Science of Remote Sensing最新文献

Over the past few years, the Optical Trapezoid Model (OPTRAM) has been widely used as a means for high-resolution mapping of surface soil moisture using optical satellite data. In this paper, we propose a new variant of OPTRAM that can map not only soil moisture, but also water bodies such as lakes and rivers. The proposed variant was tested using laboratory experimental data as well as Landsat-8 reflectance observations. Results showed the new OPTRAM variant has greater skill than the original variant in separating land and water pixels. In addition, the new variant showed less sensitivity to the model parameters, and hence, is less user dependent. To quantitatively examine the user-dependency of the model, we analyzed OPTRAM soil moisture based on Landsat-8 satellite images in California, where we varied the model parameters in a plausible range. The correlations of the resulting maps in terms of R² between two largely different sets of parameters were found in the range of 0.47-0.52 for the original variant and 0.67-0.76 for the new variant. Because some OPTRAM parameters can be quite uncertain, particularly in wet regions, the reduced sensitivity promises more consistent soil moisture estimates across the range of parameter choices.

Accurate monitoring and evaluation of root-zone soil salt content (SSC) are critical for sustainable development of irrigated agriculture in arid and semi-arid areas. Based on soil-crop water relations and farmland evapotranspiration (ET) fused through remote sensing data, this study developed an inversion method to estimate root-zone SSC using a case study from cotton fields under film mulched drip irrigation (CFFMDI) in the Manas River Basin (MRB) over 21 years (2000–2020). Two hypotheses were set as: (1) relative transpiration can be approximated by relative ET; and (2) the soil water stress response function is linearly proportional to the ratio of relative water supply. Measured data from a field experiment and collected data from regional survey and literature retrieval were used to optimize parameters and verify the hypotheses and method. The method was then applied to analyze the spatial and temporal distribution characteristics and cumulative effects of root-zone SSC. Results showed that the hypotheses and the method were reasonable and reliable in estimating root-zone SSC (with coefficient of determination R² > 0.50). Along with the popularization of film-mulched drip irrigation and the expansion of CFFMDI over the past 21 years, regional-scale root-zone SSC declined significantly with an annual attenuation rate of about 0.09 g kg⁻¹. Due to the gradual reduction of irrigation amount per unit area, the decline was more rapid before 2011 (0.18 g kg⁻¹), but slightly slowed down or even reversed at the end of the second decade (2015–2020). By 2020, the mean regional root-zone SSC reached 3.93 g kg⁻¹. At the beginning of this century, MRB was mainly composed of mildly- (59.8%) and moderately-salinized CFFMDI (39.9%). However, by 2020, non- (69.7%) and mildly-salinized cotton field (28.2%) dominated the basin. The inversion method of root-zone SSC fully considers the water consumption mechanism of soil-crop system, thus shows great potential in effective planning and management of soil and water resources in arid salinized areas such as MRB.

The Landsat global consolidated data archive now exceeds 50 years. In recognition of the need for consistently processed data across the Landsat satellite series, the U.S. Geological Survey (USGS) initiated collection-based processing of the entire archive that was processed as Collection 1 in 2016. In preparation for the data from the now successfully launched Landsat 9, the USGS reprocessed the Landsat archive as Collection 2 in 2020. This paper describes the rationale for, and the contents and advancements provided by Collection 2, and highlights the differences between the Collection 1 and Collection 2 products. Notably, the Collection 2 products have improved geolocation and, for the first time, the USGS provides a global inventory of Level 2 surface reflectance and surface temperature products. Also for the first time, the USGS used a commercial cloud computing architecture to efficiently process the archive and enable direct cloud access of the Landsat products. The paper concludes with discussion of likely improvements expected in Collection 3 in preparation for the Landsat Next mission that is planned for launch in the early 2030s.

Forest fires are able to cause significant damage to humans and the earth's fauna and flora. If a fire is not detected and extinguished before it spreads, it can have disastrous results. In addition to satellite images, recent studies have shown that exploring both weather and topography characteristics is crucial for effectively predicting the propagation of wildfires. In this paper, we present FU-NetCastV2, a deep learning convolutional neural network for fire spread and burned area mapping. This algorithm predicts which areas around wildfires are at high risk of future spread. With an accuracy of 94.6% and an AUC of 97.7%, the model surpassed the literature by 3.7% and exhibited a 1.9% improvement over our previous model. The proposed approach was implemented using consecutive forest wildfire perimeters, satellite images, Digital Elevation Model maps, aspect, slope and weather data.

