Time Series Of Imagery Research Articles

Climate change effects on waterhole persistence in rivers of the Lake Eyre Basin, Australia

The episodic, seasonal hydrology of Lake Eyre Basin (LEB) rivers produce a series of waterholes which provide critical aquatic refugia and essential water supply during predominantly dry periods. This study used direct measures of water loss, regional meteorological data, and waterhole bathymetry to develop calibrated waterhole persistence models in a range of waterhole types throughout the Queensland portion of the LEB. Evaporation was a major driver of waterhole persistence, whereas the influence of other water balance parameters (e.g. groundwater discharge, seepage, vegetation utilisation) varied between waterholes and years. Estimates of waterhole persistence were extrapolated to the regional scale using a modelled time-series of waterhole level data and Landsat satellite imagery (1988–2019). Strong relationships were calculated between waterhole level and waterhole area (r2 = 0.88) and waterhole level and number of waterholes (r2 = 0.79) for all regional waterhole zones. Using these relationships and predicted evaporation rates under three possible future greenhouse gas scenarios, modelling showed that, by 2070, the number of persistent waterholes could reduce by up to 67% over 12 month period without flow. Similarly, reductions in waterhole area of up to 72% are possible under drier climate scenarios. These results potentially represent a significant risk to the aquatic ecosystems and other waterhole-dependent users for an already limited resource.

Mapping mangrove dynamics and colonization patterns at the Suriname coast using historic satellite data and the LandTrendr algorithm

Mangroves play an important role in protecting coasts against wave energy and storms. Mangrove ecosystems provide important habitats for fauna and flora and are an important carbon sink. Loss of mangroves forest may lead to enhanced coastal erosion. Mangroves are complex ecosystems and processes of settling and development are not fully understood. Characterizing the rates and patterns of mangrove gains and losses is needed to better understand the functioning of mangrove ecosystems, how mangrove dynamics are linked to coastal morphological behaviour and how human interference with the coastal system impacts mangroves. Here we present a study of the mangrove ecosystems at the Suriname coast, which are relatively pristine and characterized by strong dynamics due to migrating mudbanks along the coast. Satellite images between 2000 and 2018, available in the historic satellite image archives, were analysed using the LandTrendr (Landsat-based detection of trends in disturbance and recovery) algorithm to identify locations of mangrove erosion, mangrove colonization, surface areas of change and patterns of settlement, as indicated by (sudden) changes in NDVI. The algorithm requires careful setting of various parameters for successful detection of (abrupt) temporal changes in mangrove coverage. The algorithm was evaluated on its robustness using various parameter settings. Results show the value of the timeseries of Landsat imagery to detect locations of coastal erosion of up to 50 m/yr and accretion where loss or settlement of mangroves is prevailing between 2000 and 2018. Locally differences are very large. An overall westward mangrove progression along the coast is apparent from the images and probably linked to mud bank migration. Various patterns of mangrove colonization and development such as arc-, zonal- and patch- arrangements were identified, although at some locations the Landsat resolution of 30 m is somewhat coarse to allow detailed analysis. The success and robustness of the LandTrendr algorithm are controlled by NDVI threshold values, number of allowed breakpoints in the timeseries and fitting parameters. The presented method requires further testing and evaluation but is a promising tool for semi-automatic detection of coastal mangrove erosion and colonization that can be applied to other mangrove ecosystems in the world. The satellite timeseries analyses generate valuable information on coastal dynamics, which is helpful to identify coastal areas prone to erosion and mangrove retreat and provide as such a valuable tool for coastal management and protection.

Improved Mapping of Long-Term Forest Disturbance and Recovery Dynamics in the Subtropical China Using All Available Landsat Time-Series Imagery on Google Earth Engine Platform

