Sentinel-2 Time Series Data Research Articles

Crops planting structure and karst rocky desertification analysis by Sentinel-1 data

Abstract Accurate crop planting structure (CPS) information and its relationship with the surrounding special environment can provide strong support for the adjustment of agricultural structure in areas with limited cultivated land resources, and it will help regional food security, social economy, and ecological balance adjustment. However, due to the perennial cloudy, rainy, and scattered arable land in Karst mountainous areas, the monitoring of planting structure by traditional remote sensing methods is greatly limited. In this regard, we focus on synthetic aperture radar (SAR) remote sensing, which can penetrate clouds and rain, without light constraints to image. In this article, based on parcel-based temporal sequence SAR, the CPS in South China karst area was extracted by deep learning technology, and the spatial coupling relationship between CPS and karst rocky desertification (KRD) was analyzed. The results showed that: (a) The overall accuracy of CPS classification was 75.98%, which proved that the geo-parcel-based time series SAR has a good effect for the CPS mapping in the karst mountainous areas; (b) Through the analysis of the spatial relationship between the planting structure and KRD, we found that the lower KRD level caused the simpler CPS and the higher KRD grade caused more complex CPS and more richer landscape types. The spatial variation trend of CPS landscape indicates the process of water shortage and the deepening of KRD in farmland; (c) The landscape has higher connectivity (Contagion Index, CI 0.52–1.73) in lower KRD level and lower connectivity (CI 0.83–2.05) in higher KRD level, which shows that the degree of fragmentation and connection of CPS landscape is positively proportional to the degree of KRD. In this study, the planting structure extraction of crops under complex imaging environment was realized by using the farmland geo-parcels-based time series Sentinel-1 data, and the relationship between planting structure and KRD was analyzed. This study provides a new idea and method for the extraction of agricultural planting structure in the cloudy and rainy karst mountainous areas of Southwest China. The results of this study have certain guiding significance for the adjustment of regional agricultural planting structure and the balance of regional development.

Fine-Resolution Mapping of Pan-Arctic Lake Ice-Off Phenology Based on Dense Sentinel-2 Time Series Data

The timing of lake ice-off regulates biotic and abiotic processes in Arctic ecosystems. Due to the coarse spatial and temporal resolution of available satellite data, previous studies mainly focused on lake-scale investigations of melting/freezing, hindering the detection of subtle patterns within heterogeneous landscapes. To fill this knowledge gap, we developed a new approach for fine-resolution mapping of Pan-Arctic lake ice-off phenology. Using the Scene Classification Layer data derived from dense Sentinel-2 time series images, we estimated the pixel-by-pixel ice break-up end date information by seeking the transition time point when the pixel is completely free of ice. Applying this approach on the Google Earth Engine platform, we mapped the spatial distribution of the break-up end date for 45,532 lakes across the entire Arctic (except for Greenland) for the year 2019. The evaluation results suggested that our estimations matched well with both in situ measurements and an existing lake ice phenology product. Based on the generated map, we estimated that the average break-up end time of Pan-Arctic lakes is 172 ± 13.4 (measured in day of year) for the year 2019. The mapped lake ice-off phenology exhibits a latitudinal gradient, with a linear slope of 1.02 days per degree from 55°N onward. We also demonstrated the importance of lake and landscape characteristics in affecting spring lake ice melting. The proposed approach offers new possibilities for monitoring the seasonal Arctic lake ice freeze–thaw cycle, benefiting the ongoing efforts of combating and adapting to climate change.

