RapidEye Satellite Images Research Articles

Sistemas aéreos no tripulados y sensores remotos pasivos para clasificar microecosistemas de pastizales alto andinos

Pastures are the fodder base for camelid and sheep production in the southern Peruvian Andes, where 80% of alpacas and 15 % of sheep live, which requires better land management and grazing programs through the classification of microecosystems. The objective of this study was to classify the microecosystems based on the grasslands of the Kayra Agronomic Center in the Region of Cusco using unmanned aerial vehicles and remote sensors. To do this, traditional evaluation and estimation methods such as modified Parker and quadrat sampling, were combined with biomass classification and estimation methods supported by multispectral images. This was done using 5 m RapidEye satellite images, and multispectral orthophotographs acquired with a Micasense sensor transported by a Matrix 300 RTK Drone with 10 cm pixels. Processing was performed by Pix 4D version 4.7.5 photogrammetry software, and ENVI and ArcGIS 10.3 image processing software. An algorithm designed in the R programming language was used to estimate the biomass. The results show three life zones, three climatic zones, four ecosystems, and four plant communities with eleven dominant species. The condition of the grasslands evaluated was regular with a tendency to poor and a carrying capacity of 0.3 UV.ha-1.year-1; 0.83 UO.ha-1.year-1 and 1.11UA.ha-1.year-1. The use of remote sensors made it possible to classify grasslands quickly and efficiently.

Mapping of Inland Excess Water Using Geographical Information System and High-Resolution Satellite Images: A Case Study of SREM, Serbia

Abstract Extreme hydrological events, such as floods and droughts, are becoming more frequent as a result of climate change, leading to negative impacts on various economic sectors. The Pannonian-Carpathian Basin is particularly affected by the increasing frequency of hazardous hydrological events. Agricultural production, which is a highly significant economic sector in the region, is particularly vulnerable to these unfavourable climatic conditions. Changes in precipitation patterns and soil moisture levels can lead to reduced crop yields, while floods can pollute water sources and erode fertile soil. Mapping of Inland Excess Water (IEW), also known as ponding water or waterlogged areas, is crucial for informed decision-making, damage compensation, risk management, and future prevention planning. Remote sensing technology and machine learning have been demonstrated to be valuable tools for the mapping of IEW. The 2014 floods in Southeastern and Central Europe serve as a reminder of the importance of effective flood risk management. This study used a Geographical Information System (GIS) and a Semi-automated Classification Processing (SCP) tool to process high-resolution RapidEye satellite images from the 2014 floods in the Srem region of Serbia. The Spectral Angle Mapping (SAM) classification model was used to produce a map of IEW. The SAM model achieved an overall accuracy of 92.68 %. The study found that IEW affected approximately 2.90 % or 99.59 km² of the territory in Srem. The obtained maps can be used by responsible water management agencies to prevent and control excessive inland water.

Himalayan hazard cascades – modern and medieval outburst floods in Pokhara, Nepal

AbstractIn May 2012, a sediment‐laden flood along the Seti Khola (= river) caused 72 fatalities and widespread devastation for > 40 km in Pokhara, Nepal's second largest city. The flood was the terminal phase of a hazard cascade that likely began with a major rock‐slope collapse in the Annapurna Massif upstream, followed by intermittent ponding of meltwater and subsequent outburst flooding. Similar hazard cascades have been reported in other mountain belts, but peak discharges for these events have rarely been quantified. We use two hydrodynamic models to simulate the extent and geomorphic impacts of the 2012 flood and attempt to reconstruct the likely water discharge linked to even larger medieval sediment pulses. The latter are reported to have deposited several cubic kilometres of sediment in the Pokhara Valley. The process behind these sediment pulses is debated. We traced evidence of aggradation along the Seti Khola during field surveys and from RapidEye satellite images. We use two steady‐state flood models, HEC‐RAS and ANUGA, and high‐resolution topographic data, to constrain the initial flood discharge with the lowest mismatch between observed and predicted flood extents. We explore the physically plausible range of simplified flood scenarios, from meteorological (1000 m3 s−1) to cataclysmic outburst floods (600,000 m3 s−1). We find that the 2012 flood most likely had a peak discharge of 3700 m3 s−1 in the upper Seti Khola and attenuated to 500 m3 s−1 when arriving in Pokhara city. Simulations of larger outburst floods produce extensive backwater effects in tributary valleys that match with the locations of upstream‐dipping medieval‐age slackwater sediments in several tributaries of the Seti Khola. Our findings are consistent with the notion that the medieval sediment pulses were linked to outburst floods with peak discharges of >50,000 m3 s−1, though discharge may have been an order of magnitude higher.