After implementations of many economic and forestry policies during recent 30 years, forest cover in China has experienced significant changes. The fine scale and long-term information on forest temporary (disturbance) loss, persistent loss (deforestation), and gains (afforestation/reforestation) will provide a guide for forest management and policies at regional scale. Using Jiangxi Province as a pilot study region, here we assessed forest cover dynamics in the subtropical China during 1986–2019 by integrating time-series Landsat images, Landsat-based Detection of Trends in Disturbance and Recovery algorithm and Random Forest classifier on the Google Earth Engine Platform. The accuracy assessment indicated a high overall accuracy (>90%) and Kappa coefficient (>0.9) for both forest gain and loss detections as evaluated against various sources of plot-level sample data and the existing global forest change product. The total forest loss area was 18 697.79 km2, with persistent loss area of 3394.31 km2 due to land conversion during 1986–2019, while persistent (net) forest gain area was 45 656.96 km2, accounting for 57.70% of the forest area in 1986. Forest loss area exhibited large interannual variations, but showed a general increase trend from 1986 to 2019. The annual variation patterns of forest gain and loss area were associated with the changes in forestry policies and large disturbance events. Our assessments on the long-term and fine scale forest dynamic patterns will help evaluate the effectiveness of forest management practices and forestry polices on forest resource sustainability, and climate change and greenhouse gases mitigation in Jiangxi Province and China.

Displacement history and potential triggering factors of Baige landslides, China revealed by optical imagery time series

In this study, on the basis of the image correlation technique and the time-series images of Landsat-8 (L8), Sentinel-2 (S2), and GaoFen-2 (GF2), a systematic technical process is designed to investigate the precursory displacement evolution of two successive slope failures occurred in Baige Village, China on Oct. 11, 2018 and Nov. 3, 2018. An innovative fusion strategy is proposed to investigate the displacement history of the Oct. 11, 2018 Baige landslide, which has two steps: (1) selecting the optimal correlation window size and search step to eliminate the inconsistency of image correlation for different sensors, and (2) inverting the fused displacement time-series from correlation results to enhance the temporal sampling density. The normalized displacement velocity indicates that the Oct. 11, 2018 Baige landslide is characterized by high-speed sliding (28 m/yr) and high shear outlet with an average elevation difference of around 82 m. The fused displacement time series indicates that the whole phase transformation can be from the secondary stage to the tertiary stage on Mar. 26, 2017. Furthermore, the displacement velocity shows four quiescence phases in the secondary stage and two acceleration phases in the tertiary stage. The seasonal precipitation is assumed as the main external triggering factor, and it combined with the brittle geological material attributes to control the precursory landslide displacement evolution and caused the catastrophic slope failure on Oct. 11, 2018. The precursory displacement signals with a magnitude above 5 m/day of the second landslide (Nov. 3, 2018) is quantified by the S2 image correlation. This study highlights the prospects of optical observation time series with medium-/high-resolution in detecting and quantifying the spatio-temporal evolution characteristics of long-term creeping landslides which may play a significant role in the early warning of the catastrophic slope failure.

Suspended sediment dynamics in a deltaic estuary controlled by subtidal motion and offshore river plumes

Two series of opportunistic observations captured the seasonal variability in the suspended sediment concentration (SSC) within the main inlet of Barataria Bay, a bar-built estuary within the Mississippi Delta, as the offshore environment transitioned from a period of low river discharge and high cold front activity to a period of high river discharge and low cold front activity. During winter, the SSC dynamics in Barataria Pass was mostly driven by resuspension in response to cold front passages, with correspondingly high SSC of up to 1.84 g L−1. In addition to cold fronts, diurnal variability associated with astronomic tides was also present in the SSC time series. The mean SSC during spring (0.23 g L−1) was significantly higher than during winter (0.15 g L−1). Higher SSC during spring, in spite of the less frequent and weaker cold fronts, was due to the strong offshore influence of the Mississippi River plume. The longitudinal salinity gradients during spring showed the existence of freshwater intrusion, which was associated with the Mississippi River plume advected north-west by the Louisiana Coastal Current. Intrusion of the river plume though the inlet into Barataria Bay was also observed in time-series of MODIS satellite imagery during March 2018. A decomposition of the sediment flux time series revealed that subtidal motion associated with cold fronts was the primary mechanism of sediment transport during both winter and spring. While cold fronts enhance sediment export during winter, they also increase sediment import when associated with the higher offshore SSC during spring. We conclude that the sediment discharged at the mouth of the Mississippi River is not completely lost to the coastal zone, but instead partially supplied to deltaic estuaries through their seaward boundaries.