Fine spatial resolution remote sensing images are crucial sources of data for monitoring the Earth's surface. Due to defects in sensors and the complicated imaging environment, however, fine spatial resolution images always suffer from various degrees of information loss. According to the basic attributes of remote sensing images, the information loss generally falls into three dimensions, that is, the spatial, temporal and spectral dimensions. In recent decades, many methods have been developed to cope with this information loss problem in the three dimensions, which are termed spatial reconstruction, temporal reconstruction and spectral reconstruction in this paper. This paper presents a comprehensive review of all three types of reconstruction. First, a systematic introduction and review of the achievements is provided, including the refined general mathematical framework and diagram for each of the three parts. Second, the applications in various areas (e.g., meteorology, ecology and environmental science) are introduced. Third, the challenges and recent advances of spatial-temporal-spectral information reconstruction are summarized, such as the efforts for dealing with abrupt land cover changes in spatial reconstruction, inconsistency in multi-scale data acquired by different sensors in temporal reconstruction, and point spread function (PSF) effect in spectral reconstruction. Finally, several thoughts are given for future prospects.

Leaf mass per area (LMA) and equivalent water thickness (EWT) are key indicators providing information on plant growth status and agricultural management, and their retrieval is commonly done through radiative transfer models (RTMs) such as the PROSPECT model. However, the PROSPECT model is frequently hampered by the ill-posed problem as a consequence of measurement and model uncertainties. Here, we propose a wavelength selection method to improve the inversion of EWT and LMA by integrating PROSPECT with a machine learning algorithm (Gaussian process regression (GPR); PROSPECT-GPR for short). The GPR model conducted sorting of wavelengths and the PROSPECT-D was used to determine the optimal number of characteristic wavelengths. The results demonstrated that the estimation of EWT (R² = 0.80; RMSE = 0.0021) and LMA (R² = 0.71; RMSE = 0.0021) using the proposed wavelengths and PROSPECT inversion all exhibited superior accuracy in comparison with those from previous studies. The efficacy of PROSPECT-GPR in exploring the spectral linkage among vegetation traits was demonstrated by selecting wavelengths associated with leaf structure parameter N and EWT (1368 nm) that turn out to contribute to the estimation of LMA. The findings lay a strong foundation for understanding the spectral linkage among vegetation traits, and the proposed wavelength selection method provides valuable insights for selecting informative spectral wavelengths for RTMs inversion and designing future remote sensors.

Monitoring and understanding forest dynamics is essential for environmental conservation and management. This is why the Swiss National Forest Inventory (NFI) provides countrywide vegetation height maps at a spatial resolution of 0.5 m. Its long update time of 6 years, however, limits the temporal analysis of forest dynamics. This can be improved by using spaceborne remote sensing and deep learning to generate large-scale vegetation height maps in a cost-effective way. In this paper, we present an in-depth analysis of these methods for operational application in Switzerland. We generate annual, countrywide vegetation height maps at a 10-m ground sampling distance for the years 2017–2020 based on Sentinel-2 satellite imagery. In comparison to previous works, we conduct a large-scale and detailed stratified analysis against a precise Airborne Laser Scanning reference dataset. This stratified analysis reveals a close relationship between the model accuracy and the topology, especially slope and aspect. We assess the potential of deep learning-derived height maps for change detection and find that these maps can indicate changes as small as 250 m². Larger-scale changes caused by a winter storm are detected with an F1-score of 0.77. Our results demonstrate that vegetation height maps computed from satellite imagery with deep learning are a valuable, complementary, cost-effective source of evidence to increase the temporal resolution for national forest assessments.

The forest reflectance and transmittance model (FRT) is applicable over a wide swath of boreal forest landscapes mainly because its stand-specific inputs can be generated from standard forest inventory variables. We quantified the accuracy of this model over an extensive region for the first time. This was done by carrying out a simulation study over a large number (12,369) of georeferenced forest plots from operational forest management inventories conducted in Southern Finland. We compared the FRT simulated bidirectional reflectance factors (BRF) with those measured by Landsat 8 satellite Operational Land Imager (OLI). We also quantified the relative importance of several explanatory factors that affected the magnitude of the discrepancy between the measured and simulated BRFs using a linear mixed effects modelling framework. A general trend of FRT overestimating BRFs is seen across all tree species and spectral bands examined: up to ∼0.05 for the red band, and ∼0.10 for the near infrared band. The important explanatory factors associated with the overestimations included the dominant tree species, understory type of the forest plot, timber volume (acts as a proxy for stand maturity), vegetation heterogeneity and time of the year. Our analysis suggests that approximately 20% of the error is caused by the non-representative spectra of canopy foliage and understory. Our results demonstrate the importance of collecting representative spectra from a diverse set of forest stands, and over the full range of seasons.