Status of mangroves land use on the Brazilian Amazon coast from RapidEye imagery and GEOBIA approach.

Given the ecological and socioeconomic importance of the mangroves of the Brazilian Amazon coast, RapidEye satellite images were analyzed to recognize mangrove forest and salt flat changes to different land use through human activities. Results show that mangroves are still very well preserved, with less than 1% of the total converted to other uses, primarily urban areas and roads. These human activities have been the principal causes of use in the mangrove forest, driven by local anthropogenic pressures resulting from human settlements in the transition zone between the mainland and tidal flats. In contrast, aquaculture, the principal driver of the loss of mangroves in other regions of South America and in Asian countries, plays only a secondary role in habitat conversion on the Amazon coast. However, these human activities demand more attention and policies need to be supported by Brazilian legislation.

Effectiveness of the red-edge band of RapidEye in land cover classification

ABSTRACT This study examined the effectiveness of the red-edge band using RapidEye satellite images for land cover classification. The analysis comprises three schemes for evaluating the effectiveness of the red-edge band: principal component analysis (PCA), vegetation index, and supervised image classification. The factor loadings computed by means of PCA were applied to analyze the importance of each band in the training samples. The analysis results of the factor loadings indicated that the red-edge band performed better than the visible band in the vegetation region. When rice paddy and peanuts were classified using the NDVI_RE, the improvement in accuracy was approximately 7%. Further, the accuracy of rice paddy classification using CMFI_RE was improved by approximately 6%. It can thus be inferred that the red-edge band made a certain contribution to vegetation classification. In land cover classification using reflectance, the accuracy of the support vector machine (SVM) was higher than that of the maximum likelihood classifier (MLC), the iterative self-organizing data analysis technique, and the K-means algorithm. When the red-edge band was included, the overall accuracy improved from 1% to 3%. The results of our experiments indicated that the red-edge band contributed marginally to land cover classification.

Evaluation of Hyperparameter Combinations of the U-Net Model for Land Cover Classification

The aim of this study was to select the optimal deep learning model for land cover classification through hyperparameter adjustment. A U-Net model with encoder and decoder structures was used as the deep learning model, and RapidEye satellite images and a sub-divided land cover map provided by the Ministry of Environment were used as the training dataset and label images, respectively. According to different combinations of hyperparameters, including the size of the input image, the configuration of convolutional layers, the kernel size, and the number of pooling and up-convolutional layers, 90 deep learning models were built, and the model performance was evaluated through the training accuracy and loss, as well as the validation accuracy and loss values. The evaluation results showed that the accuracy was higher with a smaller image size and a smaller kernel size, and was more dependent on the convolutional layer configuration and number of layers than the kernel size. The loss tended to be lower as the convolutional layer composition and number of layers increased, regardless of the image size or kernel size. The deep learning model with the best performance recorded a validation loss of 0.11 with an image size of 64 × 64, a convolutional layer configuration of C→C→C→P, a kernel size of 5 × 5, and five layers. Regarding the classification accuracy of the land cover map constructed using this model, the overall accuracy and kappa coefficient for three study cities showed high agreement at approximately 82.9% and 66.3%, respectively.

