Remote Sensing — Open Access Journal

This paper describes in detail the development of a ground-penetrating radar (GPR) model for the acquisition, processing and visualisation of underground utility infrastructure (UUI) in a controlled environment. The initiative was to simulate a subsurface urban environment through the construction of regional road, local road and pedestrian pavement in real urban field/testing pools (RUTPs). The RUTPs represented a controlled environment in which the most commonly used utilities were installed. The accuracy of the proposed kinematic GPR-TPS (terrestrial positioning system) model was analysed using all the available data about the materials, whilst taking into account the thickness of the pavement as well as the materials, dimensions and 3D position of the UUI as given reference values. To determine the reference 3D position of the UUI, a terrestrial geodetic surveying method based on the established positional and height geodetic network was used. In the first phase of the model, the geodetic network was used as a starting point for determining the 3D position of the GPR antenna with the efficient kinematic GPR surveying setup using a GPR and self-tracking (robotic) TPS. In the second phase, GPR-TPS system latency was quantified by matching radargram pairs with a set of fidelity measures based on a correlation coefficient and mean squared error. This was followed by the most important phase, where, by combining sets of “standard” processing routines of GPR signals with the support of advanced algorithms for signal processing, UUI were interpreted and visualised semi-automatically. As demonstrated by the results, the proposed GPR model with a kinematic GPR-TPS surveying setup for data acquisition is capable of achieving an accuracy of less than ten centimetres. Full article

The importance of snow cover extent (SCE) has been proven to strongly link with various natural phenomenon and human activities; consequently, monitoring snow cover is one the most critical topics in studying and understanding the cryosphere. As snow cover can vary significantly within short time spans and often extends over vast areas, spaceborne remote sensing constitutes an efficient observation technique to track it continuously. However, as optical imagery is limited by cloud cover and polar darkness, synthetic aperture radar (SAR) attracted more attention for its ability to sense day-and-night under any cloud and weather condition. In addition to widely applied backscattering-based method, thanks to the advancements of spaceborne SAR sensors and image processing techniques, many new approaches based on interferometric SAR (InSAR) and polarimetric SAR (PolSAR) have been developed since the launch of ERS-1 in 1991 to monitor snow cover under both dry and wet snow conditions. Critical auxiliary data including DEM, land cover information, and local meteorological data have also been explored to aid the snow cover analysis. This review presents an overview of existing studies and discusses the advantages, constraints, and trajectories of the current developments. Full article

Suspended solids concentration (SSC) is an important indicator of the degree of water pollution. However, when using an empirical or semi-empirical model adapted to some of the inland waters to estimate SSC on unmanned aerial vehicle (UAV)-borne hyperspectral images, the accuracy is often not sufficient. Thus, in this study, we attempted to use the particle swarm optimization (PSO) algorithm to find the optimal parameters of the least-squares support vector machine (LSSVM) model for the quantitative inversion of SSC. A reservoir and a polluted riverway were selected as the study areas. The spectral data of the 36-point and 29-point 400–900 nm wavelength range on the UAV-borne images were extracted. Compared with the semi-empirical model, the random forest (RF) algorithm and the competitive adaptive reweighted sampling (CARS) algorithm combined with partial least squares (PLS), the accuracy of the PSO-LSSVM algorithm in predicting the SSC was significantly improved. The training samples had a coefficient of determination ($R^2$) of 0.98, a root mean square error (RMSE) of 0.68 mg/L, and a mean absolute percentage error (MAPE) of 12.66% at the reservoir. For the polluted riverway, PSO-LSSVM also performed well. Finally, the established SSC inversion model was applied to UAV-borne hyperspectral remote sensing (HRS) images. The results confirmed that the distribution of the predicted SSC was consistent with the observed results in the field, which proves that PSO-LSSVM is a feasible approach for the SSC inversion of UAV-borne HRS images. Full article

