Land-use classification based on high-resolution remote sensing imagery and deep learning models.

High-resolution imagery and deep learning models have gained increasing importance in land-use mapping. In recent years, several new deep learning network modeling methods have surfaced. However, there has been a lack of a clear understanding of the performance of these models. In this study, we applied four well-established and robust deep learning models (FCN-8s, SegNet, U-Net, and Swin-UNet) to an open benchmark high-resolution remote sensing dataset to compare their performance in land-use mapping. The results indicate that FCN-8s, SegNet, U-Net, and Swin-UNet achieved overall accuracies of 80.73%, 89.86%, 91.90%, and 96.01%, respectively, on the test set. Furthermore, we assessed the generalization ability of these models using two measures: intersection of union and F1 score, which highlight Swin-UNet's superior robustness compared to the other three models. In summary, our study provides a systematic analysis of the classification differences among these four deep learning models through experiments. It serves as a valuable reference for selecting models in future research, particularly in scenarios such as land-use mapping, urban functional area recognition, and natural resource management.

Combined use of SAR images and numerical simulations to identify the source and trajectories of oil spills in coastal environments.

Remote sensing data and numerical simulation are important tools to rebuild any oil spill accident letting to identify its source and trajectory. Through these tools was identified an oil spill that affected Oaxacan coast in October 2022. The SAR images were processed with a standard method included in SNAP software, and the numerical simulation was made using Lagrangian transport model included in GNOME software. With the combining of these tools was possible to discriminate the look-alikes from true oil slicks; which are the main issue when satellite images are used. Obtained results showed that 4.3m3 of crude oil were released into the ocean from a punctual point of oil pollution. This oil spill was classified such as a small oil spill. The marine currents and weathering processes were the main drivers that controlled the crude oil displacement and its dispersion. It was estimated in GNOME that 1.6 m3 of crude oil was floating on the sea (37.2 %), 2.4 m3 was evaporated into the atmosphere (55.8 %) and 0.3 m3 reached the coast of Oaxaca (7 %). This event affected 82 km of coastline, but the most important touristic areas as well as turtle nesting zones were not affected by this small crude oil spill. Results indicated that the marine-gas-pump number 3 in Salina Cruz, Oaxaca, is a punctual point of oil pollution in the Southern Mexican Pacific Ocean. Further work is needed to assess the economic and ecological damage to Oaxacan coast caused by this small oil spill.

Use of remote sensing to assess vegetative stress as a proxy for soil contamination.

We report, for the first time, a multimodal investigation of current crude oil reprocessing and storage sites to assess their impact on the environment after 50 years of continuous operation. We have adopted a dual approach to investigate potential soil contamination. The first approach uses conventional analytical techniques i.e. energy dispersive X-ray fluorescence (ED-XRF) for metal analysis, and a complementary metabolomic investigation using hydrophilic liquid interaction chromatography hi-resolution mass spectrometry (HILIC-MS) for organic contaminants. Secondly, the deployment of an unmanned aerial vehicle (UAV) with a multispectral image (MSI) camera, for the remote sensing of vegetation stress, as a proxy for sub-surface soil contamination. The results identified high concentrations of barium (mean 21 017 ± 5950 µg g-1, n = 36) as well as metabolites derived from crude oil (polycyclic aromatic hydrocarbons), cleaning processes (surfactants) and other organic pollutants (e.g. pesticides, plasticizers and pharmaceuticals) in the reprocessing site. This data has then been correlated, with post-flight data analysis derived vegetation indices (NDVI, GNDVI, SAVI and Cl green VI), to assess the potential to identify soil contamination because of vegetation stress. It was found that strong correlations exist (an average R2 of >0.68) between the level of soil contamination and the ground cover vegetation. The potential to deploy aerial remote sensing techniques to provide an initial survey, to inform decision-making, on suspected contaminated land sites can have global implications.

Study on remote sensing inversion and temporal-spatial variation of Hulun lake water quality based on machine learning.