Aboveground Biomass Estimation in Short Rotation Forest Plantations in Northern Greece Using ESA’s Sentinel Medium-High Resolution Multispectral and Radar Imaging Missions

Plantations of fast-growing forest species such as black locust (Robinia Pseudoacacia) can contribute to energy transformation, mitigate industrial pollution, and restore degraded, marginal land. In this study, the synergistic use of Sentinel-2 and Sentinel-1 time series data is explored for modeling aboveground biomass (AGB) in black locust short-rotation plantations in northeastern Greece. Optimal modeling dates and EO sensor data are also identified through the analysis. Random forest (RF) models were originally developed using monthly Sentinel-2 spectral indices, while, progressively, monthly Sentinel-1 bands were incorporated in the statistical analysis. The highest accuracy was observed for the models generated using Sentinel-2 August composites (R2 = 0.52). The inclusion of Sentinel-1 bands in the spectral indices’ models had a negligible effect on modeling accuracy during the leaf-on period. The correlation and comparative performance of the spectral indices in terms of pairwise correlation with AGB varied among the phenophases of the forest plantations. Overall, the field-measured AGB in the forest plantations plots presented a higher correlation with the optical Sentinel-2 images. The synergy of Sentinel-1 and Sentinel-2 data proved to be a non-efficient approach for improving forest biomass RF models throughout the year within the geographical and environmental context of our study.

Mapping European Rice Paddy Fields Using Yearly Sequences of Spaceborne Radar Reflectivity: A Case Study in Italy

Although a vast literature exists on satellite-based mapping of rice paddy fields in Asia, where most of the global production takes place, little has been produced so far that focuses on the European context. Detection and mapping methods that work well in the Asian context will not offer the same performance in Europe, where different seasonal cycles, environmental contexts, and rice varieties make distinctive features dissimilar to the Asian case. In this context, water management is a key clue; watering practices are distinctive for rice with respect to other crops, and within rice there exist diverse cultivation practices including organic and non-organic approaches. In this paper, we focus on satellite-observed water management to identify rice paddy fields cultivated with a traditional agricultural approach. Building on established research results, and guided by the output of experiments on real-world cases, a new method for analyzing time-series of Sentinel-1 data has been developed, which can identify traditional rice fields with a high degree of reliability. Typical watering practices for traditional rice cultivation leave distinctive marks on the yearly sequence of spaceborne radar reflectivity that are identified by the proposed classifier. The method is tested on a small sample of rice paddy fields, built by direct collection of ground reference information. Automated setting of parameters was sufficient to achieve accuracy values beyond 90%, and scanning of a range of values led to touch full score on an independent test set. This work is a part of a broader initiative to build space-based tools for collecting additional pieces of evidence to support food chain traceability; the whole system will consider various parameters, whose analysis procedures are still at their early stages of development.

Mapping saffron fields and their ages with Sentinel-2 time series in north-east Iran

Saffron (Crocus sativus L.) is the most expensive spice worldwide and is predominantly produced in the Khorasan Province situated in north-east Iran. Climatic shifts and lowering groundwater tables negatively affect saffron yields in this region, which are determined by environmental factors, agronomical practices, and crop age. Nonetheless, spatially explicit information on changes in saffron cultivation is scarce, underlining a need for better monitoring tools. This study aims to evaluate the utility of Sentinel-2 (S2) time series in accurately mapping saffron fields and their ages (i.e., how many years saffron was cultivated in a field), based on its unique phenology. To separate saffron from other land covers, we first derived 252 spectral-temporal features by calculating 21 spectral features (10 individual bands plus 11 vegetation indices) for each of the 12 months. A Random Forest (RF) algorithm was then used in combination with field data to retain only features of high importance for saffron classification. These features comprised vegetation indices that incorporated spectral information from red, and near- and shortwave infrared bands during the phenological phases of the rapid green-up (February to March) and the dormant period (August to October). The RF classifier resulted in a saffron map for the year 2019 with a high classification accuracy based on these features. Compared against an independent in-situ saffron field dataset, 87.6% of the existing fields were correctly classified as saffron. To assess saffron field ages, we analysed the spectral separability of different age groups using the NDVI time series. We found that NDVI levels between December and May allowed for effectively separating 1st, 2nd, 3rd, 4th-6th, and 7th-8th year saffron fields. An RF-based classification of field ages resulted in an overall accuracy of 86.8%. This study demonstrated that S2 time series data allow for accurately mapping saffron fields and their age groups. Our findings provide a solid basis for mapping saffron across larger areas and for monitoring changes in saffron distribution. Such information is crucial for understanding how anthropogenic and climate change impacts will affect the future of saffron cultivation.