This paper proposes a complete framework of a machine learning-based model that detects convective initiation (CI) from geostationary meteorological satellite data. The suggested framework consists of three main processes: (1) An automated sampling tool; (2) machine learning-based CI detection modelling; (3) repeated model tuning through validation. In this study, the automated sampling tool was able to track the CI objects iteratively, even without ancillary data such as an atmospheric motion vector (AMV). The collected samples were used to train the machine learning model for CI detection. Random forest (RF) was used to classify the CI and non-CI. To enhance the advantages of the machine learning approach, we adopted model tuning to iteratively update the training dataset from each validation result by adding hits and misses to the CI samples, and false alarms and correct negatives to the non-CI samples. Using 12 interest fields from the Himawari-8 Advanced Himawari Imager (AHI) over the Korean Peninsula, this simple and intuitive tuning process increased the overall probability of detection (POD) from 0.79 to 0.82 and decreased the overall false alarm rate (FAR) from 0.46 to 0.37 with around 40 min of the lead-time. Amongst the 12 interest fields, $T_b$(11.2) µm was identified as the most significant predictor in the RF model, followed by $T_b$(8.6—11.2) µm, and $T_b$(6.2–7.3) µm. The effect of model tuning on the CI detection performance was also analyzed using spatiotemporal validation maps. By automatically collecting and updating the machine learning training dataset, the suggested framework is expected to help the maintenance of the CI detection model from an operational perspective. Full article

Maintaining the high visual recognizability of traffic signs for traffic safety is a key matter for road network management. Mobile Laser Scanning (MLS) systems provide efficient way of 3D measurement over large-scale traffic environment. This paper presents a quantitative visual recognizability evaluation method for traffic signs in large-scale traffic environment based on traffic recognition theory and MLS 3D point clouds. We first propose the Visibility Evaluation Model (VEM) to quantitatively describe the visibility of traffic sign from any given viewpoint, then we proposed the concept of visual recognizability field and Traffic Sign Visual Recognizability Evaluation Model (TSVREM) to measure the visual recognizability of a traffic sign. Finally, we present an automatic TSVREM calculation algorithm for MLS 3D point clouds. Experimental results on real MLS 3D point clouds show that the proposed method is feasible and efficient. Full article

Himalayan glaciers have shrunk rapidly in recent decades, but the spatial pattern of ice loss is highly variable and appears to be modulated by factors relating to individual glacier characteristics. This hinders our ability to predict their future evolution, which is vital for water resource management. The aim of this study is to assess recent glacier changes in the little-studied Annapurna Conservation Area (ACA; area: 7629 km²) in Nepal, and to explore local controls influencing their behaviour. We map changes in glacier area, surface elevation, and ice flow velocity on a large sample of glaciers (n = 162) in the ACA between 2000 and 2016. We found that total glacier area decreased by 8.5% between 2000 and 2014/15. Ice surface velocity changes between 2002 and 2016 were variable, with no clear trend of acceleration or deceleration. The mean surface elevation change for a smaller sample of glaciers (n = 72) was −0.33 ± 0.22 m a⁻¹ between 2000 and 2013/16, which equates to a mean mass balance of −0.28 ± 0.24 m w.e. a⁻¹. There was a trend of increasingly less negative mass balance towards the north. Glaciers that lost the most mass in the north of the ACA tended to have lower maximum elevations, bottom-heavy hypsometries, and were more likely to be avalanche-fed. However, these patterns were not apparent in glaciers in central ACA. There was no significant difference in the mean surface elevation change rate on the ablation zones of debris-covered compared with debris-free glaciers. Our work shows that glaciers in the ACA are losing area and mass at variable rates, but that the influence of local controls is complex, which introduces large uncertainties when predicting their future evolution. Full article