Time series monitoring of impervious surfaces and runoff impacts in Metro Vancouver

Urban areas are increasing rapidly worldwide, leading to widespread changes in land surfaces over time. Urbanized land cover is heterogeneous, and is characterized by a large areal proportion of manufactured impervious surfaces which are linked to ecological degradation, habitat loss, and increase in precipitation runoff leading to pollution and safety risks. Data from the Landsat series of satellites present an opportunity to characterize urban land cover and impervious surfaces, over a large spatial and temporal scale. In this study, land cover changes from 1990 to 2015 are characterized in the large metropolitan area of Metro Vancouver, Western Canada. An ordinal regression is used to link Landsat spectral data with a detailed land classification containing classes of impervious surface used by municipal planners in the region (Spearman's Rho = 0.76). The regression is then used to classify a time series of imagery where static land classifications are not available, providing a 25-year time-series of change in impervious surface area. A trend in increasing impervious surface was detected across the municipalities in the region, with an overall areal increase of 31.96%. Precipitation events were then simulated at each time step, using precipitation rates adjusted for expected changes in climate by 2050. Both runoff depths and inundated area increased over time, with a 51% increase in area inundated by at least 5 cm. Runoff depths were evaluated for each municipality in the region, and compared to projected populations for 2050 to highlight communities that may face elevated levels of runoff risk. Results show a steady increase in impervious surfaces in the region. Impacts of future extreme precipitation events vary across the region, with flat and low-lying topographies appearing to be more severely affected, along with areas containing extensive impervious development.

Rapid cloud-based temporal compositing of Sentinel-1 radar imagery for epibenthic shellfish inventory

Delineation of shellfish beds through field surveys is time consuming. Remote sensing can help in detecting the location and boundaries of shellfish beds. This can be achieved through the use of aerial photographs and optical satellite sensors during cloud-free and low-tide conditions. Cloud penetrating radar is an alternative, but still requires careful selection of low tide imagery. Manual selection becomes cumbersome for large areas, such as the entire Dutch Wadden Sea. We therefore developed a method that automatically combines dense time series of Sentinel-1 radar imagery into a useful mosaic that allows for effective delineation of shellfish beds. The method consists of temporal compositing many images in the cloud-based Google Earth Engine. We evaluated different combinations of 1) compositing options (average backscatter and various percentiles), 2) temporal periods for compositing, 3) different polarizations, and 4) imagery from ascending versus descending satellite paths. The resulting composite images were visually compared with in-situ records to identify which composite visually best allowed for shellfish bed delineation. Although the average VH-polarized backscatter for one full year provided effective demarcation, an RGB color composite containing three average VH images for four months each provided improved visibility of the shellfish beds. We then quantitatively compared which compositing option was most effective in separating pixels with shellfish beds from those without. For this purpose, the unsupervised classification approach of K-means clustering was applied to the different composite images, and outcomes (i.e. selected classes) were compared with field data derived from the annual in-situ survey. The multi-season image composites (i.e. Jan–Apr, May–Aug, and Sep–Dec) that were made by combining VH image acquisitions of only descending and of combined ascending and descending paths resulted in higher accuracies than other compositing options tested. The producer and user accuracy of the multi-season descending composite were respectively 40% and 70% for the full Dutch Wadden Sea, against 59% and 73% for a smaller subset. These accuracies are similar to what is previously achieved for small study areas with carefully selected radar images acquired around low tide. Moreover, our approach could identify various shellfish beds that were initially missed by the in-situ surveys, but for which later the presence was confirmed (so the actual producer accuracy would increase to about 82%). The composite images can be created on the fly with Google Earth Engine shortly before a field campaign and used as a base for sampling design.

First record of the occurrence of sea ice in the Cordillera Darwin fjords (54°S), Chile

AbstractThis paper provides the first evidence for sea-ice formation in the Cordillera Darwin (CD) fjords in southern Chile, which is farther north than sea ice has previously been reported for the Southern Hemisphere. Initially observed from a passenger plane in September 2015, the presence of sea ice was then confirmed by aerial reconnaissance and subsequently identified in satellite imagery. A time series of Sentinel-1 and Landsat-8 images during austral winter 2015 was used to examine the chronology of sea-ice formation in the Cuevas fjord. A longer time series of imagery across the CD was analyzed from 2000 to 2017 and revealed that sea ice had formed in each of the 13 fjords during at least one winter and was present in some fjords during a majority of the years. Sea ice is more common in the northern end of the CD, compared to the south where sea ice is not typically present. Is suggested that surface freshening from melting glaciers and high precipitation reduces surface salinity and promotes sea-ice formation within the semi-enclosed fjord system during prolonged periods of cold air temperatures. This is a unique set of initial observations that identify questions for future research in this remote area.