Hulun Lake is facing significant water quality degradation, necessitating effective monitoring for safety. Traditional methods lack the necessary spatial and temporal coverage, underscoring the need for a remote sensing model. In this study, we utilized the Landsat 8 OLI dataset, incorporating cross-section monitoring and field sampling data comprehensively. Employing the random forest algorithm, we constructed a remote sensing inversion model for six water quality parameters in Hulun Lake: chlorophyll-a (Chl-a), total nitrogen (TN), total phosphorus (TP), ammonia nitrogen (NH3-N), chemical oxygen demand (COD), and dissolved oxygen (DO). The model was applied to the non-freezing period of Hulun Lake from 2016 to 2021, exhibiting commendable performance and generating high-resolution maps. Time series analysis revealed that during the study period, the pollution levels of TN, TP, and COD in Hulun Lake were extremely serious, exceeding the Class V water standard of China's surface water environmental quality standard. Regional analysis indicated lower pollutant concentrations in the central lake area compared to the lake inlet. The inflowing rivers with high pollution adversely impacted Hulun Lake's water quality. To ensure the continued health of Hulun Lake's water quality, it is imperative to monitor lake water quality attentively and implement necessary measures to prevent further deterioration. This study holds crucial importance for shaping and executing ecological protection and restoration strategies for Hulun Lake.

A remote sensing approach for exploring the dynamics of jellyfish, relative to the water current.

Drifting in large numbers, jellyfish often interfere in the operation of nearshore electrical plants, cause disturbances to marine recreational activity, encroach upon local fish populations, and impact food webs. Understanding the dynamic mechanisms behind jellyfish behavior is of importance in order to create migration models. In this work, we focus on the small-scale dynamics of jellyfish and offer a novel method to accurately track the trajectory of individual jellyfish with respect to the water current. The existing approaches for similar tasks usually involve a surface float tied to the jellyfish for location reference. This operation may induce drag on the jellyfish, thereby affecting its motion. Instead, we propose to attach an acoustic tag to the jellyfish's bell and then track its geographical location using acoustic beacons, which detect the tag's emissions, decode its ID and depth, and calculate the tag's position via time-difference-of-arrival acoustic localization. To observe the jellyfish's motion relative to the water current, we use a submerged floater that is deployed together with the released tagged jellyfish. Being Lagrangian on the horizontal plane while maintaining an on-demand depth, the floater drifts with the water current; thus, its trajectory serves as a reference for the current's velocity field. Using an acoustic modem and a hydrophone mounted to the floater, the operator from the deploying boat remotely changes the depth of the floater on-the-fly, to align it with that of the tagged jellyfish (as reported by the jellyfish's acoustic tag), thereby serving as a reference for the jellyfish's 3D motion with respect to the water current. We performed a proof-of-concept to demonstrate our approach over three jellyfish caught and tagged in Haifa Bay, and three corresponding floaters. The results present different dynamics for the three jellyfish, and show how they can move with, and even against, the water current.

Sub-monthly time scale forecasting of harmful algal blooms intensity in Lake Erie using remote sensing and machine learning.

