Sentinel-2A Multispectral Instrument Research Articles

Modeling and Multi-Temporal Characterization of Total Suspended Matter by the Combined Use of Sentinel 2-MSI and Landsat 8-OLI Data: The Pertusillo Lake Case Study (Italy)

The total suspended matter (TSM) variability plays a crucial role in a lake’s ecological functioning and its biogeochemical cycle. Sentinel-2A MultiSpectral Instrument (MSI) and Landsat 8 Operational Land Instrument (OLI) data offer unique opportunities for investigating certain in-water constituents (e.g., TSM and chlorophyll-a) owing to their spatial resolution (10–60 m). In this framework, we assessed the potential of MSI–OLI combined data in characterizing the multi-temporal (2014–2018) TSM variability in Pertusillo Lake (Basilicata region, Southern Italy). We developed and validated a customized MSI-based TSM model (R2 = 0.81) by exploiting ground measurements acquired during specific measurement campaigns. The model was then exported as OLI data through an intercalibration procedure (R2 = 0.87), allowing for the generation of a TSM multi-temporal MSI–OLI merged dataset. The analysis of the derived multi-year TSM monthly maps showed the influence of hydrological factors on the TSM seasonal dynamics over two sub-regions of the lake, the west and east areas. The western side is more influenced by inflowing rivers and water level fluctuations, the effects of which tend to longitudinally decrease, leading to less sediment within the eastern sub-area. The achieved results can be exploited by regional authorities for better management of inland water quality and monitoring systems.

Mapping Seasonal Variations of Grazing Land Above-ground Biomass with Sentinel 2A Satellite Data

Seasonal variations have brought about significant changes in vegetation cover and spatial distribution in the past decade. Globally, grazing lands are experiencing a significant warming and drying process more especially the grazing lands in the Savannah and Sahel regions. This paper reports the study undertaken for mapping changes on the grass above ground biomass (GAB) due to these seasonal changes using Sentinel 2A Multispectral Instrument (MSI) data. Emphasising on the GAB, the main objective of this study is to map and model monthly GAB variations to their corresponding meteorological data. A set of selected widely used vegetation indices were applied to satellite data, and later were further regressed against corresponding in-situ GAB samples and weather data, hence, producing a predictor of GAB from satellite data. Sentinel 2A MSI data were acquired monthly from January to December 2018. Combined with precipitation and temperature data, the GAB variations on monthly scales were analysed. The results indicated that GAB determined and its seasonal variations shown good agreement (r = 0.8, p < 0.001) with corresponding in-situ verifications. Temperature was found inversely proportionally to GAB for the whole grazing calendar. Therefore, it was concluded that mapping GAB seasonal variations is achievable with Sentinel2 MSI, vast potential for input to grazing land management.

Cross-Calibration of MODIS Reflective Solar Bands With Sentinel 2A/2B MSI Instruments

Moderate resolution imaging spectroradiometer (MODIS) is a 36-band spectroradiometer that measures the Earth's surface from 0.4 to 14.4 μm at three spatial resolutions, 250 m (two bands), 500 m (five bands), and 1000 m (29 bands). The wide-scale use of the science products derived from the two MODIS instruments on the Terra and Aqua spacecrafts is a result of their excellent on-orbit performance, calibration stability, and accuracy through the life of the mission, making them a benchmark against which the performance of newer instruments is frequently evaluated. The recently launched multispectral instrument (MSI) aboard the Sentinel 2A and Sentinel 2B spacecrafts are part of the European Union's Copernicus program designed to acquire high spatial resolution imagery in the reflective spectrum from 0.4 to 2.2 μm. One of the popular techniques to evaluate the on-orbit calibration is by comparing the top-of-atmosphere (TOA) reflectance with an independent (well-calibrated) sensor while viewing a pseudoinvariant desert target, such as Libya 4. In this work, the TOA reflectances from Terra and Aqua MODIS and Sentinel 2A and Sentinel 2B MSI are compared using the same-day scenes from Libya 4. The corrections for spectral response function mismatch and the bidirectional reflectance distribution function (BRDF) are formulated and applied to obtain an effective TOA reflectance difference between the spectrally matching bands of these sensors. The availability of off-nadir MODIS overpass pairs with MSI facilitates the comparison across the entire MODIS scanangle range and in turn an on-orbit evaluation of the response versus scan-angle (RVS) corrections is performed. Additionally, each MODIS instrument is used as a transfer mechanism to evaluate the calibration differences between the two MSIs, with an agreement to within 1% observed between the two MSIs. The radiometric calibration differences between Terra MODIS and the two MSIs at nadir is generally within 4%, with only the red-band pair (Terra MODIS band 1 and MSI band 4) showing disagreement beyond 4%. In general, the reflectance ratios are in better agreement with Aqua MODIS than with Terra MODIS. A better agreement between MODIS and the two MSIs is observed at nadir, indicating some residual effects associated with the BRDF correction that are observed in the off-nadir scene pairs. Also included in this article is a description of the various uncertainties associated with this cross-calibration.