Analysis of Interannual and Seasonal Nearshore Bar Behaviour Observed from Decadal Optical Satellite Data in the Curonian Spit, Baltic Sea

Long-term observations of nearshore bar behaviour are a vital component of coastal monitoring, management, and prediction. Optical satellite remote sensing enables the possibility of such observations over large spatial areas, but its full potential remains unexploited. This study assessed alongshore variability in cross-shore nearshore bar behaviour on a wave-dominated multi-bar coast of the Curonian Spit (south-eastern Baltic Sea) between 2011 and 2021, using satellite-derived bar data. Nearshore bars were extracted from a time series of PlanetScope and RapidEye satellite images with an automated GIS-based algorithm, previously proposed by the study authors. The cross-shore behaviour of a multiple bar system was analysed by adapting traditional bathymetry-based analysis techniques to satellite-derived data that included bar crestlines and images of multi-scale Relative Bathymetric Position Index (RBPI). The analysis was performed on 1071 shore-perpendicular transects. Multi-bar onshore and offshore migration rates were quantified on interannual and seasonal timescales. The results show that, on an interannual timescale, bars migrated offshore at rates up to 9.7 m/month, while the rates of onshore migration reached up to 11 m/month. During the months of low wave energy, bars moved offshore at rates up to 6.2 m/month, and during the months of high wave energy, up to 12.9 m/month. However onshore migration rates, during the months of low and high wave energy, reached up to 7.0 and 13.4 m/month, respectively. A complex empirical orthogonal function (CEOF) analysis was performed on RBPI-derived cross-shore profiles, and cyclic offshore directed bar behaviour was examined. For the first time, the net offshore migration (NOM) cycle with bar cycle return periods of 1.8 to 13.5 years was investigated on the south-eastern Baltic Sea coast. Bar cycle return periods increased and rates of bar cross-shore migration decreased from north to south along the Curonian Spit. Similar nearshore bar behaviour regions were identified using clustering analysis based on quantified temporal and morphological characteristics of the bars. Factors controlling alongshore variability in bar cross-shore behaviour were determined. The study results suggest that small alongshore variations in nearshore hydrodynamics, caused by the local wave climate and its interplay with the shoreline orientation, determine the morphological and temporal variability of the multi-bar system in the Curonian Spit.

Landslide Segmentation with Deep Learning: Evaluating Model Generalization in Rainfall-Induced Landslides in Brazil

Automatic landslide mapping is crucial for a fast response in a disaster scenario and improving landslide susceptibility models. Recent studies highlighted the potential of deep learning methods for automatic landslide segmentation. However, only a few works discuss the generalization capacity of these models to segment landslides in areas that differ from the ones used to train the models. In this study, we evaluated three different locations to assess the generalization capacity of these models in areas with similar and different environmental aspects. The model training consisted of three distinct datasets created with RapidEye satellite images, Normalized Vegetation Index (NDVI), and a digital elevation model (DEM). Here, we show that larger patch sizes (128 × 128 and 256 × 256 pixels) favor the detection of landslides in areas similar to the training area, while models trained with smaller patch sizes (32 × 32 and 64 × 64 pixels) are better for landslide detection in areas with different environmental aspects. In addition, we found that the NDVI layer helped to balance the model’s results and that morphological post-processing operations are efficient for improving the segmentation precision results. Our research highlights the potential of deep learning models for segmenting landslides in different areas and is a starting point for more sophisticated investigations that evaluate model generalization in images from various sensors and resolutions.