Harmful algal blooms of cyanobacteria (CyanoHAB) have emerged as a serious environmental concern in large and small water bodies including many inland lakes. The growth dynamics of CyanoHAB can be chaotic at very short timescales but predictable at coarser timescales. In Lake Erie, cyanobacteria blooms occur in the spring-summer months, which, at annual timescale, are controlled by the total spring phosphorus (TP) load into the lake. This study aimed to forecast CyanoHAB cell count at sub-monthly (e.g., 10-day) timescales. Satellite-derived cyanobacterial index (CI) was used as a surrogate measure of CyanoHAB cell count. CI was related to the in-situ measured chlorophyll-a and phycocyanin concentrations and Microcystis biovolume in the lake. Using available data on environmental and lake hydrodynamics as predictor variables, four statistical models including LASSO (Least Absolute Shrinkage and Selection Operator), artificial neural network (ANN), random forest (RF), and an ensemble average of the three models (EA) were developed to forecast CI at 10-, 20- and 30-day lead times. The best predictions were obtained by using the RF and EA algorithms. It was found that CyanoHAB growth dynamics, even at sub-monthly timescales, are determined by coarser timescale variables. Meteorological, hydrological, and water quality variations at sub-monthly timescales exert lesser control over CyanoHAB growth dynamics. Nutrients discharged into the lake from rivers other than the Maumee River were also important in explaining the variations in CI. Surprisingly, to forecast CyanoHAB cell count, average solar radiation at 30 to 60 days lags were found to be more important than the average solar radiation at 0 to 30 days lag. Other important variables were TP discharged into the lake during the previous 10 years, TP and TKN discharged into the lake during the previous 120 days, the average water level at 10-day lag and 60-day lag.

Aerial remote sensing of sub-sea dispersant injection effects during the Deepwater Horizon (MC-252) oil spill.

During the Deepwater Horizon oil spill in 2010, subsea dispersant injection (SSDI) was utilized for the first time in an effort to reduce the amount of oil reaching the sea surface and thus potentially decrease its environmental impact and enhance responders' safety. Since then, controversy has developed about SSDI's effectiveness. Most of the analysis is based on modeling, with some models concluding SSDI significantly reduced surfacing oil volumes, and others predicting that processes unrelated to the dispersant caused most of the subsurface oil retention. This study utilized a multispectral aerial sensor image time series to correlate the surface area covered by freshly upwelled oil with changes in SSDI rates, accounting for an approximate 4 hour oil rise time lag. A significant negative correlation was found between oil-covered surface area and SSDI rates, providing direct observation support that the technique did reduce the amount of surfacing oil around the wellhead.

Oil spills detection from SAR Earth observations based on a hybrid CNN transformer networks.

Oil spills are the main threats to marine and coastal environments. Due to the increase in the marine transportation and shipping industry, oil spills have increased in recent years. Moreover, the rapid spread of oil spills in open waters seriously affects the fragile marine ecosystem and creates environmental concerns. Effective monitoring, quick identification, and estimation of the volume of oil spills are the first and most crucial steps for a successful cleanup operation and crisis management. Remote Sensing observations, especially from Synthetic Aperture Radar (SAR) sensors, are a very suitable choice for this purpose due to their ability to collect data regardless of the weather and illumination conditions and over far and large areas of the Earth. Owing to the relatively complex nature of SAR observations, machine learning (ML) based algorithms play an important role in accurately detecting and monitoring oil spills and can significantly help experts in faster and more accurate detection. This paper uses SAR images from ESA's Copernicus Sentinel-1 satellite to detect and locate oil spills in open waters under different environmental conditions. To this end, a deep learning framework has been presented to identify oil spills automatically. The SAR images were segmented into two classes, the oil slick and the background, using convolutional neural networks (CNN) and vision transformers (ViT). Various scenarios for the proposed architecture were designed by placing ViT networks in different parts of the CNN backbone. An extensive dataset of oil spill events in various regions across the globe was used to train and assess the performance of the proposed framework. After the detection performance assessments, the F1-score values for the standard DeepLabV3+, FC-DenseNet, and U-Net networks were 75.08 %, 73.94 %, and 60.85, respectively. In the combined networks models (combination of CNN and ViT), the best F1-score results were obtained as 78.48 %. Our results showed that these hybrid models could improve detection accuracy and have a high ability to distinguish oil spill borders even in noisy images. Evaluation metrics are increased in all the combined networks compared to the original CNN networks.

Application and recent progress of inland water monitoring using remote sensing techniques.