Assessment of Workflow Feature Selection on Forest LAI Prediction with Sentinel-2A MSI, Landsat 7 ETM+ and Landsat 8 OLI.

The European Space Agency (ESA)’s Sentinel-2A (S2A) mission is providing time series that allow the characterisation of dynamic vegetation, especially when combined with the National Aeronautics and Space Administration (NASA)/United States Geological Survey (USGS) Landsat 7 (L7) and Landsat 8 (L8) missions. Hybrid retrieval workflows combining non-parametric Machine Learning Regression Algorithms (MLRAs) and vegetation Radiative Transfer Models (RTMs) were proposed as fast and accurate methods to infer biophysical parameters such as Leaf Area Index (LAI) from these data streams. However, the exact design of optimal retrieval workflows is rarely discussed. In this study, the impact of five retrieval workflow features on LAI prediction performance of MultiSpectral Instrument (MSI), Enhanced Thematic Mapper Plus (ETM+) and Operational Land Imager (OLI) observations was analysed over a Dutch beech forest site for a one-year period. The retrieval workflow features were the (1) addition of prior knowledge of leaf chemistry (two alternatives), (2) the choice of RTM (two alternatives), (3) the addition of Gaussian noise to RTM produced training data (four and five alternatives), (4) possibility of using Sun Zenith Angle (SZA) as an additional MLRA training feature (two alternatives), and (5) the choice of MLRA (six alternatives). The features were varied in a full grid resulting in 960 inversion models in order to find the overall impact on performance as well as possible interactions among the features. A combination of a Terrestrial Laser Scanning (TLS) time series with litter-trap derived LAI served as independent validation. The addition of absolute noise had the most significant impact on prediction performance. It improved the median prediction Root Mean Square Error (RMSE) by 1.08 m2 m−2 when 5 % noise was added compared to inversions with 0 % absolute noise. The choice of the MLRA was second most important in terms of median prediction performance, which differed by 0.52 m2 m−2 between the best and worst model. The best inversion model achieved an RMSE of 0.91 m2 m−2 and explained 84.9% of the variance of the reference time series. The results underline the need to explicitly describe the used noise model in future studies. Similar studies should be conducted in other study areas, both forest and crop systems, in order to test the noise model as an integral part of hybrid retrieval workflows.

Radiometric Cross Calibration and Validation Using 4 Angle BRDF Model between Landsat 8 and Sentinel 2A

This work describes a proposed radiometric cross calibration between the Landsat 8 Operational Land Imager (OLI) and Sentinel 2A Multispectral Instrument (MSI) sensors. The cross-calibration procedure involves (i) correction of the MSI data to account for spectral band differences with OLI and (ii) normalization of Bidirectional Reflectance Distribution Function (BRDF) effects in the data from both sensors using a new model accounting for the view zenith/azimuth angles in addition to the solar zenith/view angles. Following application of the spectral and BRDF normalization, standard least-squares linear regression is used to determine the cross-calibration gain and offset in each band. Uncertainties related to each step in the proposed process are determined, as is the overall uncertainty associated with the complete processing sequence. Validation of the proposed cross-calibration gains and offsets is performed on image data acquired over the Algodones Dunes site. The results of this work indicate that the blue band has the most significant offset, requiring use of the estimated cross-calibration offset in addition to the estimated gain. The highest difference was observed in the blue and red bands, which are 2.6% and 1.4%, respectively, while other bands shows no significant difference. Overall, the net uncertainty in the proposed process was estimated to be on the order of 6.76%, with the largest uncertainty component due to each sensor’s calibration uncertainty on the order of 5% and 3% for the MSI and OLI, respectively. Other significant contributions to the uncertainty include seasonal changes in solar zenith and azimuth angles, target site nonuniformity, variability in atmospheric water vapor, and/or aerosol concentration.