Net Primary Productivity (NPP) is a basis of material and energy flows in terrestrial ecosystems, and it is also an important component in the research on carbon cycle and carbon budget. This paper evaluated the spatial distribution pattern and temporal change trends for forest NPP simulated by the LPJ (Lund-Potsdam-Jena) model and NDVI (normalized difference vegetation index) in the Yangtze River basin from 1982 to 2013. The results revealed that: (1) the spatial distribution of the forest NPP and NDVI in the Yangtze River basin has gradually decreased from the southeast coast to the northwest. The forest NPP and NDVI in the mid-lower Yangtze were higher than that of the upper Yangtze; (2) the forest NPP and NDVI in most areas of the Yangtze River basin were positively correlated with the temperature and precipitation. Moreover, the correlations among the temperature with the forest NPP and NDVI were stronger than that of correlations among precipitation with forest NPP and NDVI. Moreover, the extreme drought event in the year of 2004–2005 led the NPP to decrease in the middle and lower Yangtze River basin; (3) human activity such as major ecological projects would have a certain impact on the NPP and NDVI. The increase in forest areas from 2000 to 2010 was larger than that from 1990 to 2000. Moreover, the increasing rate for the NDVI was higher than that of NPP, especially after the year 2000, which indicates that the major ecological projects might have great impacts on the vegetation dynamics. Moreover, more attention should be paid on the joint impacts of human activity and climate change on terrestrial NPP and NDVI. Full article

Satellite-based multi-sensor data coupled with field and microscopic investigations are used to unravel the setting and controls of gold mineralization in the Wadi Beitan–Wadi Rahaba area in the South Eastern Desert of Egypt. The satellite-based multispectral and Synthetic Aperture Radar (SAR) data promoted a vibrant litho-tectonic understanding and abetted in assessing the regional structural control of the scattered gold occurrences in the study area. The herein detailed approach includes band rationing, principal component and independent component analyses, directional filtering, and automated and semi-automated lineament extraction techniques to Landsat 8- Operational Land Imager (OLI), Advanced Spaceborne Thermal Emission and Reflection Radiomete (ASTER), Phased Array L-band Synthetic Aperture Radar (PALSAR), and Sentinel-1B data. Results of optical and SAR data processed as grayscale raster images of band ratios, Relative Absorption Band Depth (RBD), and (mafic–carbonate–hydrous) mineralogical indices are used to extract the representative pixels (regions of interest). The extracted pixels are then converted to vector shape files and are finally imported into the ArcMap environment. Similarly, manually and automatically extracted lineaments are merged with the band ratios and mineralogical indices vector layers. The data fusion approach used herein reveals no particular spatial association between gold occurrences and certain lithological units, but shows a preferential distribution of gold–quartz veins in zones of chlorite–epidote alteration overlapping with high-density intersections of lineaments. Structural features including en-echelon arrays of quartz veins and intense recrystallization and sub-grain development textures are consistent with vein formation and gold deposition syn-kinematic with the host shear zones. The mineralized, central-shear quartz veins, and the associated strong stretching lineation affirm vein formation amid stress build-up and stress relaxation of an enduring oblique convergence (assigned as Najd-related sinistral transpression; ~640–610 Ma). As the main outcome of this research, we present a priority map with zones defined as high potential targets for undiscovered gold resources. Full article