Hyperspectral remote sensing, which retrieves the water quality parameters by direct high-resolution analysis of the electromagnetic spectrum reflected from the water surface, has been widely applied for inland water quality detection. Such a new approach provides an opportunity to generate real-time data from water with the noncontact method, largely improving working efficiency. By summarizing the development and current applications of hyperspectral remote sensing, we compare the relative merits of varying remote sensing platforms, popular inversion models, and the application of hyperspectral monitoring of chlorophyll-a (Chl-a), transparency, total suspended solids (TSS), colored dissolved organic matter (CDOM), phycocyanin (PC), total phosphorus (TP), and total nitrogen (TN) water quality parameters. Most studies have focused on spaceborne remote sensing, which is usually used to monitor large waterbodies for Chl-a and other water quality parameters with optical properties; semiempirical, bio-optical, and semianalytical models are frequently used. With the rapid development of aerospace technology and near-surface remote sensing, the spectral resolution of remote sensing imaging technology has been dramatically improved and has begun to be applied to small waterbodies. In the future, the multiplatform linkage monitoring approach may become a new research direction. Advanced computer technology has also enabled machine learning models to be applied to water quality parameter inversion, and machine learning models have higher robustness than the three commonly used models mentioned above. Although nitrogen and phosphorus, with nonoptical properties, have also received attention and research from some scholars in recent years, the uncertainty of their mechanisms makes it necessary to maintain a cautious attitude when treating such research.

Experimenting Agriculture 4.0 with Sensors: A Data Fusion Approach between Remote Sensing, UAVs and Self-Driving Tractors.

Geomatics is important for agriculture 4.0; in fact, it uses different types of data (remote sensing from satellites, Unmanned Aerial Vehicles-UAVs, GNSS, photogrammetry, laser scanners and other types of data) and therefore it uses data fusion techniques depending on the different applications to be carried out. This work aims to present on a study area concerning the integration of data acquired (using data fusion techniques) from remote sensing techniques, UAVs, autonomous driving machines and data fusion, all reprocessed and visualised in terms of results obtained through GIS (Geographic Information System). In this work we emphasize the importance of the integration of different methodologies and data fusion techniques, managing data of a different nature acquired with different methodologies to optimise vineyard cultivation and production. In particular, in this note we applied (focusing on a vineyard) geomatics-type methodologies developed in other works and integrated here to be used and optimised in order to make a contribution to agriculture 4.0. More specifically, we used the NDVI (Normalized Difference Vegetation Index) applied to multispectral satellite images and drone images (suitably combined) to identify the vigour of the plants. We then used an autonomous guided vehicle (equipped with sensors and monitoring systems) which, by estimating the optimal path, allows us to optimise fertilisation, irrigation, etc., by data fusion techniques using various types of sensors. Everything is visualised on a GIS to improve the management of the field according to its potential, also using historical data on the environmental, climatic and socioeconomic characteristics of the area. For this purpose, experiments of different types of Geomatics carried out individually on other application cases have been integrated into this work and are coordinated and integrated here in order to provide research/application cues for Agriculture 4.0.

Building a UAV Based System to Acquire High Spatial Resolution Thermal Imagery for Energy Balance Modelling.

High spatial resolution and geolocation accuracy canopy evapotranspiration (ET) maps are well suited tools for evaluation of small plot field trials. While creating such a map by use of an energy balance model is routinely performed, the acquisition of the necessary imagery at a suitable quality is still challenging. An UAV based thermal/RGB integrated imaging system was built using the RaspberryPi (RPi) microcomputer as a central unit. The imagery served as input to the two-source energy balance model pyTSEB to derive the ET map. The setup's flexibility and modularity are based on the multiple interfaces provided by the RPi and the software development kit (SDK) provided for the thermal camera. The SDK was installed on the RPi and used to trigger cameras, retrieve and store images and geolocation information from an onboard GNSS rover for PPK processing. The system allows acquisition of 8 cm spatial resolution thermal imagery from a 60 m height of flight and less than 7 cm geolocation accuracy of the mosaicked RGB imagery. Modelled latent heat flux data have been validated against latent heat fluxes measured by eddy covariance stations at two locations with RMSE of 75 W/m2 over a two-year study period.