Clasificación de usos y cubiertas del suelo y análisis de cambios en los alrededores de la Reserva Ecológica Manglares Churute (Ecuador) mediante una serie de imágenes Sentinel-1

<p class="p1">Management practices adopted in protected natural areas often ignore the relevance of the territory surrounding the actual protected land (buffer area). These areas can be the source of impacts that threaten the protected ecosystems. This paper reports a case study where a time series of Sentinel-1 imagery was used to classify the land-use/land-cover and to evaluate its change between 2015 and 2018 in the buffer area around the Manglares Churute Ecological Reserve (REMCh) in Ecuador. Sentinel-1 scenes were processed and ground-truth data were collected consisting of samples of the main land-use/land-cover classes in the region. Then, a Random Forests (RF) classification algorithm was built and optimized, following a five-fold cross validation scheme using the training dataset (70% of the ground truth). The remaining 30% was used for validation, achieving an Overall Accuracy of 84%, a Kappa coefficient of 0.8 and successful class performance metrics for the main crops and land use classes. Results were poorer for heterogeneous and minor classes, nevertheless the performance of the classification was deemed sufficient for the targeted change analysis. Between 2015 and 2018, an increase in the area covered by intensive land uses was evidenced, such as shrimp farms and sugarcane, which replaced traditional crops (mainly rice and banana). Even though such changes only affected the land area around the natural reserve, they might affect its water quality due to the use of fertilizers and pesticides that easily. Therefore, it is recommended that these buffer areas around natural protected areas be taken into account when designing adequate environmental protection measures and polices.</p>

ФРАКТАЛЬНИЙ АНАЛІЗ КОСМІЧНИХ ЗНІМКІВ SENTINEL-2 ДЛЯ МОНІТОРИНГУ СІЛЬСЬКОГОСПОДАРСЬКИХ КУЛЬТУР

The subject of the study in the article is using the new approach to the processing of spatial information from satellites for more effective and operational evaluation of crops. This is due to the growing trend of access to remote sensing data, due to the improvement of spatial and temporal resolution, which can be used in the analysis of vegetation cover and other related work. The goal of the article is the capability assessment of processing the Sentinel-2 satellite imagery using fractal dimensions to agricultural plant monitoring at different phases of the vegetative. The tasks: to research the method of constructing fractal dimensions for the Sentinel-2 satellite imagery to assess the state of crops during the vegetative phase; to assess the relationship between changes in FD averages and changes in the NDVI index of different time series remote images, to determine the advantage of calculation method fractal dimensions compared to the NDVI index. The following results were obtained. It was found that the NDVI index is most often used to quantify the state of biomass during different time intervals. But this index becomes ineffective during periods of weakening of the vegetation active phase. Accordingly, it is of practical interest to evaluate the possibility of using fractal analysis of agricultural crop satellite imagery at different vegetation phases. The basis of fractal analysis of digital images is the formation of fractal dimensions fields. The analysis of changes in the FD values on different remote images time series of the grain cornfields from the «sliding window» values is carried out. The dependences of the maximum and minimum values of FD, which are in the images, on the «window» size are investigated. It is shown that the homogeneity of the underlying surface can be estimated from the magnitude of changes in the maximum values of FD with the increasing size of the «window». It is established that the pattern of the change of the FD minimum values when changing the «window» size is due to the large sharpness of the underlying surface in the images, and the anomalous behavior of these values allows determining anomalous areas of different sizes in satellite imagery. The pattern of the change in the range of FD with increasing size of the «window», which can be used to determine the homogeneity of the underlying surface in satellite imagery, as well as during the detection of abnormal areas on them. The change analysis of FD average values with an increase in the sizes of «sliding window» is carried out. It is shown that with the same size of the «window» for different image time series, the average FD will be different, which can be used to characterize the agriculture crop vegetation phase. It is established that the pattern of changes in the FD average values is the same as the NDVI indices for different satellite imagery time series of the corn crop fields and that the magnitudes of the FD average values depend on the size of the «window». The size of the «window» is recommended, which provides accommodation between the speed of image processing and the quality of the assessment state vegetation crop. It is shown that to increase the speed of formation of the FFD during the processing of large images, it is advisable to use a «jumping window» instead of a «sliding window». It is mentioned that the «jump» value can be equal to the «window» size. This «jump» value provides maximum speed and does not affect the crop satellite imagery processing quality. Conclusions. The recommended approach to the processing of spatial data from satellites allows assessing the crops' consistency using FD. The pattern of the change in the FD mean values is identical to the NDVI change in different satellite imagery time series of corn crops. In that event, when forming the FFD, data from only one channel of the Sentinel-2 satellite can be used (for example, from the near-infrared channel – b8), and to calculate the NDVI index it is necessary to obtain data from two channels (from the near-infrared and red channels – channels b8 and b4 of the satellite Sentinel-2, respectively), which will reduce the processing time. The scale of FD average values allows detecting a qualitative change in biomass. During further research, it is advisable to perform fractal analysis of Sentinel-2 satellite imagery for other crops at different phases of the vegetation.