Comparison of Different Multispectral Sensors for Photosynthetic and Non-Photosynthetic Vegetation-Fraction Retrieval

It is very difficult and complex to acquire photosynthetic vegetation (PV) and non-PV (NPV) fractions (fPV and fNPV) using multispectral satellite sensors because estimations of fPV and fNPV are influenced by many factors, such as background-noise interference of pixel-, spatial-, and spectral-scale effects. In this study, comparisons between Sentinel-2A Multispectral Instrument (S2 MSI), Landsat-8 Operational Land Imager (L8 OLI), and GF1 Wide Field View (GF1 WFV) sensors for retrieving sparse photosynthetic and non-photosynthetic vegetation coverage are presented. The analysis employed a linear spectral-mixture model (LSMM) and nonlinear spectral-mixture model (NSMM) to unmix pixels with different spectral and spatial resolution images based on field endmembers; the estimated endmember fractions were later validated with reference to fraction measurements. The results demonstrated that: (1) with higher spatial and spectral resolution, the S2 MSI sensor had a clear advantage for retrieving PV and NPV fractions compared to L8 OLI and GF1 WFV sensors; (2) through incorporating more red edge (RE) and near-infrared (NIR) bands, the accuracy of NPV fraction estimation could be greatly improved; (3) nonlinear spectral mixing effects were not obvious on the 10–30 m spatial scale for desert vegetation; (4) in arid regions, a shadow endmember is a significant factor for sparse vegetation coverage estimated with remote-sensing data. The estimated NPV fractions were especially affected by the shadow effects and could increase root mean square by 50%. The utilized approaches in the study could effectively assess the performance of major multispectral sensors to extract fPV and fNPV through the novel method of spectral-mixture analysis.

Sentinel-2A Image Fusion Using a Machine Learning Approach

The multispectral instrument (MSI) carried by Sentinel-2A has 13 spectral bands with various spatial resolutions (i.e., four 10-m, six 20-m, and three 60-m bands). A wide range of applications requires a 10-m resolution for all spectral bands, including the 20- and 60-m bands. To achieve this requirement, previous studies used conventional pansharpening techniques, which require a simulated 10-m panchromatic (PAN) band from four 10-m bands [blue, green, red, and near infrared (NIR)]. The simulated PAN band may not have all the information from the original four bands and may have no spectral response function that overlaps the 20- or 60-m bands to be sharpened, which may degrade fusion quality. This paper presents a machine learning method that can directly use the information from multiple 10-m resolution bands for fusion. The method first learns the spectral relationship between the 20- or 60-m band to be sharpened and the selected 10-m bands degraded to 20 or 60 m using the support vector regression (SVR) model. The model is then applied to the selected 10-m bands to predict the 10-m-resolution version of the 20- or 60-m band. The image degradation process was tuned to closely match the Sentinel-2A MSI modulation transfer function (MTF). We applied our method to three data sets in Guangzhou, China, New South Wales, Australia, and St. Louis, USA, and achieved better fusion results than other commonly used pansharpening methods in terms of both visual and quantitative factors.

Crop classification using spectral indices derived from Sentinel-2A imagery

ABSTRACTOptical remote sensing is one of the most attractive options for generating crop cover maps because it enables computation of vegetation indices, which are useful for assessing the condition of vegetation. The Sentinel-2A Multispectral Instrument (MSI), which is a multispectral sensor with 13 bands covering the visible, near infrared and short-wave infrared (SWIR) wavelength regions, offers a vast number of vegetation indices. Spectral indices, which are combinations of spectral measurements at different wavelengths, have been used in the previous studies and they sometimes contributed to improve classification accuracies. In this study, 91 published spectral indices were calculated from the MSI data. Additionally, classification algorithms are essential for generating accurate maps and the random forests classifier is one of which possesses the five hyperparameters were applied. The improvements in classification accuracies were confirmed achieving an overall accuracy of 93.1% based on the reflectance at 4 bands and 8 spectral indices.