COMPARISON OF PIXEL AND OBJECT BASED CLASSIFICATION METHODS ON RAPIDEYE SATELLITE IMAGE

The aim of this study is to evaluate the classification performances of land use/land cover (LULC) classification methods by comparing the results of pixel and object-based classification approaches on RapidEye satellite image. Pixel-based classification was carried out in ERDAS Imagine 10.4 using the Maximum Likelihood-supervised approach, whilst object-based classification was performed in e-Cognition Developer 64 using the nearest neighbour-supervised classification method. A LULC map of eight classes was created in both methods. While the accuracy for thematic LULC classes varied in both methods, the overall accuracy and kappa values of LULC maps for pixel and object-based classification methods were 58.39%-0.45 and 89.58%-0.86, respectively. Accuracy assessments and comparative results showed that object-based classification gives better results for thematic LULC classes as well as the overall accuracy of LULC maps. Even though pixel-based classification method was good at mapping many thematic classes, there were misclassifications between natural/semi-natural LULC classes. These results can be attributed to parameters set by users, such as the number of control points, etc. However, the capacity of object-based classification method to include auxiliary data (e.g. DEM, NDVI) increases the accuracy of LULC maps with high-resolution satellites.

Is Precision Viticulture Beneficial for the High-Yielding Lambrusco (Vitis Vinifera L.) Grapevine District?

The best Lambrusco wines are often obtained by blending a representative of the Lambrusco family (i.e. Lambrusco Salamino) with a smaller fraction of Ancellotta, a teinturier variety possessing an extraordinary quality of accumulating color. Due to the economic importance of the Lambrusco business and the rising interest in precision viticulture, a two-year trial was carried out in seven vineyard plots growing both the named varieties. A RapidEye satellite image taken on 9 August 2018, led to vigor maps based on unfiltered normalized difference vegetation index (NDVI). In both the years, ground truthing was performed on the test vines chosen within each vigor area for soil features, vegetative growth, yield, grape, and final wine composition. For data pooled over sites and years, Ancellotta showed a very clear response to NDVA-based vigor mapping as low vigor areas always achieved improved ripening in terms of higher total soluble solids (+1.24 Brix), color and phenols (+0.36 mg/kg and + 0.44mg/kg, respectively), and lower malate (-1.79 g/L) vs high vigor. Such a behavior was shown even in those cases where NDVI of different vigor levels and pruning weight were not closely correlated and, most notably, low vigor matched with a slightly higher yield as compared to high vigor plots. Overall, the high yielding Lambrusco Salamino was less responsive in terms of vine performance and grape composition versus intra-vineyard variability. This study highlights that, in Ancellotta, adjusting the vine balance toward ostensible lower vigor (i.e pruning weight ≤ 1 kg/m) would result in a superior choice in terms of improved ripening and wine profiles would not be detrimentally impacted by the yield level which, in fact, increased in some cases.

A METHODOLOGY FOR DETERMINING AND ANALYZING PERMANENT PROTECTION AREAS OF PROPERTIES DECLARED IN THE RURAL ENVIRONMENTAL REGISTER- CAR

The Permanent Protection Areas (PPA) are relevant to ensure vegetation around the drainage network. This paper presents an automated methodology for the extraction of drainage from the river and automated generation of PPA, and analysis of environmental adequacy. The methodology is based on geoprocessing and remote sensing techniques applied to RapidEye satellite images. The analyzed area covers a portion of the Paraíba do Sul river basin, located in the city of São José dos Campos (Southern Brazil). Land use and land cover were determined using a digital classifier and estimated within the APP of four rural properties bordering the river. The digital classification of the RapidEye images was evaluated based on the visual interpretation of high spatial resolution airborne orthophotos, as well as through random points that enabled the generation of the Kappa index and global accuracy, showing high agreement. The analysis shows the inadequate land use practice in some properties analyzed, indicating changes in the areas of PPA over the years analyzed. The results of this research show that the proposed methodology can be used for supervision purposes in properties declared in the Rural Environmental Registry (CAR), thus assisting in the decision-making process.