Sentinel-1 interferometric coherence and backscattering analysis for crop monitoring

In this study, we investigate the synergic use of synthetic aperture radar (SAR) backscattering (i.e., sigma nought σ0) and InSAR coherence (γ) maps as a tool for crop growth monitoring. Experiments were carried out using Sentinel-1 TOPS SAR data and field observations in one of the State General Directorate of Agriculture Enterprise farms in Konya (Central Turkey). The phenological stages of maize, sunflower, and wheat have been analyzed and compared to coherence and backscatter time series of Sentinel-1 data on multiple tracks and polarizations. The results evidence a strong correlation between different phenological stages of the crops and the InSAR coherence. Specifically, the observed coherence values are the highest for the maize (γasc, desc = 0.47) and sunflower (γasc = 0.49, γdesc = 0.48) after plowing the fields and seeding the crops. The coherence decreases with the plants' growth and reaches the lowest values for maize, sunflower, and wheat (γ = 0.08, γ = 0.09 and γ = 0.07, respectively) when the ground is completely covered by plants. Then, a coherence increase is observed after the harvesting time (γ = 0.51, γ = 0.50 and γ = 0.42 for maize, sunflower, and wheat, respectively). In terms of multi-temporal SAR backscattering, we find significant changes of the σ0 values during the crops' growing stages due to the changes in their leaf geometry and physical structure. The highest σ0 values for the maize, sunflower, and wheat are obtained as −9.18 dB, −5.24 dB and −10.05 dB, respectively, for the ascending orbit, in mid growing stages. Results show the improved capacity of SAR-driven measurements for agriculture monitoring and precise farming activities when InSAR coherence and backscattering are synergistically used. Specifically, the coherence allows estimating the main growth stages of the different crop types. Moreover, SAR backscattering provides reliable information on the whole growth stages during the agricultural season, and it might be profitably exploited for crop assessment.

Mapping Paddy Field Extent and Temporal Pattern Variation in a Complex Terrain Area using Sentinel 1-Time Series Data: Case Study of Magelang District, Indonesia

The availability of free Synthetic Aperture Radar (SAR) data of Sentinel 1A/B, with the high temporal resolution, has provoked the usage of time-series backscatter values from the SAR data for mapping paddy field extent and crop phenology. However, paddy field extent mapping over complex terrain areas is rarely conducted, and the effect of terrain shadows on the accuracy of paddy field classification has not been addressed yet. This study attempted to identify the effect of terrain shadows on the paddy field mapping accuracy using monthly median composites Sentinel S1A data in 2018 and to perform the effort to minimize the misclassification by incorporating dem-derived terrain ruggedness index (TRI) in the random forest classifier. Lastly, this research also aimed to identify the general variation of the time-series backscatter of VV-and-VH polarization in the paddy field by employing unsupervised K-means classification. Our study showed that terrain shadow contributed to the misclassification of the identified paddy field due to the similar low backscatter values to the paddy field. Incorporating TRI can eliminate the misclassified paddy field, which subsequently increased the accuracy of paddy field mapping by 8 – 9% (user accuracy) and 4 – 5% (producer accuracy). In addition, means of clustered classes from unsupervised classification over paddy field areas generated temporal patterns related to the cropping frequencies and agreement with the weather pattern of the study area. This indicated that the cropping phase of paddy could be identified using the temporal pattern from 12-month median composites of the radar backscatter.