Improving burn severity retrieval by integrating tree canopy cover into radiative transfer model simulation

Burn severity mapping greatly informs fire management and can be used to predict post-fire vegetation recovery. Satellite remote sensing is a cost-effective method for estimating burn severity, providing a comprehensive spatially explicit view of whole landscapes. However, the proportion of tree canopy cover (TCC) affects the reflectance signal, obscuring background char and ash. Consequently, traditional optical satellite remote sensing methods that do not account for variation in TCC misclassify burn severity, especially in areas with extremely low or high TCC. In this study, TCC data served to parameterize and constrain the inversion of the Forest Reflectance and Transmittance (FRT) radiative transfer model (RTM) to alleviate spectral confusion when retrieving burn severity. The methodology was evaluated using field measurements of burn severity for a series of wildfires in the fire-prone tropical savannas of northern Australia and the western United States. Burn severity classes were used for Australia while the Composite Burn Index (CBI) for US. Reflectance data from Sentinel-2A Multi-Spectral Instrument (MSI) and Landsat-5 Thematic Mapper (TM) corresponding to post-fire field survey dates were used to retrieve burn severity using FRT RTM (with and without using TCC information in its parameterization and inversion) and two standard empirical burn indices, dNBR and RdNBR, for comparison. Using FRT RTM without TCC constraint produced an overestimation for low burn severity in regions with low TCC and an underestimation for moderate and high burn severity in regions with high TCC. Burn severity estimation accuracy significantly improved by integrating TCC in the parameterization and inversion of FRT RTM. The overall accuracy in northern Australia increased from 65% to 81%, and the kappa coefficient increased from 0.35 to 0.55. In the western United States, R2 between estimated and observed CBI, increased from 0.33 to 0.54, root mean square error (RMSE) reduced from 0.53 to 0.43, and in all instances, the method performed better than dNBR and RdNBR. The method used in this study achieved more accurate burn severity mapping, thus assisting land managers to better understand post-fire vegetation resilience and forest management.

Monitoring of Urbanization and Analysis of Environmental Impact in Stockholm with Sentinel-2A and SPOT-5 Multispectral Data

There has been substantial urban growth in Stockholm, Sweden, the fastest-growing capital in Europe. The intensifying urbanization poses challenges for environmental management and sustainable development. Using Sentinel-2 and SPOT-5 imagery, this research investigates the evolution of land-cover change in Stockholm County between 2005 and 2015, and evaluates urban growth impact on protected green areas, green infrastructure and urban ecosystem service provision. One scene of 2015 Sentinel-2A multispectral instrument (MSI) and 10 scenes of 2005 SPOT-5 high-resolution instruments (HRI) imagery over Stockholm County are classified into 10 land-cover categories using object-based image analysis and a support vector machine algorithm with spectral, textural and geometric features. Reaching accuracies of approximately 90%, the classifications are then analyzed to determine impact of urban growth in Stockholm between 2005 and 2015, including land-cover change statistics, landscape-level urban ecosystem service provision bundle changes and evaluation of regional and local impact on legislatively protected areas as well as ecologically significant green infrastructure networks. The results indicate that urban areas increased by 15%, while non-urban land cover decreased by 4%. In terms of ecosystem services, changes in proximity of forest and low-density built-up areas were the main cause of lowered provision of temperature regulation, air purification and noise reduction. There was a decadal ecosystem service loss of 4.6 million USD (2015 exchange rate). Urban areas within a 200 m buffer zone around the Swedish environmental protection agency’s nature reserves increased 16%, with examples of urban areas constructed along nature reserve boundaries. Urban expansion overlapped the deciduous ecological corridor network and green wedge/core areas to a small but increasing degree, often in close proximity to weak but important green links in the landscape. Given these findings, increased conservation/restoration focus on the region’s green weak links is recommended.