Semantically Enriched Crop Type Classification and Linked Earth Observation Data to Support the Common Agricultural Policy Monitoring

During the last decades, massive amounts of satellite images are becoming available that can be enriched with semantic annotations for the creation of value-added earth observation products. One challenge is to extract knowledge from the raw satellite data in an automated way and to effectively manage the extracted information in a semantic way, to allow fast and accurate decisions of spatiotemporal nature in a real operational scenario. In this work, we present a framework that combines supervised learning for crop type classification on satellite imagery time-series with semantic web and linked data technologies to assist in the implementation of rule sets by the European common agricultural policy (CAP). The framework collects georeferenced data that are available online and satellite images from the Sentinel-2 mission. We analyze image time-series that cover the entire cultivation period and link each parcel with a specific crop. On top of that, we introduce a semantic layer to facilitate a knowledge-driven management of the available information, capitalizing on ontologies for knowledge representation and semantic rules, to identify possible farmers noncompliance according to the Greening 1 (crop diversification) and SMR 1 rule (protection of waters against pollution caused by nitrates) rules of the CAP. Experiments show the effectiveness of the proposed integrated approach in three different scenarios for crop type monitoring and consistency checking for noncompliance to the CAP rules: the smart sampling of on-the-spot checks; the automatic detection of CAP's Greening 1 rule; and the automatic detection of susceptible parcels according to the CAP's SMR 1 rule.

Subpixel Analysis of MODIS Imagery Time Series using Transfer Learning and Relative Calibration

Transfer learning reuses a pre-trained model on a new related problem, which can be useful for monitoring large areas such as the Amazon biome. A given object must have similar spectral characteristics in the data used for this type of analysis, which can be achieved using relative calibration techniques. In this article, we present a relative calibration process in multitemporal images and evaluate its impacts on a subpixel classification process. MODIS images from the Amazon region, collected between 2013 and 2017, were relatively calibrated using a 2012 image as reference and classified by transfer learning. Classifications of calibrated and uncalibrated images were compared with data from the PRODES project, focusing on forest areas. A great variation was observed in the spectral responses of the forest class, even in images of proximate dates and from the same sensor. These variations significantly impacted the land cover classifications in the subpixel, with cases of agreement between the uncalibrated data maps and PRODES of 0%. For calibrated data, the agreement values ​​were greater than 70%. The results indicate that the method used, although quite simple, is adequate and necessary for the subpixel classification of MODIS images by transfer learning.

A Gated Recurrent Units (GRU)-Based Model for Early Detection of Soybean Sudden Death Syndrome through Time-Series Satellite Imagery

In general, early detection and timely management of plant diseases are essential for reducing yield loss. Traditional manual inspection of fields is often time-consuming and laborious. Automated imaging techniques have recently been successfully applied to detect plant diseases. However, these methods mostly focus on the current state of the crop. This paper proposes a gated recurrent unit (GRU)-based model to predict soybean sudden death syndrome (SDS) disease development. To detect SDS at a quadrat level, the proposed method uses satellite images collected from PlanetScope as the training set. The pixel image data include the spectral bands of red, green, blue and near-infrared (NIR). Data collected during the 2016 and 2017 soybean-growing seasons were analyzed. Instead of using individual static imagery, the GRU-based model converts the original imagery into time-series data. SDS predictions were made on different data scenarios and the results were compared with fully connected deep neural network (FCDNN) and XGBoost methods. The overall test accuracy of classifying healthy and diseased quadrates in all methods was above 76%. The test accuracy of the FCDNN and XGBoost were 76.3–85.5% and 80.6–89.2%, respectively, while the test accuracy of the GRU-based model was 82.5–90.4%. The calculation results show that the proposed method can improve the detection accuracy by up to 7% with time-series imagery. Thus, the proposed method has the potential to predict SDS at a future time.