Assimilation of Remote Sensing Data into Crop Growth Model for Yield Estimation: A Case Study from India

Crop yield estimation is important to inform logistics management such as the prescription of nutrient inputs, financing, storage and transport, marketing as well as to inform for crop insurance appraisals due to loss incurred by abiotic and biotic stresses. In this study, we used a suite of methods to assess yields at the village level (< 5 km2) using remote sensing technology and crop modeling in Indian states of Telangana, Andhra Pradesh and Odisha. Remote sensing products were generated using Sentinel-2 and Landsat 8 time series data and calibrated with data collected from farmers’ fields. We derived maps showing spatial variation in crop extent, crop growth stages and leaf area index (LAI), which are crucial in yield assessment. Crop classification was performed on Sentinel-2 time series data using spectral matching techniques (SMTs) and crop management information collected from field surveys along with ground data. The locations of crop cutting experiments (CCEs) was identified based on crop extent maps. LAI was derived based on the SAVI (soil-adjusted vegetation index) equation were using Landsat 8-time series data. We used the technique of re-parametrization of crop simulation models based on the several iterations using remote sensing leaf area index (LAI). The data assimilation approach helps in fine-tuning the initial parameters of the crop growth model and improving simulation with the help of remotely sensed observations. Results clearly show a good correlation between observed and simulated crop yields (R2 is greater than 0.7) for all the crops studied. Our study showed that by assimilation of remotely sensed data in to crop models, crop yields at harvest could be successfully predicted.

An automated early-season method to map winter wheat using time-series Sentinel-2 data: A case study of Shandong, China

Timely and accurate information on winter wheat distribution and planting area is of great significance to food security, policy-making, and ecological function evaluation. However, several problems exist in the traditional winter wheat mapping approaches using remote sensing data, such as the limited spatial resolution of the remote sensing image data, the utilization of full-season remote sensing data, and the heavy dependence on training data. In this context, we propose a method based on the Sentinel-2 time series data with a 10-m spatial resolution to map winter wheat in Shandong, China. This is a novel, easy-to-operate, and effective mapping method, which is called the automated early-season method to map winter wheat using the Sentinel-2 data (AEMMS). The model is based on the assumption that the biomass accumulated by winter crops (mainly consisting of winter wheat and garlic in Shandong province) is gradually increasing and other vegetation is gradually decreasing in the early-season phenological window phase. In addition, winter wheat is the crop that accumulates the most biomass among all the winter crops in Shandong province, and the normalized difference vegetation index (NDVI) value of winter wheat is generally higher than that of garlic. We designe five phenological metrics and a series of classification rules for winter wheat discrimination. The AEMMS method has the following advantages: (1) it achieves high spatial resolution winter wheat mapping with a 10-m spatial resolution; (2) it is an early-season mapping method, which provides winter wheat maps nearly 5 months before harvest; and (3) it is automatic and needs no training sample data. The AEMMS method was applied in Shandong, China, to discriminate winter wheat for the 2017–2018 season. Winter wheat areas were derived in all 17 of the municipal administrative regions of Shandong province, and a strong correlation was observed between the derived winter wheat areas and the official statistics, with the coefficient of determination reaching 0.8973. A high mapping accuracy was also achieved in Jiaxiang County using the AEMMS method, with an overall accuracy of 97.80% and a kappa coefficient of 0.9368.

Estimating Pasture Biomass Using Sentinel-2 Imagery and Machine Learning

Effective dairy farm management requires the regular estimation and prediction of pasture biomass. This study explored the suitability of high spatio-temporal resolution Sentinel-2 imagery and the applicability of advanced machine learning techniques for estimating aboveground biomass at the paddock level in five dairy farms across northern Tasmania, Australia. A sequential neural network model was developed by integrating Sentinel-2 time-series data, weekly field biomass observations and daily climate variables from 2017 to 2018. Linear least-squares regression was employed for evaluating the results for model calibration and validation. Optimal model performance was realised with an R2 of ≈0.6, a root-mean-square error (RMSE) of ≈356 kg dry matter (DM)/ha and a mean absolute error (MAE) of 262 kg DM/ha. These performance markers indicated the results were within the variability of the pasture biomass measured in the field, and therefore represent a relatively high prediction accuracy. Sensitivity analysis further revealed what impact each farm’s in situ measurement, pasture management and grazing practices have on the model’s predictions. The study demonstrated the potential benefits and feasibility of improving biomass estimation in a cheap and rapid manner over traditional field measurement and commonly used remote-sensing methods. The proposed approach will help farmers and policymakers to estimate the amount of pasture present for optimising grazing management and improving decision-making regarding dairy farming.