Urban Heat Islands (UHIs) at the surface and canopy levels are major issues in urban planification and development. For this reason, the comprehension and quantification of the influence that the different land-uses/land-covers have on UHIs is of particular importance. In order to perform a detailed thermal characterisation of the city, measures covering the whole scenario (city and surroundings) and with a recurrent revisit are needed. In addition, a resolution of tens of meters is needed to characterise the urban heterogeneities. Spaceborne remote sensing meets the first and the second requirements but the Land Surface Temperature (LST) resolutions remain too rough compared to the urban object scale. Thermal unmixing techniques have been developed in recent years, allowing LST images during day at the desired scales. However, while LST gives information of surface urban heat islands (SUHIs), canopy UHIs and SUHIs are more correlated during the night, hence the development of thermal unmixing methods for night LSTs is necessary. This article proposes to adapt four empirical unmixing methods of the literature, Disaggregation of radiometric surface Temperature (DisTrad), High-resolution Urban Thermal Sharpener (HUTS), Area-To-Point Regression Kriging (ATPRK), and Adaptive Area-To-Point Regression Kriging (AATPRK), to unmix night LSTs. These methods are based on given relationships between LST and reflective indices, and on invariance hypotheses of these relationships across resolutions. Then, a comparative study of the performances of the different techniques is carried out on TRISHNA synthesized images of Madrid. Since TRISHNA is a mission in preparation, the synthesis of the images has been done according to the planned specification of the satellite and from initial Aircraft Hyperspectral Scanner (AHS) data of the city obtained during the DESIREX 2008 capaign. Thus, the coarse initial resolution is 60 m and the finer post-unmixing one is 20 m. In this article, we show that: (1) AATPRK is the most performant unmixing technique when applied on night LST, with the other three techniques being undesirable for night applications at TRISHNA resolutions. This can be explained by the local application of AATPRK. (2) ATPRK and DisTrad do not improve significantly the LST image resolution. (3) HUTS, which depends on albedo measures, misestimates the LST, leading to the worst temperature unmixing. (4) The two main factors explaining the obtained performances are the local/global application of the method and the reflective indices used in the LST-index relationship. Full article

Satellite ocean color products from the Visible Infrared Imaging Radiometer Suite (VIIRS) onboard the Suomi National Polar-orbiting Partnership (SNPP) since 2012 and in situ water turbidity measurements from the U.S. Environmental Protection Agency’s Great Lakes Environmental Database System are used to develop a water turbidity algorithm for satellite ocean color applications in the Great Lakes for water quality monitoring and assessments. Results show that the proposed regional algorithm can provide reasonably accurate estimations of water turbidity from satellite observations in the Great Lakes. Therefore, VIIRS-derived water turbidity data are used to investigate spatial and temporal variations in water turbidity for the entirety of the Great Lakes. Water turbidity values are overall the highest in Lake Erie, moderate in Lake Ontario, and relatively low in lakes Superior, Michigan, and Huron. Significantly high values in water turbidity appear in the nearshore regions, particularly in Thunder Bay (Lake Superior), Green Bay (Lake Michigan), and Saginaw Bay (Lake Huron). Seasonal patterns of water turbidity are generally similar in lakes Superior, Michigan, Huron, and Ontario, showing relatively high values in the spring and autumn months and lows in the winter season, while the seasonal pattern in Lake Erie is apparently different from the other lakes, with the highest value in the winter season and the lowest in the summer season. A strong interannual variability in water turbidity is shown in the time series of the VIIRS-derived water turbidity data for most of the lakes. Full article

Forest structure is a crucial component in the assessment of whether a forest is likely to act as a carbon sink under changing climate. Detailed 3D structural information about the tundra–taiga ecotone of Siberia is mostly missing and still underrepresented in current research due to the remoteness and restricted accessibility. Field based, high-resolution remote sensing can provide important knowledge for the understanding of vegetation properties and dynamics. In this study, we test the applicability of consumer-grade Unmanned Aerial Vehicles (UAVs) for rapid calculation of stand metrics in treeline forests. We reconstructed high-resolution photogrammetric point clouds and derived canopy height models for 10 study sites from NE Chukotka and SW Yakutia. Subsequently, we detected individual tree tops using a variable-window size local maximum filter and applied a marker-controlled watershed segmentation for the delineation of tree crowns. With this, we successfully detected 67.1% of the validation individuals. Simple linear regressions of observed and detected metrics show a better correlation (R²) and lower relative root mean square percentage error (RMSE%) for tree heights (mean R² = 0.77, mean RMSE% = 18.46%) than for crown diameters (mean R² = 0.46, mean RMSE% = 24.9%). The comparison between detected and observed tree height distributions revealed that our tree detection method was unable to representatively identify trees <2 m. Our results show that plot sizes for vegetation surveys in the tundra–taiga ecotone should be adapted to the forest structure and have a radius of >15–20 m to capture homogeneous and representative forest stands. Additionally, we identify sources of omission and commission errors and give recommendations for their mitigation. In summary, the efficiency of the used method depends on the complexity of the forest’s stand structure. Full article