Optimal season for discriminating C3 and C4 grass functional types using multi-date Sentinel 2 data

ABSTRACTThe ability of remote sensing systems to optimally discriminate and map C3 and C4 grass species varies over time, due to environmental changes, which influence their phenological, physiological and morphological characteristics. In this regard, the discrimination of C3 and C4 grasses is insufficient when using a single image acquired at a specific period. In this study, multi-date Sentinel 2A MultiSpectral Instrument (MSI) data were explored to determine the optimal period for classifying and mapping Festuca costata, C3 and Themeda Triandra, C4 grasses in the montane grasslands of South Africa. The study further assessed how seasonal variations in species classification can be explained by climatic variability (rainfall and temperature). Results showed that image acquisition dates influence the discrimination accuracy, spatial representation of the two grass species, as well as the performance of spectral bands. The winter period also presents a better temporal window for discriminating C3 and C4 target grass species, with higher overall classification accuracies (between 91.8% and 95.3%), than summer (between 81.4% and 90.3%). Lower omission (between 2.8% and 11.6%) and commission (between 2.5% and 14.2%) errors were also observed when discriminating using winter images, as compared to those acquired in summer. Summer images showed large grass species areal coverage (e.g. in November and March, C3 and C4 covered ±25%), whereas in winter (mainly August), a notable decrease was observed. Overall, findings of the study have demonstrated the relevance of multi-date Sentinel data in discriminating C3 and C4 grass species. There is, however, a need to explore the classification ability of Sentinel 2 derivatives, especially during early summer and winter fall.

An inter-comparison exercise of Sentinel-2 radiometric validations assessed by independent expert groups

Copernicus is the European Union's Earth Observation and Monitoring programme, delivering free access to operational and historical environmental data to support applications in a wide range of societal benefit areas. To allow meaningful long-term environmental monitoring and robust decision-making, it is essential to ensure that satellite-retrieved products are of high quality and consistency. This paper describes the outputs of an international workshop on the radiometric calibration validation of the Copernicus Sentinel-2A and Sentinel-2B Multi-Spectral Instrument. A wide range of vicarious methodologies have been applied independently and then compared per type of target. All methods agree on the good radiometric performance of both Sentinel-2A and Sentinel-2B with respect to the mission requirements as well as on evidence of a slight bias between the two instruments. Comparisons of all these results are discussed to highlight the benefits and advantages of the methods as well as to propose potential improvements either for the methods themselves and/or for the comparison exercise.

Direct, ECOC, ND and END Frameworks—Which One Is the Best? An Empirical Study of Sentinel-2A MSIL1C Image Classification for Arid-Land Vegetation Mapping in the Ili River Delta, Kazakhstan

To facilitate the advances in Sentinel-2A products for land cover from Moderate Resolution Imaging Spectroradiometer (MODIS) and Landsat imagery, Sentinel-2A MultiSpectral Instrument Level-1C (MSIL1C) images are investigated for large-scale vegetation mapping in an arid land environment that is located in the Ili River delta, Kazakhstan. For accurate classification purposes, multi-resolution segmentation (MRS) based extended object-guided morphological profiles (EOMPs) are proposed and then compared with conventional morphological profiles (MPs), MPs with partial reconstruction (MPPR), object-guided MPs (OMPs), OMPs with mean values (OMPsM), and object-oriented (OO)-based image classification techniques. Popular classifiers, such as C4.5, an extremely randomized decision tree (ERDT), random forest (RaF), rotation forest (RoF), classification via random forest regression (CVRFR), ExtraTrees, and radial basis function (RBF) kernel-based support vector machines (SVMs) are adopted to answer the question of whether nested dichotomies (ND) and ensembles of ND (END) are truly superior to direct and error-correcting output code (ECOC) multiclass classification frameworks. Finally, based on the results, the following conclusions are drawn: 1) the superior performance of OO-based techniques over MPs, MPPR, OMPs, and OMPsM is clear for Sentinel-2A MSIL1C image classification, while the best results are achieved by the proposed EOMPs; 2) the superior performance of ND, ND with class balancing (NDCB), ND with data balancing (NDDB), ND with random-pair selection (NDRPS), and ND with further centroid (NDFC) over direct and ECOC frameworks is not confirmed, especially in the cases of using weak classifiers for low-dimensional datasets; 3) from computationally efficient, high accuracy, redundant to data dimensionality and easy of implementations points of view, END, ENDCB, ENDDB, and ENDRPS are alternative choices to direct and ECOC frameworks; 4) surprisingly, because in the ensemble learning (EL) theorem, “weaker” classifiers (ERDT here) always have a better chance of reaching the trade-off between diversity and accuracy than “stronger” classifies (RaF, ExtraTrees, and SVM here), END with ERDT (END-ERDT) achieves the best performance with less than a 0.5% difference in the overall accuracy (OA) values, but is 100 to 10000 times faster than END with RaF and ExtraTrees, and ECOC with SVM while using different datasets with various dimensions; and, 5) Sentinel-2A MSIL1C is better choice than the land cover products from MODIS and Landsat imagery for vegetation species mapping in an arid land environment, where the vegetation species are critically important, but sparsely distributed.

