growth information for determining the timing and strategy of FHB prevention and plant protection operations.