Coupling with high-resolution remote sensing data to evaluate urban non-point source pollution in Tongzhou, China.

Urban non-point source (NPS) pollution has gradually become one of the important factors affecting the urban water environment. The quantitative evaluation of urban NPS pollution is the priority to identify key control area of urban NPS pollution. Current model applied in China is mainly focused on small-scale area, large-scale spatial continuous simulation is lacking. In this study A spatial continuous evaluation method coupled with high-resolution remote sensing data has been established and the method was applied to Tongzhou, China. With the spatial distribution of land-use type and built-up area which were been obtained by remote sensing technology, the accumulative and wash-off load of urban NPS nitrogen and phosphorus were estimated for the prominent problems of nitrogen and phosphorus nutrient pollution in the rivers in the study area. The main sources of urban NPS Nitrogen and phosphorus pollution are roof and road rainfall runoff respectively. Compared to other urban NPS pollution models, the method developed in this study can quickly realize spatial visualization assessment of urban NPS pollution and provide a means to estimate urban NPS loads in entire city or urban agglomeration, it is applicable for common urban NPS pollutants and also has advantages in areas without data.

Development of remote sensing algorithm for total phosphorus concentration in eutrophic lakes: Conventional or machine learning?

Phosphorus is a limiting nutrient in freshwater ecosystems. Therefore, the estimation of total phosphorus (TP) concentration in eutrophic water using remote sensing technology is of great significance for lake environmental management. However, there is no TP remote sensing model for lake groups, and thus far, specific models have been used for specific lakes. To address this issue, this study proposes a framework for TP estimation. First, three algorithm development frameworks were compared and applied to the development of an algorithm for Lake Taihu, which has complex water environment characteristics and is a representative of eutrophic lakes. An Extremely Gradient Boosting (BST) machine learning framework was proposed for developing the Taihu TP algorithm. The machine learning algorithm could mine the relationship between FAI and TP in Lake Taihu, where the optical properties of the water body are dominated by phytoplankton. The algorithm exhibited robust performance with an R2 value of 0.6 (RMSE = 0.07 mg/L, MRE = 43.33%). Then, a general TP algorithm (R2 = 0.64, RMSE = 0.06 mg/L, MRE = 34.13%) was developed using the proposed framework and tested in seven other lakes using synchronous image data. The algorithm accuracy was found to be affected by aquatic vegetation and enclosure aquaculture. Third, compared with field investigations in other studies on Lake Taihu, the Taihu TP algorithm showed good performance for long-term TP estimation. Therefore, the machine learning framework developed in this study has application potential in large-scale spatio-temporal TP estimation in eutrophic lakes.

Land use classification of open-pit mine based on multi-scale segmentation and random forest model.

The mining industry production is an important pillar industry in China, while its extensive production activities have led to several ecological and environmental problems. Earth observation technology using high-resolution satellite imagery can help us efficiently obtain information on surface elements, surveying and monitoring various land occupation issues arising from open-pit mining production activities. Conventional pixel-based interpretation methods for high-resolution remote sensing images are restricted by "salt and pepper" noise caused by environmental factors, making it difficult to meet increasing requirements for monitoring accuracy. With the Jingxiang phosphorus mining area in Jingmen Hubei Province as the studied area, this paper uses a multi-scale segmentation algorithm to extract large-scale main characteristic information using a layered mask method based on the hierarchical structure of the image object. The remaining characteristic elements were classified and extracted in combination with the random forest model and characteristic factors to obtain land occupation information related mining industry production, which was compared with the results of the Classification and Regression Tree model. 23 characteristic factors in three aspects were selected, including spectral, geometric and texture characteristics. The methods employed in this study achieved 86% and 0.78 respectively in overall extraction accuracy analysis and the Kappa coefficient analysis, compared to 79% and 0.68 using the conventional method.

Investigation into the geometry and distribution of oil inclusions in sea ice using non-destructive X-ray microtomography and its implications for remote sensing and mitigation potential.