New Approach for Temporal Stability Evaluation of Pseudo-Invariant Calibration Sites (PICS)

Pseudo-Invariant Calibration Sites (PICS) are one of the most popular methods for in-flight vicarious radiometric calibration of Earth remote sensing satellites. The fundamental question of PICS temporal stability has not been adequately addressed. However, the main purpose of this work is to evaluate the temporal stability of a few PICS using a new approach. The analysis was performed over six PICS (Libya 1, Libya 4, Niger 1, Niger 2, Egypt 1 and Sudan 1). The concept of a “Virtual Constellation” was developed to provide greater temporal coverage and also to overcome the dependence limitation of any specific characteristic derived from one particular sensor. TOA reflectance data from four sensors consistently demonstrating “stable” calibration to within 5%—the Landsat 7 ETM+ (Enhanced Thematic Mapper Plus), Landsat 8 OLI (Operational Land Imager), Terra MODIS (Moderate Resolution Imaging Spectroradiometer) and Sentinel-2A MSI (Multispectral Instrument)–were merged into a seamless dataset. Instead of using the traditional method of trend analysis (Student’s T test), a nonparametric Seasonal Mann-Kendall test was used for determining the PICS stability. The analysis results indicate that Libya 4 and Egypt 1 do not exhibit any monotonic trend in six reflective solar bands common to all of the studied sensors, indicating temporal stability. A decreasing monotonic trend was statistically detected in all bands, except SWIR 2, for Sudan 1 and the Green and Red bands for Niger 1. An increasing trend was detected in the Blue band for Niger 2 and the NIR band for Libya 1. These results do not suggest abandoning PICS as a viable calibration source. Rather, they indicate that PICS temporal stability cannot be assumed and should be regularly monitored as part of the sensor calibration process.

Empirical Absolute Calibration Model for Multiple Pseudo-Invariant Calibration Sites

This work extends an empirical absolute calibration model initially developed for the Libya 4 Pseudo-Invariant Calibration Site (PICS) to five additional Saharan Desert PICS (Egypt 1, Libya 1, Niger 1, Niger 2, and Sudan 1), and demonstrates the efficacy of the resulting models at predicting sensor top-of-atmosphere (TOA) reflectance. It attempts to generate absolute calibration models for these PICS that have an accuracy and precision comparable to or better than the current Libya 4 model, with the intent of providing additional opportunities for sensor calibration. In addition, this work attempts to validate the general applicability of the model to other sites. The method uses Terra Moderate Resolution Imaging Spectroradiometer (MODIS) as the reference radiometer and Earth Observing-1 (EO-1) Hyperion image data to provide a representative hyperspectral reflectance profile of the PICS. Data from a region of interest (ROI) in an “optimal region” of 3% temporal, spatial, and spectral stability within the PICS are used for developing the model. The developed models were used to simulate observations of the Landsat 7 (L7) Enhanced Thematic Mapper Plus (ETM+), Landsat 8 (L8) Operational Land Imager (OLI), Sentinel 2A (S2A) MultiSpectral Instrument (MSI) and Sentinel 2B (S2B) MultiSpectral Instrument (MSI) from their respective launch date through 2018. The models developed for the Egypt 1, Libya 1 and Sudan 1 PICS have an estimated accuracy of approximately 3% and precision of approximately 2% for the sensors used in the study, comparable to the current Libya 4 model. The models developed for the Niger 1 and Niger 2 sites are significantly less accurate with similar precision.