Monitoring of flood water propagation based on microwave and optical imagery

The economy of northern part of Bangladesh mainly depends upon agriculture based production. Flood disasters in Dinajpur along the Punarbhaba, Atrai Rivers result in significant loss of life and economic damage. This paper finds the optimum technique to use microwave (Sentinel-1A) and optical (Landsat-8) imagery alternatively along with GIS for fast and effective flood mapping, semi-real-time monitoring and analyzing the propagation of flood water in a flood prone area. Since flooded areas usually are covered by dense clouds, Synthetic Aperture Radar (SAR) has the advantage over optical data by operating at wavelengths not hampered by cloud cover or a lack of illumination. It has capability to continuous observation of flood events for producing rapid, accurate and cost effective flood mapping. This characteristic makes SAR a potential alternative to optical sensors. The focal steps in flood monitoring using optical satellite data are: identifying water bodies efficiently; disregarding some cloud influences; and, monitoring the flood process dynamically. The purpose of this study is to evaluate the effectiveness of using freely available time series of Copernicus Sentinel-1 SAR and Landsat-8 imagery. This study also assess the TRMM data for precipitation measurement in the study area for hydrological analyses of watersheds. The SAR images were first calibrated, geometrically corrected and filtered. Afterward, to extract the inundated areas from the SAR images threshold method was applied. In threshold method and MNDWI index was used to separate the open water and nonwater (land) areas from the images. This research proposed the combination model of SAR and optical data that describes how we can get water extraction result with higher accuracy during flood using optical and microwave data. The results showed that the VH of SAR backscattering coefficient achieved higher accuracy than VV of SAR backscattering coefficient, and the combination model provides the highest accuracy of flood water extraction. The highest number of populations are affected by flood that is around 827822 in 2017. Therefore, the findings of this study will help to require initiative to scale back the flood hazard impact and increase the pliability within the process of flood management. This study proposed a method to automatically generate periodic surface water at 10 m resolution, which may contribute to monitoring surface water in a timely manner. This has also played a significant role in the monitoring and early warning of emerging major natural disasters at home and abroad.

Automatic Extraction and Filtering of OpenStreetMap Data to Generate Training Datasets for Land Use Land Cover Classification

This paper tests an automated methodology for generating training data from OpenStreetMap (OSM) to classify Sentinel-2 imagery into Land Use/Land Cover (LULC) classes. Different sets of training data were generated and used as inputs for the image classification. Firstly, OSM data was converted into LULC maps using the OSM2LULC_4T software package. The Random Forest classifier was then trained to classify a time-series of Sentinel-2 imagery into 8 LULC classes with samples extracted from: (1) The LULC maps produced by OSM2LULC_4T (TD0); (2) the TD1 dataset, obtained after removing mixed pixels from TD0; (3) the TD2 dataset, obtained by filtering TD1 using radiometric indices. The classification results were generalized using a majority filter and hybrid maps were created by merging the classification results with the OSM2LULC outputs. The accuracy of all generated maps was assessed using the 2018 official “Carta de Ocupação do Solo” (COS). The methodology was applied to two study areas with different characteristics. The results show that in some cases the filtering procedures improve the training data and the classification results. This automated methodology allowed the production of maps with overall accuracy between 55% and 78% greater than that of COS, even though the used nomenclature includes classes that can be easily confused by the classifiers.