Efficacy of PCR-bassed techniqes and remote sensing technology for detecting potato brown rot infection caused by R. solanacearum

Bacterial wilt of potatoes (Solanum tuberosum) caused by Ralstonia solanacearum phylotype II sequevar I (race 3 biovar 2), is one of the most serious diseases of potatoes. This study aimed to detect such deteriorations through the remotely sensed data. In this regard, naturally infested potato fields at Abo Alkhawi, Al Tayareya and Om Saber villages at Behera governorate, Egypt, were used for selective isolation of R. solanacearum. On Semi Selective South Africa (SMSA) medium, 30 isolates were randomly screened and tested by immunofluorescence antibody staining (IFAS) and confirmed biologically by tomato plantlets (cv. Pinto) reaction. All isolates showed positive results with Real-time PCR. Based on the phylotype specific multiplex (Pmx) – PCR data, all isolates in concern were R. solanacearum and affiliated to phylotype II as 372- bp amplicon. Spectral indices were applied on the Landsat 8 and RapidEye satellite images acquired in January and April 2019 including Normalized Difference Moisture Index (NDMI), Normalized Difference Vegetation Index (NDVI), Normalized Difference Salinity Index (NDSI) and Modified Chlorophyll Absorption Ratio Index (MCARI). Results from the remote sensing analyses showed medium to relatively high salinity and moisture content of soils concerened. Moreover, plants chlorophyll levels are in medium levels which is not typical to condition associated with usual circumstances. Further investigation, in a greater academic depth, are needed for pathological field surveys.

Can Water Constituents Be Used as Proxy to Map Microplastic Dispersal Within Transitional and Coastal Waters?

Due to high spatiotemporal variability of aquatic ecosystems, relationships between microplastic sources and sinks are highly complex and transportation pathways yet to be understood. Field data acquisitions are a necessary component for monitoring of microplastic contamination but alone cannot capture such complex relationships. Remote sensing is a key technology for environmental monitoring through which extrapolation of spatially limited field data to larger areas can be obtained. We tested whether microplastic distribution follows the same movement pattern as water constituents depictable from Landsat-8, Sentinel-2, and RapidEye satellite images, namely chlorophyll-a, suspended particulate matter, and colored dissolved organic matter, and if they can be utilized as proxies. As rivers are a major source for marine microplastic contamination, we sampled three example river systems: the lower courses and river mouths of the Trave and Elbe estuary in Germany and the Po delta in Italy. For a full quantitative analysis of microplastics (>300 µm), ATR- and FPA-based µFTIR spectroscopy and NIR imaging spectroscopy were utilized. A comparison of water constituents with in-situ data using regression analysis showed no consistent pattern. Only for the Trave river, a positive relationship between microplastics and water constituents was present. Differences in hydrodynamic conditions and spatiotemporal dynamics of water constituents and microplastic emissions among the rivers are possible explanations for the contrary results. The lower detection limit of 300 µm for microplastics could also have influenced relationships as microplastic abundance exponentially increases with decreasing size class. Further studies with improved sampling methods are necessary to assess our proposed method.

Integrating various satellite images for identification of the water bodies through using machine learning: A case study of Salah Adin, Iraq

This research aims to extract water bodies from several types of satellite images by using machine learning. There are several methods to apply the extraction of information about water bodies such as unsupervised classification, supervised classification. This project applied the supervised classification method to extract water bodies and building geodatabase for the water bodies in Salah Adin, Iraq by applying remote sensing and GIS technique. The satellite images which have been used in this research include satellite images from RapideEye satellite with spatial resolution (5×5) m at 2011, where these images used to extract canals of water, also used images from Sentinel-2 satellite with spatial resolution (10 × 10) m at 2017 to extract lakes, and broad rivers. In addition, DEM raster (90) m at 2011 from ASTER satellite was used to extract the water streams, which are expected to be channels of water discharge in flood situations. During implementation of the methodology of the research the most important issue which appear is using multi data to extract all types of water bodies in study area also to avoid the pixel mixed problem, this case was evident when using RapidEye satellite images in the confluence of the river in the surrounding wetlands, which led to inaccurate results in the geometric dimensions of the river, although the high spatial resolution but the influenced element on the accuracy of results is radiometric resolution. The result shows the random forest of machine learning algorithm is overcome on the other algorithms such as decision tree machine learning, maximum likelihood and support vector machine. The high accuracy of image classification to extract the water bodies depend on integrating the three satellite images.