The 10-m crop type maps in Northeast China during 2017\u20132019

Northeast China is the leading grain production region in China where one-fifth of the national grain is produced; however, consistent and reliable crop maps are still unavailable, impeding crop management decisions for regional and national food security. Here, we produced annual 10-m crop maps of the major crops (maize, soybean, and rice) in Northeast China from 2017 to 2019, by using (1) a hierarchical mapping strategy (cropland mapping followed by crop classification), (2) agro-climate zone-specific random forest classifiers, (3) interpolated and smoothed 10-day Sentinel-2 time series data, and (4) optimized features from spectral, temporal, and texture characteristics of the land surface. The resultant maps have high overall accuracies (OA) spanning from 0.81 to 0.86 based on abundant ground truth data. The satellite estimates agreed well with the statistical data for most of the municipalities (R2 ≥ 0.83, p < 0.01). This is the first effort on regional annual crop mapping in China at the 10-m resolution, which permits assessing the performance of the soybean rejuvenation plan and crop rotation practice in China.

Thermal Noise Removal From Polarimetric Sentinel-1 Data

This study proposes, for the first time, an approach to remove thermal noise from the wave coherency matrix, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathrm {C_{2}}$ </tex-math></inline-formula> , estimated from single-look complex dual-polarization Interferometric Wide Swath mode Sentinel-1 synthetic aperture radar data. The approach is straightforward; it exploits the ThermalNoiseRemoval module, provided by the European Space Agency (ESA) in its Sentinel Application Platform (SNAP) software, to remove thermal noise from the channel intensities. Then, noise correction on the complex data is applied, in order to estimate the noise-free <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathrm {C_{2}}$ </tex-math></inline-formula> matrix. As a further novelty, the proposed approach can be implemented in SNAP, through the use of a processing graph that is here provided. The method is applied on a dense time series of Sentinel-1 data, collected on an agricultural area located near Seville, Spain. The impact of thermal noise on the estimation of the eigendecomposition parameters of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathrm {C_{2}}$ </tex-math></inline-formula> , i.e., entropy ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$H_{2}$ </tex-math></inline-formula> ), average alpha angle ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\overline {\alpha _{2}}$ </tex-math></inline-formula> ), and anisotropy ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$A_{2}$ </tex-math></inline-formula> ), is assessed for different land-cover types, namely river, rice, forest, and urban areas. Monte Carlo simulations are implemented to assess the performance of the proposed approach in estimating <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$H_{2}$ </tex-math></inline-formula> , <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\overline {\alpha _{2}}$ </tex-math></inline-formula> , and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$A_{2}$ </tex-math></inline-formula> . Results show that the proposed noise removal method improves the estimation of these parameters for the considered land-cover classes.