10-m-resolution mangrove maps of China derived from multi-source and multi-temporal satellite observations

China has lost 50% of its mangroves since the 1950s, while the remaining mangroves are exhibiting an increase in fragmentation. While a detailed mangrove map of China derived from remote sensing imagery is crucial for protecting mangroves, fragmented mangrove patches are hardly captured by Landsat-derived 30-m-resolution maps. To overcome this limitation, we proposed 10-m-resolution mangrove maps for 2017 using Sentinel SAR and optical time-series imagery, together with terrain data. We first classified coastal land covers based on quantile synthesis; following this, the mangrove category was refined by applying the tidal flat constraint and manual editing. The proposed approach considered factors that limit the accuracy of mangrove maps, including tidal inundation, intertidal terrain, cloudiness, and phenological variations of vegetation. A pixel-based mangrove map (PXL_Map) and an object-based mangrove map (OBJ_Map) were produced using the Google Earth Engine cloud computing platform. PXL_Map achieved an overall accuracy of 95.3 ± 0.2% based on validation samples, and a 91.6% accuracy based on field sample plots. OBJ_Map had a slightly lower performance than PXL_Map. The maps produced in this study improved the delineation of small mangrove patches along tidal creeks relative to recent Landsat-based maps. The total mangrove area was estimated to be between 21,148 ha (by PXL_Map) and 24,801 ha (by OBJ_Map). Furthermore, the spatial differences among the mangrove maps produced in this study were quantitatively compared with three other existing mangrove maps. The discrepancies among maps could have been due to the size and the location of the mangrove patches. The 10-m-resolution mangrove maps from this study can help manage and protect mangroves in China, especially the small fragmented mangrove patches scattered along the coast.

Monitoring tropical forest succession at landscape scales despite uncertainty in Landsat time series.

Forecasting rates of forest succession at landscape scales will aid global efforts to restore tree cover to millions of hectares of degraded land. While optical satellite remote sensing can detect regional land cover change, quantifying forest structural change is challenging. We developed a state-space modeling framework that applies Landsat satellite data to estimate variability in rates of natural regeneration between sites in a tropical landscape. Our models work by disentangling measurement error in Landsat-derived spectral reflectance from process error related to successional variability. We applied our modeling framework to rank rates of forest succession between 10 naturally regenerating sites in Southwestern Panama from about 2001 to 2015 and tested how different models for measurement error impacted forecast accuracy, ecological inference, and rankings of successional rates between sites. We achieved the greatest increase in forecasting accuracy by adding intra-annual phenological variation to a model based on Landsat-derived normalized difference vegetation index (NDVI). The best-performing model accounted for inter- and intra-annual noise in spectral reflectance and translated NDVI to canopy height via Landsat-lidar fusion. Modeling forest succession as a function of canopy height rather than NDVI also resulted in more realistic estimates of forest state during early succession, including greater confidence in rank order of successional rates between sites. These results establish the viability of state-space models to quantify ecological dynamics from time series of space-borne imagery. State-space models also provide a statistical approach well-suited to fusing high-resolution data, such as airborne lidar, with lower-resolution data that provides better temporal and spatial coverage, such as the Landsat satellite record. Monitoring forest succession using satellite imagery could play a key role in achieving global restoration targets, including identifying sites that will regain tree cover with minimal intervention.

Advancing the Mapping of Mangrove Forests at National-Scale Using Sentinel-1 and Sentinel-2 Time-Series Data with Google Earth Engine: A Case Study in China

A high resolution mangrove map (e.g., 10-m), including mangrove patches with small size, is urgently needed for mangrove protection and ecosystem function estimation, because more small mangrove patches have disappeared with influence of human disturbance and sea-level rise. However, recent national-scale mangrove forest maps are mainly derived from 30-m Landsat imagery, and their spatial resolution is relatively coarse to accurately characterize the extent of mangroves, especially those with small size. Now, Sentinel imagery with 10-m resolution provides an opportunity for generating high-resolution mangrove maps containing these small mangrove patches. Here, we used spectral/backscatter-temporal variability metrics (quantiles) derived from Sentinel-1 SAR (Synthetic Aperture Radar) and/or Sentinel-2 MSI (Multispectral Instrument) time-series imagery as input features of random forest to classify mangroves in China. We found that Sentinel-2 (F1-Score of 0.895) is more effective than Sentinel-1 (F1-score of 0.88) in mangrove extraction, and a combination of SAR and MSI imagery can get the best accuracy (F1-score of 0.94). The 10-m mangrove map was derived by combining SAR and MSI data, which identified 20003 ha mangroves in China, and the area of small mangrove patches (<1 ha) is 1741 ha, occupying 8.7% of the whole mangrove area. At the province level, Guangdong has the largest area (819 ha) of small mangrove patches, and in Fujian, the percentage of small mangrove patches is the highest (11.4%). A comparison with existing 30-m mangrove products showed noticeable disagreement, indicating the necessity for generating mangrove extent product with 10-m resolution. This study demonstrates the significant potential of using Sentinel-1 and Sentinel-2 images to produce an accurate and high-resolution mangrove forest map with Google Earth Engine (GEE). The mangrove forest map is expected to provide critical information to conservation managers, scientists, and other stakeholders in monitoring the dynamics of the mangrove forest.

