Thematic Accuracy Research Articles

Using of high spatial resolution images to evaluate the thematic accuracy of land use and occupation maps with the Kappa index

The objective from this article evaluates the thematic quality of automatic mappings with supervised classification for land use and land cover, using high spatial resolution satellite images as "ground truth". This consider the advancement of remote sensing technologies has enabled the acquisition of satellite images with various spatial resolutions, which are essential for thematic mapping and automatic classifiers in the context of land use and land cover mapping. In fact, the wide availability of high, medium, and low spatial resolution satellite images has significantly optimized the time and resources required by using more accurate classifiers during data processing. The image used for verification of this paper was GeoEye, with a spatial resolution of 0.5m, dated October 2023. The images submitted to the automatic classifier were Sentinel-2A with a spatial resolution of 10m and Planet with a spatial resolution of 5m, both from the same satellite pass period (October 2023) over the study area, aiming to avoid seasonal and phenological variations in vegetation, as well as changes in the environment due to anthropogenic intervention. The classification method adopted was Maximum Likelihood (MAXVER). The classification accuracy was rigorously evaluated to ensure the reliability of the results using the Kappa index, assessing the agreement between the observed and expected classifications. Based on the methods presented, the set of mapped classes in this study showed good accuracy for the Planet image and very good accuracy for the Sentinel image.

Evaluation of the accuracy and readability of ChatGPT-4 and Google Gemini in providing information on retinal detachment: a multicenter expert comparative study

BackgroundLarge language models (LLMs) such as ChatGPT-4 and Google Gemini show potential for patient health education, but concerns about their accuracy require careful evaluation. This study evaluates the readability and accuracy of ChatGPT-4 and Google Gemini in answering questions about retinal detachment.MethodsComparative study analyzing responses from ChatGPT-4 and Google Gemini to 13 retinal detachment questions, categorized by difficulty levels (D1, D2, D3). Masked responses were reviewed by ten vitreoretinal specialists and rated on correctness, errors, thematic accuracy, coherence, and overall quality grading. Analysis included Flesch Readability Ease Score, word and sentence counts.ResultsBoth Artificial Intelligence tools required college-level understanding for all difficulty levels. Google Gemini was easier to understand (p = 0.03), while ChatGPT-4 provided more correct answers for the more difficult questions (p = 0.0005) with fewer serious errors. ChatGPT-4 scored highest on most challenging questions, showing superior thematic accuracy (p = 0.003). ChatGPT-4 outperformed Google Gemini in 8 of 13 questions, with higher overall quality grades in the easiest (p = 0.03) and hardest levels (p = 0.0002), showing a lower grade as question difficulty increased.ConclusionsChatGPT-4 and Google Gemini effectively address queries about retinal detachment, offering mostly accurate answers with few critical errors, though patients require higher education for comprehension. The implementation of AI tools may contribute to improving medical care by providing accurate and relevant healthcare information quickly.

Comparative Analysis between Remote Sensing Burned Area Products in Brazil: A Case Study in an Environmentally Unstable Watershed

Forest fires can profoundly impact the hydrological response of river basins, modifying vegetation characteristics and soil infiltration. This results in a significant increase in surface flow and channel runoff. In response to these effects, many researchers from different areas of earth sciences are committed to determining emergency measures to rehabilitate river basins, intending to restore their functions and minimize damage to soil resources. This study aims to analyze the mapping detection capacity of burned areas in a river basin in Brazil based on images acquired by AMAZÔNIA-1/WFI and the AQ1KM product. The effectiveness of the AMAZÔNIA-1 satellite in this regard is evaluated, given the importance of the subject and the relatively recent introduction of the satellite. The AQ1KM data were used to analyze statistical trends and spatial patterns in the area burned from 2003 to 2023. The U-Net architecture was used for training and classification of the burned area in AMAZÔNIA-1 images. An increasing trend in burned area was observed through the Mann–Kendall test map and Sen’s slope, with the months of the second semester showing a greater occurrence of burned areas. The NIR band was found to be the most sensitive spectral resource for detecting burned areas. The AMAZÔNIA-1 satellite demonstrated superior performance in estimating thematic accuracy, with a correlation of above 0.7 achieved in regression analyses using a 10 km grid cell resolution. The findings of this study have significant implications for the application of Brazilian remote sensing products in ecology, water resources, and river basin management and monitoring applications.

Accuracy Assessment and Comparison of National, European and Global Land Use Land Cover Maps at the National Scale—Case Study: Portugal

The free availability of Sentinel-1 and 2 imageries enables the production of high resolution (10 m) global Land Use Land Cover (LULC) maps by a wide range of institutions, which often make them publicly available. This raises several issues: Which map should be used for each type of application? How accurate are these maps? What is the level of agreement between them? This motivated us to assess the thematic accuracy of six LULC maps for continental Portugal with 10 m spatial resolution with reference dates between 2017 and 2020, using the same method and the same reference database, in a bid to make the results comparable. The overall accuracy and the per class user’s and producer’s accuracy are compared with the ones reported by the map producers, at the national, European, or global level, according to their availability. The nomenclatures of the several maps were then analyzed and compared to generate a harmonized nomenclature to which all maps were converted into. The harmonized products were compared directly with a visual analysis and the proportion of regions equally classified was computed, as well as the area assigned per product to each class. The accuracy of these harmonized maps was also assessed considering the previously used reference database. The results show that there are significant differences in the overall accuracy of the original products, varying between 42% and 72%. The differences between the user’s and producer’s accuracy per class are very large for all maps. When comparing the obtained results with the ones reported by the map producers for Portugal, Europe or globally (depending on what is available) the results obtained in this study have lower accuracy metrics values for all maps. The comparison of the harmonized maps shows that they agree in 83% of the study area, but there are differences in terms of detail and area of the classes, mainly for the class “Built up” and “Bare land”.