Forest biomass is an important descriptor for studying carbon storage, carbon cycles, and global change science. The full-waveform spaceborne Light Detection And Ranging (LiDAR) Geoscience Laser Altimeter System (GLAS) provides great possibilities for large-scale and long-term biomass estimation. To the best of our knowledge, most of the existing research has utilized average tree height (or height metrics) within a GLAS footprint as the key parameter for biomass estimation. However, the vertical distribution of tree height is usually not as homogeneous as we would expect within such a large footprint of more than 2000 m², which would limit the biomass estimation accuracy vastly. Therefore, we aim to develop a novel canopy height layering biomass estimation model (CHL-BEM) with GLAS data in this study. First, all the trees with similar height were regarded as one canopy layer within each GLAS footprint. Second, the canopy height and canopy cover of each layer were derived from GLAS waveform parameters. These parameters were extracted using a waveform decomposition algorithm (refined Levenberg–Marquardt—RLM), which assumed that each decomposed vegetation signal corresponded to a particular canopy height layer. Third, the biomass estimation model (CHL-BEM) was established by using the canopy height and canopy cover of each height layer. Finally, the CHL-BEM was compared with two typical biomass estimation models of GLAS in the study site located in Ejina, China, where the dominant species was Populus euphratica. The results showed that the CHL-BEM presented good agreement with the field measurement biomass (R² = 0.741, RMSE = 0.487, %RMSE = 24.192) and achieved a significantly higher accuracy than the other two models. As a whole, we expect our method to advance all the full-waveform LiDAR development and applications, e.g., the newly launched Global Ecosystem Dynamics Investigation (GEDI). Full article

High spatial resolution benthic habitat information is essential for coral reef protection and coastal environmental management. Satellite-based shallow benthic composition mapping offers a more efficient approach than traditional field measurements, especially given the advancements in high spatial and temporal resolution satellite imagery. The Planet Dove satellite constellation now has more than 150 instruments in orbit that offer daily coverage at high spatial resolution (3.7 m). The Dove constellation provides regularly updated imagery that can minimize cloud in tropical oceans where dense cloud cover persists. Daily image acquisition also provides an opportunity to detect time-sensitive changes in shallow benthic habitats following coral bleaching events, storms, and other disturbances. We developed an object-based coral reef habitat mapping approach for Dove and similar multispectral satellites that provides bathymetry estimation, bottom reflectance retrieval, and object-based classification to identify different benthic compositions in shallow coastal environments. We tested our approach in three study sites in the Dominican Republic using 18 Dove images. Benthic composition classification results were validated by field measurements (overall accuracy = 82%). Bathymetry and bottom reflectance significantly contributed to identifying benthic habitat classes with similar surface reflectance. This new object-based approach can be effectively applied to map and manage coral reef habitats. Full article

Remote sensing data can be utilized to help developing countries monitor the use of land. However, the problem of constant cloud coverage prevents us from taking full advantage of satellite optical images. Therefore, we instead opt to use data from synthetic-aperture radar (SAR), which can capture images of the Earth’s surface regardless of the weather conditions. In this study, we use SAR data to identify newly built constructions. Most studies on change detection tend to detect all of the changes that have a similar temporal change characteristic occurring on two occasions, while we want to identify only the constructions and avoid detecting other changes such as the seasonal change of vegetation. To do so, we study various deep learning network techniques and have decided to propose the fully convolutional network with a skip connection. We train this network with pairs of SAR data acquired on two different occasions from Bangkok and the ground truth, which we manually create from optical images available from Google Earth for all of the SAR pairs. Experiments to assign the most suitable patch size, loss weighting, and epoch number to the network are discussed in this paper. The trained model can be used to generate a binary map that indicates the position of these newly built constructions precisely with the Bangkok dataset, as well as with the Hanoi and Xiamen datasets with acceptable results. The proposed model can even be used with SAR images of the same specific satellite from another orbit direction and still give promising results. Full article