Assessing Canopy Responses to Thinnings for Sweet Chestnut Coppice with Time-Series Vegetation Indices Derived from Landsat-8 and Sentinel-2 Imagery

Forest management treatments often translate into changes in forest structure. Understanding and assessing how forests react to these changes is key for forest managers to develop and follow sustainable practices. A strategy to remotely monitor the development of the canopy after thinning using satellite imagery time-series data is presented. The aim was to identify optimal remote sensing Vegetation Indices (VIs) to use as time-sensitive indicators of the early response of vegetation after the thinning of sweet chestnut (Castanea Sativa Mill.) coppice. For this, the changes produced at the canopy level by different thinning treatments and their evolution over time (2014–2019) were extracted from VI values corresponding to two trials involving 33 circular plots (r = 10 m). Plots were subjected to one of the following forest management treatments: Control with no intervention (2800–3300 stems ha−1), Treatment 1, one thinning leaving a living stock density of 900–600 stems ha−1 and Treatment 2, a more intensive thinning, leaving 400 stems ha−1. Time series data from Landsat-8 and Sentinel-2 were collected to calculate values for different VIs. Canopy development was computed by comparing the area under curves (AUCs) of different VI time-series annually throughout the study period. Soil-Line VIs were compared to the Normalized Vegetation Index (NDVI) revealing that the Second Modified Chlorophyll Absorption Ratio Index (MCARI2) more clearly demonstrated canopy evolution tendencies over time than the NDVI. MCARI2 data from both L8 and S2 reflected how the influence of treatment on the canopy cover decreases over the years, providing significant differences in the thinning year and the year after. Metrics derived from the MCARI2 time-series also demonstrated the capacity of the canopy to recovery to pretreatment coverage levels. The AUC method generates a specific V-shaped time-signature, the vertex of which coincides with the thinning event and, as such, provides forest managers with another tool to assist decision making in the development of sustainable forest management strategies.

Mapping of Cotton Fields Within-Season Using Phenology-Based Metrics Derived from a Time Series of Landsat Imagery

A phenology-based crop type mapping approach was carried out to map cotton fields throughout the cotton-growing areas of eastern Australia. The workflow was implemented in the Google Earth Engine (GEE) platform, as it is time efficient and does not require processing in multiple platforms to complete the classification steps. A time series of Normalised Difference Vegetation Index (NDVI) imagery were generated from Landsat 8 Surface Reflectance Tier 1 (L8SR) and processed using Fourier transformation. This was used to produce the harmonised-NDVI (H-NDVI) from the original NDVI, and then phase and amplitude values were generated from the H-NDVI to visualise active cotton in the targeted fields. Random Forest (RF) models were built to classify cotton at early, mid and late growth stages to assess the ability of the model to classify cotton as the season progresses, with phase, amplitude and other individual bands as predictors. Results obtained from leave-one-season-out cross validation (LOSOCV) indicated that Overall Accuracy (OA), Kappa, Producer’s Accuracies (PA) and User’s Accuracy (UA), increased significantly when adding amplitude and phase as predictor variables to the model, than prediction using H-NDVI or raw bands only. Commission and omission errors were reduced significantly as the season progressed and more in-season imagery was available. The methodology proposed in this study can map cotton crops accurately based on the reconstruction of the unique cotton reflectance trajectory through time. This study confirms the importance of phenological metrics in improving in-season cotton fields mapping across eastern Australia. This model can be used in conjunction with other datasets to forecast yield based on the mapped crop type for improved decision making related to supply chain logistics and seasonal outlooks for production.