Forest disturbance alerts for the Congo Basin using Sentinel-1

A humid tropical forest disturbance alert using Sentinel-1 radar data is presented for the Congo Basin. Radar satellite signals can penetrate through clouds, allowing Sentinel-1 to provide gap-free observations for the tropics consistently every 6–12 days at 10 m spatial scale. In the densely cloud covered Congo Basin, this represents a major advantage for the rapid detection of small-scale forest disturbances such as subsistence agriculture and selective logging. Alerts were detected with latest available Sentinel-1 images and results are presented from January 2019 to July 2020. We mapped 4 million disturbance events during this period, totalling 1.4 million ha with nearly 80% of events smaller than 0.5 ha. Monthly distribution of alert totals varied widely across the Congo Basin countries and can be linked to regional differences in wet and dry season cycles, with more forest disturbances in the dry season. Results indicated high user’s and producer’s accuracies and the rapid confirmation of alerts within a few weeks. Our disturbance alerts provide confident detection of events larger than or equal to 0.2 ha but do not include smaller events, which suggests that disturbance rates in the Congo Basin are even higher than presented in this study. The new alert product can help to better study the forest dynamics in the Congo Basin with improved spatial and temporal detail and near real-time detections, and highlights the value of dense Sentinel-1 time series data for large-area tropical forest monitoring. The research contributes to the Global Forest Watch initiative in providing timely and accurate information to support a wide range of stakeholders in sustainable forest management and law enforcement. The alerts are available via the https://www.globalforestwatch.org and http://radd-alert.wur.nl.

Mapping coastal salt marshes in China using time series of Sentinel-1 SAR

Salt marshes provide crucial ecological functions and services and are experiencing rapid losses and degradation under global climate changes and high-intensity human activities in coastal zone. However, mapping salt marsh distributions and compositions with high accuracy on the national or global scale remains challenging. Here, we used Sentinel-1 time-series data and knowledge-based automatic decision tree classifiers to produce a 10-m map of the salt marshes in China coastal zones. We established annual synthetic aperture radar (SAR) composite features to address seasonal variations and tidal dynamics in salt marshes and created five partitioned-region classifiers according to the climatic and environmental characteristics. In total, 1729 Sentinel-1 images in 2019 and 3418 in situ sites were included for the classification. The results showed that (1) the annual mean SAR composite is more stable and practical than the seasonal composite for salt marsh classification along the China coastal zones in terms of these features addressing variations in tidal fluctuation, climate zones, and salt marsh compositions; (2) SAR surface roughness can effectively separate salt marshes into sparse density and red phenotype salt marshes; (3) the total area of Chinese coastal salt marshes was 127,477.37 ha with an overall accuracy of 87.30%, it included 48.30% Spartina alterniflora, 29.19% Phragmites australis, 13.76% Suaeda spp. and 8.75% Scirpus spp., and more than 85% of salt marshes were distributed from Shandong Province to Zhejiang Province. This study showed the potential application of time-series Sentinel-1 SAR data, annual composite data and knowledge-based classifiers for large-scale coastal zones, and these data will be valuable for coastal ecological restoration and sustainability management.

Google Earth Engine Based Approach for Assessment and Management of Flood in Ganga Sub Basin – Ghaghra Confluence to Gomti Confluence

Floods are an intense and frequent disaster happening in numerous portions of the world. Flood is a excess of water that submerges surroundings that is normally dry. It is the sever problem in Ganga Sub Basin - Ghaghra Confluence to Gomti Confluence of Uttar Pradesh. Near real time mapping of inundated areas is very important for figuring out the flood extent, deployment of emergency reaction teams, and evaluation of damages and casualties. In this thesis work, a real time flood mapping and monitoring online WEB based application using Sentinel-1 time-series data has been developed on Google Earth Engine (GEE) platform. The SAR Data has Capability to see through the cloud and during the flood the major problem with optical data is, it can not see through the cloud. In this thesis work the SAR data has been used to identify the inundated pixels before and during flood to identify the extent and calamities due to flood. Flood is a hazardous natural phenomenon which severely affect lives, goods and services. Monitoring of flood affected area becomes compulsory for emergency responses but due to damage of network and risk for lives in going such areas it becomes a very typical task. Google Earth Engine (GEE) is an open-source cloud based online platform which reduces this problem. In this thesis work the GEE platform has been used for Near-Real time flood mapping in the study area. The GEE based app “FLOOD MONITORING SYSTEM: GANGA SUB BASIN – GHAGHRA CONFLUENCE TO GOMTI CONFLUENCE” has been developed for Near-Real time flood Mapping.