A New Deep-Learning-Based Approach for Earthquake-Triggered Landslide Detection From Single-Temporal RapidEye Satellite Imagery

Accurate landslide detection and mapping are essential for land use planning, management/assessment, and geo-disaster risk mitigation as well as post-disaster reconstructions. Till now, visual interpretation and field survey are still the most widely adopted techniques for landslide mapping, which are often criticized labor-intensive, time-consuming, and costly. With the rapid advancement of artificial intelligence, deep-learning-based approach for landslide detection and mapping has drawn great attention for its significant advantages over the traditional techniques. However, lack of sufficient training samples has constrained the application of deep-learning-based approach in landslide detection from satellite images for a long time. The present study aimed to examine the feasibility of a new deep-learning-based approach to intelligently detect and map earthquake-triggered landslides from single-temporal RapidEye satellite images. Specifically, the proposed approach consists of three steps. First of all, a standard data preprocessing workflow to automatically generate training samples was designed and some data augmentation strategies were implemented to alleviate the lack of training samples. Then, a cascaded end-to-end deep learning network, namely LandsNet, was constructed to learn various features of landslides. Finally, the identified landslide maps were further optimized with morphological processing. Experiments in two spatially independent earthquake-affected regions showed our proposed approach yielded the best F1 value of about 86.89%, which was about 7% and 8% higher than that obtained by ResUNet and DeepUNet, respectively. Comparative studies on the feasibility and robustness of the proposed approach with ResUNet and DeepUNet demonstrated its strong application potentials in the emergency response of natural disasters.

SPATIAL DYNAMICS OF ATLANTIC FOREST FRAGMENTS IN A RIVER BASIN

Considering the intrinsic relationship between vegetation and conservation of water resources, we evaluated the natural vegetation of an Atlantic Forest fragment through landscape metrics and spectral vegetation indices, indicating ways for its conservation in the Una river basin, Ibiúna, São Paulo, Brazil. We analyzed landscape metrics and the extraction of mean altitude, slope, and distance from the drainage network for each fragment based on the vectors of forest fragments in the river basin, using ArcGis 10.3. Normalized difference vegetation indices, photochemical production, and carbon flux were estimated by combining red, green, blue, and infrared spectral bands near the RapidEye satellite image. The results showed 197 fragments with sizes ranging from 1 to 306 hectares and irregularly shaped. The largest fragments are few in number, but account for 52% of the river basin's vegetation cover area. The vegetation presented high levels of biomass but moderate photosynthetic production, coinciding for a moderate carbon flux. The best scores were related to fragments at higher altitudes, in sloping areas next to drainage sites. Actions to conserve the fragments of larger areas and draw up environmental awareness measures should be encouraged, since high agricultural vocation may incite the reduction of these areas to increase economic production.

Correlation analysis of vegetation indices based on multi-temporal satellite images and unmanned aerial vehicle images with wheat protein contents

Satellite remote sensing has been proved to be an effective way of monitoring crop growth status and yield prediction. Recently near-ground remote sensing using an unmanned aerial vehicle (UAV) witnessed wide applications in obtaining field information. In this research, four Rapideye satellite images and eight UAV images are used from early June to the end of July 2015, covering two different varieties of winter wheat fields, in order to analyze the correlation among satellite images based normalized difference vegetation index (NDVI) with the UAV images based visible-band difference vegetation index (VDVI) and ground variable of grain protein contents. The results of relational analysis of NDVI with sampled wheat grain protein content showed that NDVI related most to the grain protein content at the later stage of wheat growing season, one week prior to harvesting. And the linear-regression models for field No.1 and field No.2 of NDVI with VDVI showed good consistency at the early stage of wheat growing season, with the coefficient of determination R2 = 0.62 and 0.86, respectively.