The unmanned aerial vehicle (UAV) sensors and platforms nowadays are being used in almost every application (e.g., agriculture, forestry, and mining) that needs observed information from the top or oblique views. While they intend to be a general remote sensing (RS) tool, the relevant RS data processing and analysis methods are still largely ad-hoc to applications. Although the obvious advantages of UAV data are their high spatial resolution and flexibility in acquisition and sensor integration, there is in general a lack of systematic analysis on how these characteristics alter solutions for typical RS tasks such as land-cover classification, change detection, and thematic mapping. For instance, the ultra-high-resolution data (less than 10 cm of Ground Sampling Distance (GSD)) bring more unwanted classes of objects (e.g., pedestrian and cars) in land-cover classification; the often available 3D data generated from photogrammetric images call for more advanced techniques for geometric and spectral analysis. In this paper, we perform a critical review on RS tasks that involve UAV data and their derived products as their main sources including raw perspective images, digital surface models, and orthophotos. In particular, we focus on solutions that address the “new” aspects of the UAV data including (1) ultra-high resolution; (2) availability of coherent geometric and spectral data; and (3) capability of simultaneously using multi-sensor data for fusion. Based on these solutions, we provide a brief summary of existing examples of UAV-based RS in agricultural, environmental, urban, and hazards assessment applications, etc., and by discussing their practical potentials, we share our views in their future research directions and draw conclusive remarks. Full article

Over the last few decades, small-scale irrigation has been implemented in Burkina Faso as a strategy to mitigate the impacts of adverse climate conditions. However, the development of irrigated perimeters around small and medium water reservoirs has put the water resources under significant pressure, given the uncontrolled exploitation and lack of efficacious management plan. Insights into changes in irrigated areas around these reservoirs are therefore crucial for their sustainable management while meeting the different agricultural water needs. They will help to center policy priorities in terms of major impacts on the reservoirs; and thereby elaborate relevant mitigation and/or adaptation strategies. The main objectives of this study were to (1) quantify the changes in irrigated land areas surrounding the Mogtedo water reservoir between 1987 and 2015; and (2) determine whether the irrigable potential of this reservoir could sustainably meet the agricultural water needs under a more variable and changing climate. A low-cost remote sensing method based on Landsat imagery (Thematic Mapper, Enhanced Thematic Mapper Plus, and Operational Land Imager) and using Support Vector Machine (SVM) classification was developed to detect the changes in proportion of land use/land cover (LULC) in the Mogtedo region. A forward and backward change detection analysis requiring agronomic expertise was also applied to correct the pixels temporal trajectories. In addition, an intensity analysis was performed to assess land changes at time intervals, category, and transition levels. Five main LULC classes were identified: bare and hydromorphic soils, irrigated and rainfed agricultural areas, and water bodies. Overall, the classification of LULC was satisfactory with the overall accuracy and kappa coefficients ranging from 94.22 to 95.60% and 0.92 to 0.94, respectively. Results showed that LULC transformations were faster between 2000 and 2015, compared to the 1987–2000 period. The majority of categories (LULC classes) were active in terms of intensity of change (gain or loss) during the 1987–2000 and 2000–2015 periods, except hydromorphic soils. During these periods, the transition from rainfed agricultural areas to irrigated agricultural areas were targeted and stationary. Our findings revealed a 54% increase in irrigated areas between 1987 and 2015. The reservoir water volume decreased markedly from 9,077,000 m³ to 7,100,000 m³ during the same period. Such a decrease threatens the satisfaction of agricultural water requirements, since the reservoir is the unique source of irrigation water in the region. It could potentially lead to conflicts between users if adequate strategies for the sustainable management of the Mogtedo reservoir are not implemented. The methodology used in this study also addressed the challenge of building up historical spatial information database in data-scarce environments, and could be replicated readily in regions or countries like Burkina Faso. Full article