As climate change brings reduced sea ice cover and longer ice-free summers to the Arctic, northern Canada is experiencing an increase in shipping and industrial activity in this sensitive region. Disappearing sea ice, therefore, makes the Arctic region susceptible to accidental releases of different types of oil and fuel pollution resulting in a pressing need for the development of appropriate scientific knowledge necessary to inform regulatory policy formulation. In this study, we examine the microstructure of the surficial layers of sea ice exposed to oil using X-ray microtomography. Through analysis, 3D imaging of the spatial distribution of the ice's components (brine, air, and oil) were made. Additional quantitative information regarding the size, proximity, orientation, and geometry of oil inclusions were computed to ascertain discernable relationships between oil and the other components of the ice. Our results indicate implications for airborne remote sensing and bioremediation of the upper sea ice layers.

Guidelines for the use of the Attend Anywhere Platform for Telecommunications within the Pain Service.

Introduction Remote consultation is of growing in importance and gaining popularity in both primary and secondary healthcare settings. Reduced necessity for a physical presence of the patient within the healthcare setting is of particular benefit in the current COVID-19 era. It is also of benefit to a diverse group of patients, for example: those who are geographically distant from the base hospital, those suffering from mobility issues or chronic illness, those who require chaperoning as well as those with limited access to transport. We have developed guidelines for the use of the medical telecommunications platform, Attend Anywhere, which has been utilised across the English and Scottish National Health Services, as well as with the Australian Health service, and is now available in Health Service Executive (HSE) settings. Herein we describe and recommend a process that we have found helpful, and we propose guidelines on how a Health Care Worker (HCW) might consider approaching a virtual consultation when initiating and safely executing a patient encounter on Attend Anywhere, in a secure and efficient manner. The guidelines were created following review of the literature on previous experience by others with this software, as well as recent guidance published by the Irish Medical Council. A proportion of this guidance is transferable to other platforms. Methods We also undertook a short survey of our patients and physicians in Sligo University Hospital, who used Attend Anywhere over a six-week period to gauge their satisfaction levels with the experience., We estimated distance that our patients would have travelled for their appointment had the traditional face-to-face consultation been carried out. We noted whether we considered the medium appropriate for the patient consultations. Results 53 patients took part and satisfaction was rated from satisfied to very satisfied on a 3-point scale for all stakeholders. In addition, we found that remote consultation, when compared to face-to-face consultation, alleviated an average of 144km of unnecessary travel per appointment. Remote consultation was deemed appropriate in all cases and no rescheduled face-to-face appointments were required due to failure of the consultation due to difficulties encountered. Conclusion The authors recommend the implementation of the described guidance, with suggested Checklist, Information leaflet and Consent form, as a means of ensuring the confidentiality of the consultation and to ensure that processes are adhered to that optimise protection for both the patient and the clinician, while reducing the burden of attendance to the healthcare location.

[Application of remote sensing technology in medicinal plant resources].

The sustainable use of medicinal plants is the foundation of the inheritance of traditional Chinese medicine(TCM) and the acquisition of information on medicinal plants is the basis for the development of TCM. The traditional methods of investigating medicinal plant resources are disadvantageous in strong subjectivity and poor timeliness, making it difficult to real-time monitor medicinal plant resources. In recent years, remote sensing technology has become an important means of obtaining information on medicinal plants. The application of this technology has made up for the shortcomings of traditional methods. The open-access remote sensing data with medium spatial resolution satellites provide an opportunity for extracting information on medicinal plant resources. This study firstly introduced the principles of remote sensing technology, summarized the satellites and the parameters commonly used in the field of medicinal plant resources, and compared the survey methods of remote sensing technology with traditional methods. Secondly, it reviewed the applications of remote sensing technology in the extraction of information on the cultivation of medicinal plants and the common methods for extracting the planting structure information of medicinal plants based on remote sensing technology. Thirdly, the applications of remote sensing technology in the investigation and monitoring of medicinal plants were further analyzed with the research objects divided into wild and cultivated medicinal plants according to the characteristics of the habitats. Finally, it pointed out the key unsolved technical problems in the remote sensing monitoring of medicinal plant resources, and proposed solutions for the intelligent information processing of medicinal plants based on remote sensing big data, which is expected to provide references for the development of remote sensing technology in derivative application in medicinal plant resources.