Factors Affecting the Health Condition of Spruce Forests in Central European Mountains-Study Based on Multitemporal RapidEye Satellite Images

Spruce stands in the mountains of Central Europe are particularly valuable, not only because of their natural and scenic values but also because of their role in the protection of watersheds and soil. Over the past decades, these stands were frequently exposed to massive deforestation caused by various biotic, abiotic and anthropogenic factors. The current health condition of spruce stands in the Eastern Sudetes shows that the next phase of deterioration of these stands has begun. Bearing in mind past experiences and the current situation in these mountain areas, it is particularly important to understand the processes and factors that may short-term (2012–2016) spruce health analyses based on the normalized difference red-edge index and RapidEye satellite imagery, which is being delivered annually. Aided by remote sensing data and Boosted Regression Trees, it was possible to determine the topographic and tree stand features having the greatest impact on the vitality of spruce in all analyzed areas during 2012–2016. As the results show, the highest impact on the value of the normalized difference red-edge index indicator comes from a height above sea level, age of stands, terrain slopes, and exposure. In various areas, these factors may affect the vitality of spruce to varying degrees but not always in the same way. Our models accurately explained 74–81% of the randomly selected input data (predicted 72–80% of the excluded data) for site A, 50–58% for site B (48–57% for tested data) and 54–70% for site C (52–69% for excluded data).

LAND-COVER MAPS USING MULTIPLE CLASSIFIER SYSTEM FOR POST-DISASTER LANDSCAPE MONITORING

Abstract. Recently, land cover maps created from high resolution satellite images have been used for landscape analysis, in order to understand the impact of natural disasters on biodiversity and ecosystems. Conventional land cover classification methods, however, suffer from problems with isolated pixels (salt and pepper effect). Filtering can remove the isolated pixels, but can also result in loss of accurate information. The purpose of this study is to create a land cover map for landscape analysis of large-scale disturbances caused by the Great East Japan Earthquake of 2011, utilizing a Multiple Classifier System (MCS), which allows for reduction of isolated pixels while maintaining classification accuracy. RapidEye satellite images covering the Pacific Ocean side of the Tohoku district damaged by the earthquake and subsequent tsunami were obtained for 2010, 2011, 2012 and 2016, and land cover classification was implemented using individual classifiers and the MCS method. The results showed that the MCS land cover map was able to reduce the number of isolated pixels significantly (61-71%) compared with the individual classifiers, while maintaining very high accuracy (0.976-0.986) for all four years. These results indicate that MCS land cover maps have a great potential for analyzing disturbances following infrequent largescale natural disasters such as earthquakes and tsunami, and for monitoring the process of recovery afterwards. We expect that the results of this research will be useful in managing the recovery process in the region disturbed by the Great Eastern Japan Earthquake and Tsunami of 2011, and also for developing future Ecosystem-based Disaster Risk Reduction programs for the region.

Detecting advanced stages of winter wheat yellow rust and aphid infection using RapidEye data in North China Plain

Yellow rust (Puccinia striiformis f. sp. Tritici) and aphid (Sitobion avenae F.) are two major biotic factors threatening winter wheat growth in the main growing region in northern China. The goal of this study was to develop a remote sensing based approach to reliably detect and discriminate yellow rust and aphid infection. The study was conducted in the North China Plain in 2017 based on RapidEye satellite images using three supervised classification algorithms, the maximum-likelihood classifier, the support vector machine, and the random forest. An overall accuracy of above 60% for aphid and above 70% for yellow rust can be achieved using a single image (May 7 or 10, 2017) with any of the three algorithms. With multi-temporal images, the overall accuracies both increased for aphid and yellow rust (above 70% and above 78%). Using the image acquired on 23 May 2017, joint infections by yellow rust and aphid can be detected with satisfaction (>73% overall accuracy), although confusion exists between the two infections. This study demonstrates that winter wheat disease/pest infection can be detected with remote sensing technologies, providing decision support to farmers, insurance companies, and government organizations in the agriculture sector.