Monitoring agricultural crops is necessary for decision-making in the field. However, it is known that in some regions and periods, cloud cover makes this activity difficult to carry out in a systematic way throughout the phenological cycle of crops. This circumstance opens up opportunities for techniques involving radar sensors, resulting in images that are free of cloud effects. In this context, the objective of this work was to obtain a normalized different vegetation index (NDVI) cloudless product (NDVI_nc) by modeling Sentinel 2 NDVI using different regression techniques and the Sentinel 1 radar backscatter as input. To do this, we used four pairs of Sentinel 2 and Sentinel 1 images on coincident days, aiming to achieve the greatest range of NDVI values for agricultural crops (soybean and maize). These coincident pairs were the only ones in which the percentage of clouds was not equal to 100% for 33 central pivot areas in western Bahia, Brazil. The dataset used for NDVI_nc modeling was divided into two subsets: training and validation. The training and validation datasets were from the period from 24 June 2017 to 19 July 2018 (four pairs of images). The best performing model was used in a temporal analysis from 02 October 2017 to 08 August 2018, totaling 55 Sentinel 2 images and 25 Sentinel 1 images. The selection of the best regression algorithm was based on two validation methodologies: K-fold cross-validation (k = 10) and holdout. We tested four modeling approaches with eight regression algorithms. The random forest was the algorithm that presented the best statistical metrics, regardless of the validation methodology and the approach used. Therefore, this model was applied to a time series of Sentinel 1 images in order to demonstrate the robustness and applicability of the model created. We observed that the data derived from Sentinel 1 allowed us to model, with great reliability, the NDVI of agricultural crops throughout the phenological cycle, making the methodology developed in this work a relevant solution for the monitoring of various regions, regardless of cloud cover. Full article

Vegetation water content (VWC) is recognized as an important parameter in vegetation growth studies, natural disasters such as forest fires, and drought prediction. Recently, the Global Navigation Satellite System Interferometric Reflectometry (GNSS-IR) has emerged as an important technique for monitoring vegetation information. The normalized microwave reflection index (NMRI) was developed to reflect the change of VWC based on this fact. However, NMRI uses local site-based data, and the sparse distribution hinders the application of NMRI. In this study, we obtained a 500 m spatially continuous NMRI product by integrating GNSS-IR site data with other VWC-related products using the point–surface fusion technique. The auxiliary data in the fusion process include the normalized difference vegetation index (NDVI), gross primary productivity (GPP), and precipitation. Meanwhile, the fusion performance of three machine learning methods, i.e., the back-propagation neural network (BPNN), generalized regression neural network (GRNN), and random forest (RF) are compared and analyzed. The machine learning methods achieve satisfactory results, with cross-validation R values of 0.71–0.83 and RMSEs of 0.025–0.037. The results show a clear improvement over the traditional multiple linear regression method, which achieves R (RMSE) values of only about 0.4 (0.045). It indicates that the machine learning methods can better learn the complex nonlinear relationship between NMRI and the input VWC-related index. Among the machine learning methods, the RF model obtained the best results. Long time-series NMRI images with a 500 m spatial resolution in the western part of the continental U.S. were then obtained. The results show that the spatial distribution of the NMRI product is consistent with a drought situation from 2012 to 2014 in the U.S., which verifies the feasibility of analyzing and predicting drought times and distribution ranges by using the 500 m fusion product. Full article