Influence of Dispersed Oil on the Remote Sensing Reflectance-Field Experiment in the Baltic Sea.

Remote sensing techniques currently used to detect oil spills have not yet demonstrated their applicability to dispersed forms of oil. However, oil droplets dispersed in seawater are known to modify the local optical properties and, consequently, the upwelling light flux. Theoretically possible, passive remote detection of oil droplets was never tested in the offshore conditions. This study presents a field experiment which demonstrates the capability of commercially available sensors to detect significant changes in the remote sensing reflectance Rrs of seawater polluted by six types of dispersed oils (two crude oils, cylinder lubricant, biodiesel, and two marine gear lubricants). The experiment was based on the comparison of the upwelling radiance Lu measured in a transparent tank floating in full immersion in seawater in the Southern Baltic Sea. The tank was first filled with natural seawater and then polluted by dispersed oils in five consecutive concentrations of 1-15 ppm. After addition of dispersed oils, spectra of Rrs noticeably increased and the maximal increase varied from 40% to over three-fold at the highest oil droplet concentration. Moreover, the most affected Rrs band ratios and band differences were analyzed and are discussed in the context of future construction of algorithms for dispersed oil detection.

A Cyber-Physical Data Collection System Integrating Remote Sensing and Wireless Sensor Networks for Coffee Leaf Rust Diagnosis.

Coffee Leaf Rust (CLR) is a fungal epidemic disease that has been affecting coffee trees around the world since the 1980s. The early diagnosis of CLR would contribute strategically to minimize the impact on the crops and, therefore, protect the farmers' profitability. In this research, a cyber-physical data-collection system was developed, by integrating Remote Sensing and Wireless Sensor Networks, to gather data, during the development of the CLR, on a test bench coffee-crop. The system is capable of automatically collecting, structuring, and locally and remotely storing reliable multi-type data from different field sensors, Red-Green-Blue (RGB) and multi-spectral cameras (RE and RGN). In addition, a data-visualization dashboard was implemented to monitor the data-collection routines in real-time. The operation of the data collection system allowed to create a three-month size dataset that can be used to train CLR diagnosis machine learning models. This result validates that the designed system can collect, store, and transfer reliable data of a test bench coffee-crop towards CLR diagnosis.

Optimal Bands Combination Selection for Extracting Garlic Planting Area with Multi-Temporal Sentinel-2 Imagery.

Garlic is one of the main economic crops in China. Accurate and timely extraction of the garlic planting area is critical for adjusting the agricultural planting structure and implementing rural policy actions. Crop extraction methods based on remote sensing usually use spectral-temporal features. Still, for garlic extraction, most methods simply combine all multi-temporal images. There has been a lack of research on each band's function in each multi-temporal image and optimal bands combination. To systematically explore the potential of the multi-temporal method for garlic extraction, we obtained a series of Sentinel-2 images in the whole garlic growth cycle. The importance of each band in all these images was ranked by the random forest (RF) method. According to the importance score of each band, eight different multi-temporal combination schemes were designed. The RF classifier was employed to extract garlic planting area, and the accuracy of the eight schemes was compared. The results show that (1) the Scheme VI (the top 39 bands in importance score) achieved the best accuracy of 98.65%, which is 6% higher than the optimal mono-temporal (February, wintering period) result, and (2) the red-edge band and the shortwave-infrared band played an essential role in accurate garlic extraction. This study gives inspiration in selecting the remotely sensed data source, the band, and phenology for accurately extracting garlic planting area, which could be transferred to other sites with larger areas and similar agriculture structures.