Improving the Accuracy of Fractional Evergreen Forest Cover Estimation at Subpixel Scale in Cloudy and Rainy Areas by Harmonizing Landsat-8 and Sentinel-2 Time-Series Data

Evergreen forest provides essential ecosystem services such as maintaining the balance in carbon and oxygen cycles and air purification. However, under cloudy and rainy weather conditions, it is difficult to obtain optical remote sensing images with high spatial resolution and complete time series. In addition, surfaces underlying the forest canopy can be complex and fragmented. To solve this problem, we developed a new approach (NDVI <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CV</sub> -LS) for mapping urban evergreen forest at the subpixel scale. In order to capture more accurate growth characteristics of evergreen forest, we harmonized Landsat-8 and Sentinel-2 images with cloud cover less than 10% acquired within 1 year to denser the time-series dataset. In view of the time series fluctuation stability of evergreen forest, the coefficient of variation (CV) of the normalized difference vegetation index (NDVI) was used to distinguish evergreen forest from other vegetation. Meanwhile, the annual minimum NDVI (NDVI <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ann-min</sub> ) was used as the parameter in a dimidiate pixel model for estimating fractional evergreen forest cover (FVC <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ever</sub> ). Hefei, a cloudy and rainy subtropical city in China, was selected as a case study to evaluate the validity of the model. The verification results revealed that harmonizing Landsat-8 and Sentinel-2 time-series images to extract evergreen forest improved the overall accuracy by 8% compared with using Landsat-8 images alone, indicating that the NDVI <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CV</sub> -LS model can improve the accuracy of FCV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ever</sub> estimation, especially for areas with complex underlying surfaces under cloudy and rainy conditions.

Dynamic Retrieval of Olive Tree Properties Using Bayesian Model and Sentinel-2 Images

The goal of this study is to provide a fine detection and monitoring of olive orchard trees over large areas to anticipate any damage. We developed an original method to assess the spatiotemporal dynamics of biophysical parameters in the olive orchards using satellite observations and radiative transfer models. Sentinel-2 time-series data collected over a four-year period were fused with Planet images from the same time period to enhance the temporal trends in olive orchards in the Sfax region located in southern Tunisia. These images also served to extract soil spectrum variations required by the 3-D discrete anisotropic radiative transfer model to account for canopy background effect. As a backward model, we developed an original technique based on the Markov chain Monte Carlo method that has the advantage of being able to model sensor noise and account for spatial and temporal regularization. It allows retrieving key parameters such as leaf area index (LAI), chlorophyll content, water content, and mesophyll structure. Taking advantage of 1) the Sentinel-2 images downscaled to a moderate resolution of 80 m to ensure representative pixels of the local mixing (i.e., trees and soil); 2) the appropriate soil signature derived from high spatial and spectral resolution image; and 3) the accuracy of the direct and inverse modeling, it was possible to retrieve the plant properties even when LAI values are less than 0.14. Indeed, our inversion results show that the estimated parameters are strongly correlated especially with the LAI field measurements with $R^{2}=0.9937$ .

Characterizing ecosystem functional type patterns based on subtractive fuzzy cluster means using Sentinel-2 time-series data