Precise quantification of forest fire impacts is critical for management strategies in support of post-fire mitigation. In this regard, optical remote sensing imagery in combination with spectral unmixing has been widely used to measure fire severity by means of fractional cover of photosynthetic vegetation (PV), non-photosynthetic vegetation (NPV), charcoal (CH) and further ground components such as ash, bare soil and rocks. However, most unmixing analyses have made use of a single post-fire image without accounting for the pre-fire state. We aim to assess fire severity from Sentinel-2 data using a bi-temporal spectral unmixing analysis that provides a quantitative fire impact description and is oriented towards the process of change by including pre-fire and post-fire information. Unmixing was based on Random Forest Regression (RFR) modeling using synthetic training data from a bi-temporal spectral library. We describe fire severity as changes associated with the combustion of photosynthetic vegetation (PV–CH fraction) and dieback of photosynthetic vegetation (PV-NPV fraction). Unburned forest was mapped as stable photosynthetic vegetation (PV-PV fraction). We evaluated our approach on a forest fire that burned in a temperate forest region in eastern Germany in 2018. Independent validation was carried out based on reference fractions obtained from very high-resolution (VHR) imagery such as Plante Scope, SPOT6, orthophotos, aerial photos, and Google Earth. The results underline the effectiveness of our unmixing approach, with Root Mean Squared Errors (RMSE) of 0.072 for PV-CH, 0.09 for PV-NPV, and 0.08 for PV-PV fractions. Most of the errors were caused by spectral similarity between charcoal and shadow effects caused by trees, and the coloring of foliage and NPV in the late phenological season of the post-fire Sentinel-2 image. Based on the two-dimensional feature space of PV-CH and PV-NPV fractions, we calculated two metrics to characterize fire impacts: distance, an indicator of disturbance severity (sum of combustion and dieback), and angle, a measure of disturbance composition (gradient between combustion and dieback). Furthermore, we compared the fraction-based metrics with the difference Normalized Burn Ratio (dNBR). Since the dNBR is most sensitive to combustion and presence of charcoal, it does not fully characterize fire-related vegetation loss associated with dieback. The bi-temporal fraction-based indices provide more ecologically meaningful information on fire severity, particularly for regions that are less prone to severe wildfires such as Central Europe.

Accurate maps of gains in tree cover are necessary to quantify carbon storage, wildlife habitat, and land use changes. Satellite-based mapping of emerging smallholder woodlots in heterogeneous landscapes of sub-Saharan Africa is challenging. Our goal was to evaluate the use of time series to detect and map small woodlots (<1 ha) in Tanzania. We distinguished woodlots from other land cover types by woodlots' distinct multi-year spectral time series. Woodlots exhibit greening from planting to maturity followed by browning at harvest. We compared two time series approaches: 1) a linear model of Tasseled Cap Wetness (TCW) and other indices, and 2) LandTrendr temporal segmentation metrics. The approaches had equivalent woodlot detection accuracy, but LandTrendr segments had lower accuracy for characterizing woodlot age. We tested the effect of the following factors on woodlot detection and mapping accuracy: the length of the time series (2009–2019), frequency of observations (all Landsat vs. only Landsat-8), spatial resolution (30-m Landsat vs. 10-m Sentinel-2), and woodlot age and size. Woodlot mapping accuracies were higher with longer time series (54% at 3-yrs vs 77% at 7-yrs). The accuracies also improved with more observations, especially when the time series was short (3-yrs Landsat-8 only: 54% vs. all-Landsat: 64%, p-value <0.001). Sentinel-2's higher spatial resolution minimized commission errors even for short time series. Finally, less than half of young and small (<0.4 ha) woodlots were detected, suggesting considerable omission errors in our and other woodlot maps. Our results suggest that the accurate detection of woodlots is possible by analyzing multi-year time series of Landsat and Sentinel-2 data. Given the region's woodlot boom, accurate maps are needed to better quantify woodlots' contribution to carbon sequestration, livelihoods enhancement, and landscape management.