Mapping stocks of soil total nitrogen using remote sensing data: A comparison of random forest models with different predictors

Estimation of soil total nitrogen (STN) is important for quantifying the spatial distribution of soil N (Nitrogen) stocks, determining the amount absorbed by crops and avoiding yield losses and environmental pollution. The objective of this research was to determine the spatial distribution of STN using a remote sensing approach and to investigate the importance of different predictors under two conditions: the bare soil condition and the vegetated condition. We also investigated whether the Sentinel-2A red-edge bands and relative spectral indices were suitable for STN estimation. Soil data were collected from the topsoil (0–20 cm) at 104 sampling sites on 06/15/2016 from farmland in a black soil region in Northeast China. Two Sentinel-2A Multispectral Instrument (MSI) remote sensing images were acquired on 05/26/2016 and 08/03/2016 to represent the two conditions. Environmental variables included the terrain attributes, temperature and precipitation. Then, 21 predictors, including the original bands (O), normal spectral indices (S), red-edge indices (R) and environmental variables (E), were employed to estimate the spatial distribution of the STN content using a random forest (RF) model. Finally, different predictors were combined to construct RF models, and the prediction model with the best performance was selected to determine the spatial pattern of the STN content. The results showed that the predictors had different levels of importance under the two conditions. However, most environmental variables and normal spectral indices always play a significant role in the estimation of STN. The model with the combination of the original bands, normal spectral indices, red-edge indices and environmental variables (O + S + E + R) under the bare soil condition had the best prediction performance, and the combination of the original bands, normal spectral indices and red-edge indices (O + S + R) model had a performance similar to the O + S + E + R model. Therefore, the selection of suitable predictors is necessary to predict STN. The spatial pattern of STN was related to the crop type and elevation of the study area. The results of this study suggested that the proposed RF-based remote sensing method was able to accurately capture the variation in STN and that the performance of the prediction model can be improved by providing enough types of suitable predictors.

Land use and land cover change detection using geospatial techniques in the Sikkim Himalaya, India

Mapping and monitoring of land use land cover (LULC) changes in the Himalayas is vital for sustainable development, planning and management. Based on remote sensing (RS) and geographic information system (GIS) techniques, the study is an attempt to monitor the changes in LULC patterns of Rani Khola watershed of Sikkim Himalaya for the periods 1988–1996, 1996–2008 and 2008–2017. Images from Landsat-5 Thematic Mapper (TM) and Sentinel 2A (Multispectral Instrument) MSI data were used to extract land cover maps. Supervised classification using Maximum Likelihood Classifier (MLC) was applied to prepare LULC maps of the watershed. The accuracy of the classified map was assessed using a High Resolution Planet scope image and ground realities have been verified and ascertained through field observations and site specific interviews. As a result of policy changes and traditional agroforestry systems, LULC in the study watershed has undergone a series of complicated changes over the past three decades. Six major LULC classes viz; agriculture, barren land, built-up area, dense forest, open forests and water bodies have been identified and indicate that major land use in the watershed is forestry. Results shows, dense forest, built-up area and water bodies have increased by 16.40% (41.76 km2), 2.13% (5.41 km2) and 0.11% (0.28 km2) while open forest, agriculture and barren land have decreased by −13.98% (−35.59 km2), 2.83% (−7.22 km2) and −1.82% (0.4.64 km2) respectively. The analysis and findings of the study highlights important policy implications for the sustainable LULC management in the Rani Khola watershed of the Sikkim Himalaya.