The characteristics of ecosystem functions are of great significance for biodiversity conservation and ecosystem services. Ecosystem functional types (EFTs) are land surface areas similar in carbon dynamics that are not defined by the structure and composition of vegetation and represent the spatial heterogeneity of ecosystem functions. However, identification of EFTs based on low-resolution remote sensing data cannot satisfy the needs of fine-scale characterization of regional ecosystem functionality patterns, and a more accurate and optimized method of identifying EFTs also deserves attention. Here, we characterize EFTs at a county scale based on subtractive fuzzy cluster means (SUBFCM) and Sentinel-2 time-series data. The normalized difference vegetation index, the fraction of absorbed photosynthetically active radiation, and canopy water content and their derived variables in the growing season were selected as ecosystem functional indicators to characterize regional EFT diversity patterns. The correspondence analysis method was used to reveal relationships between the EFTs and land cover structure information, and further analysis of EFTs was performed with soil type data. Our results showed that the selected variables indicating carbon and water flux of the regional ecosystems could be adopted in ecosystem functional classification. The SUBFCM algorithm can automatically divide EFTs with faster convergence speed and reduced subjectivity. The obtained EFTs based on Sentinel-2 images reflected the internal structure of carbon balance well and the distribution pattern of ecosystem functional diversity at a fine scale. A reference for further optimization of the EFT identification algorithm and development of the understanding of spatial heterogeneity of temperate terrestrial ecosystem functions was provided.

An Approach to High-Resolution Rice Paddy Mapping Using Time-Series Sentinel-1 SAR Data in the Mun River Basin, Thailand

Timely and accurate regional rice paddy monitoring plays a significant role in maintaining the sustainable rice production, food security, and agricultural development. This study proposes an operational automatic approach to mapping rice paddies using time-series SAR data. The proposed method integrates time-series Sentinel-1 data, auxiliary data of global surface water, and rice phenological characteristics with Google Earth Engine cloud computing platform. A total of 402 Sentinel-1 scenes from 2017 were used for mapping rice paddies extent in the Mun River basin. First, the calculated minimum and maximum values of the backscattering coefficient of permanent water (a classification type within global surface water data) in a year was used as the threshold range for extracting the potential extent. Then, three rice phenological characteristics were extracted based on the time-series curve of each pixel, namely the date of the beginning of the season (DBS), date of maximum backscatter during the peak growing season (DMP), and length of the vegetative stage (LVS). After setting a threshold for each phenological parameter, the final rice paddy extent was identified. Rice paddy map produced in this study was highly accurate and agreed well with field plot data and rice map products from the International Rice Research Institute (IRRI). The results had a total accuracy of 89.52% and an F1 score of 0.91, showing that the spatiotemporal pattern of extracted rice cover was consistent with ground truth samples in the Mun River basin. This approach could be expanded to other rice-growing regions at the national scale, or even the entire Indochina Peninsula and Southeast Asia.

RUESVMs: An Ensemble Method to Handle the Class Imbalance Problem in Land Cover Mapping Using Google Earth Engine

Timely and accurate Land Cover (LC) information is required for various applications, such as climate change analysis and sustainable development. Although machine learning algorithms are most likely successful in LC mapping tasks, the class imbalance problem is known as a common challenge in this regard. This problem occurs during the training phase and reduces classification accuracy for infrequent and rare LC classes. To address this issue, this study proposes a new method by integrating random under-sampling of majority classes and an ensemble of Support Vector Machines, namely Random Under-sampling Ensemble of Support Vector Machines (RUESVMs). The performance of RUESVMs for LC classification was evaluated in Google Earth Engine (GEE) over two different case studies using Sentinel-2 time-series data and five well-known spectral indices, including the Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), Soil-Adjusted Vegetation Index (SAVI), Normalized Difference Built-up Index (NDBI), and Normalized Difference Water Index (NDWI). The performance of RUESVMs was also compared with the traditional SVM and combination of SVM with three benchmark data balancing techniques namely the Random Over-Sampling (ROS), Random Under-Sampling (RUS), and Synthetic Minority Over-sampling Technique (SMOTE). It was observed that the proposed method considerably improved the accuracy of LC classification, especially for the minority classes. After adopting RUESVMs, the overall accuracy of the generated LC map increased by approximately 4.95 percentage points, and this amount for the geometric mean of producer’s accuracies was almost 3.75 percentage points, in comparison to the most accurate data balancing method (i.e., SVM-SMOTE). Regarding the geometric mean of users’ accuracies, RUESVMs also outperformed the SVM-SMOTE method with an average increase of 6.45 percentage points.