Study of permafrost distribution in Sikkim Himalayas using Sentinel-2 satellite images and logistic regression modelling

In this study, Sentinel-2A (S-2A) Multi-Spectral Instrument (MSI) Level 1C (L1C) products are used for identification and mapping of rock glaciers within a transect under observation in Sikkim Himalayas. The mapped rock glaciers were classified on the basis of their activity (intact or relict) and their origin (glacier-derived or talus-derived). Logistic regression models were applied to calculate the probability of permafrost distribution within transect under observation. We observed that freely available high resolution S-2A MSI L1C products are a valuable addition to the already existing freely available high resolution satellite images that can be exploited to study rock glaciers in Himalayan region. The distribution of rock glaciers in Sikkim Himalayas is found to be strongly controlled by elevation and aspect. Topoclimatic conditions above 5000 m a.s.l. are favourable for the presence of permafrost in Sikkim Himalayas. More glacier-derived rock glaciers than talus-derived rock glaciers were observed within transect under observation. Outputs from logistic regression models were capable to reasonably estimate the spatially-distributed probability of permafrost occurrence in the region. However, outputs from these models depended upon the selection of the initiation line location points of rock glaciers and topoclimatic and topographic variables used as predictors for model computation. Based on probability distribution maps, more than 60% (>800 km2) area of transect under observation could contain permafrost. This high areal abundance of permafrost in Sikkim Himalayas suggests increased risks of thawing permafrost under the influence of climate warming in future.

Combinational shadow index for building shadow extraction in urban areas from Sentinel-2A MSI imagery

Images from Multispectral Instrument (MSI) on Sentinel-2A are useful for studies of urban environments and their spatio-temporal changes. However, the frequent occurrence of building shadows in urban areas brings about great challenges in urban studies. Existing building shadow indices are not effective for Sentinel-2A MSI images due to that these indices do not make full use of the rich spectral information contained in Sentinel-2A. In this study, we propose a combinational shadow index (CSI) to address this challenge. In the formulation of CSI, three features, including the proposed shadow enhancement index (SEI), the normalized difference water index (NDWI) and the NIR band (B8), were combined to separate spectrally similar objects, such as building shadows, water and low albedo features at the Earth surface. The accuracy and robustness of CSI were tested by using six data sets from four cities, including Beijing, Shanghai, Guangzhou and Shenzhen in China. The performance of CSI was compared with three existing shadow indices, i.e., P algorithm, the normalized saturation-value difference index (NSVDI) and the shadow index (SI). Results show that CSI can detect building shadows with fine structures in both clear and cloudy images more effectively and worked well on large areas too. CSI can separate shadows from water and low albedo features as measured by a spectral discrimination index (SDI). Compared with the existing building shadow indices, CSI can improve the performance of building shadow detection by combining the feature information of SEI, NDWI and NIR band, and yielded satisfactory results for extracting building shadows from Sentinel-2A MSI imagery.

Shelterbelt Agroforestry Systems Inventory and Removal Analyzed by Object-based Classification of Satellite Data in Saskatchewan, Canada

Shelterbelt agroforestry systems inventory is challenging given their narrow linear feature and extensive distribution. The objective of this study was to evaluate the capability of Sentinel-2A Multispectral Instrument (MSI) and Sentinel-1B Synthetic Aperture Radar (SAR) imagery in (i) delineating shelterbelt tree rows on managed agricultural land in Saskatchewan, Canada, and (ii) detecting shelterbelt removal during the period 2008–2016. Contrast split segmentation for the normalized difference vegetation index and Gaussian filter (line filter) data sets from SAR were used to delineate feature borders. Several feature variables from the spectral bands of MSI were used as inputs for an object-based classification using the random forest classifier. A resulting land cover map, including the linear features of existing shelterbelts, was created with an overall accuracy of 80% and kappa value of 0.69. Shelterbelt change detection analysis using the land cover map (2016) and a legacy shelterbelt inventory map (2008) showed that 354 km of shelterbelts were removed within the study area (1,400 km2), accounting for 29.8% of the total shelterbelt length present in 2008. Our results demonstrated that the combination of Sentinel SAR and MSI imagery can provide sufficient information for mapping future shelterbelt planting, as well as allow the detection of shelterbelt removal.