Time-first approach for land cover mapping using big Earth observation data time-series in a data cube – a case study from the Lake Geneva region (Switzerland)

ABSTRACT Accurate, consistent, and high-resolution Land Use & Cover (LUC) information is fundamental for effectively monitoring landscape dynamics and better apprehending drivers, pressures, state, and impacts on land systems. Nevertheless, the availability of such national products with high thematic accuracy is still limited and consequently researchers and policymakers are constrained to work with data that do not necessarily reflect on-the-ground realities impending to correctly capture details of landscape features as well as limiting the identification and quantification of drivers and rate of change. Hereafter, we took advantage of the Switzerland’s official LUC statistical sampling survey and dense time-series of Sentinel-2 data, combining them with Machine and Deep Learning methods to produce an accurate and high spatial resolution land cover map over the Lake Geneva region. Findings suggest that time-first approach is a valuable alternative to space-first approaches, accounting for the intra-annual variability of classes, hence improving classification performances. Results demonstrate that Deep Learning methods outperform more traditional Machine Learning ones such as Random Forest, providing more accurate predictions with lower uncertainty. The produced land cover map has a high accuracy, an improved spatial resolution, while at the same time preserving the statistical significance (i.e. class proportion) of the official national dataset. This work paves the way towards the objective to produce a yearly high resolution land cover map of Switzerland and potentially implement a continuous land change monitoring capability. However further work is required to properly address challenges such as the need for increased temporal resolution for LUC information or the quality of training samples.

A Planning Support System for Monitoring Aging Neighborhoods in Germany

Many single-family homes built in Germany in the first decades following the Second World War are now occupied by elderly residents. If local conditions are unfavorable, a large number of these buildings may enter the real estate market in a short period of time and put pressure on the local housing market. Planners and decision-makers therefore need detailed spatiotemporal information about these neighborhoods to effectively address and counteract such developments. We present the design and implementation of a planning support system that can generate the required information. The architecture of this newly developed software consists of a composite, multitier framework to perform the complex tasks of data importation, data processing, and visualization. Legally mandated municipal population registers provide the key data for the calculation of indicators as a base for spatiotemporal analyses and visualizations. These registers offer high data quality in terms of completeness, logical consistency, spatial, and temporal and thematic accuracy. We demonstrate the implemented method using population data from a local government in a rural area in southwestern Germany. The results show that the new tool, which relies on open software components, is capable to identify and prioritize areas with particularly high levels of problem pressure. The tool can be used not only for analyses in a local context, but also at a regional level.

Spatial planning needs towards Copernicus Land Monitoring Services: Case studies from Poland and Norway

The objective of this study is to identify the needs related to geospatial LC, LU, and LCLUC information for spatial planning in Poland and Norway, and examine the usefulness of CLMS products in the context of these planning systems. The research has conducted based on a comparative analysis of two planning systems, to indicate areas where CLMS can improve or supplement national spatial data. The study shows that CLMS can provide information on up-to-date spatial data showing actual LC/LU/LCLUC, but that the degree of detail and the accuracy may be insufficient. CLMS data is harmonised across Europe and thus meets the need expressed by international organisations, for data that are consistent at a continental level. This is not a requirement in national planning systems in Poland and Norway, where the needs are regulated by national legislation. The thematic and geometric accuracy of national data sources are usually better than the data provided by CLMS, but CLMS might fill gaps when specific topics are missing in national mapping programs.

OSM Sidewalkreator: A QGIS plugin for an automated drawing of sidewalk networks for OpenStreetMap

Sidewalks are a relevant part of the living space in urban environments but are still rarely mapped. In recent years, the mapping of sidewalks has grown in importance among the OSM and academic communities as a matter of concern for many UN SDGs. To cover this gap, we propose a GitHub-hosted, fully open-source QGIS Plugin entitled "OSM SidewalKreator" to automatically draw the geometries of sidewalks for OSM crossings and curb crossing interfaces. The plugin workflow encompasses the steps of input area selection; data fetching, data cleaning, sidewalk geometries generation; crossings and kerbs generation; optional sidewalk splitting; and data exporting. Furthermore, the tool gives the user the capacity to have control over the process. Our tests revealed that the proposed method embodied by the plugin surpasses the manual process in many contexts, highlighting completeness, topological and thematic accuracies. We conclude that deepening, improving, and increasing the amount of open sidewalk mapping, mainly in the widely available OSM, can be a valuable asset to improve the development of accessibility and mobility worldwide.

Producing context specific land cover and land use maps of human-modified tropical forest landscapes for infectious disease applications

Satellite-based land cover mapping plays an important role in understanding changes in ecosystems and biodiversity. There are global land cover products available, however for region specific studies of drivers of infectious disease patterns, these can lack the spatial and thematic detail or accuracy required to capture key ecological processes. To overcome this, we produced our own Landsat derived 30 m maps for three districts in India's Western Ghats (Wayanad, Shivamogga and Sindhudurg). The maps locate natural vegetation types, plantation types, agricultural areas, water bodies and settlements in the landscape, all relevant to functional resource use of species involved in infectious disease dynamics. The maps represent the mode of 50 classification iterations and include a spatial measure of class stability derived from these iterations. Overall accuracies for Wayanad, Shivamogga and Sindhudurg are 94.7 % (SE 1.2 %), 88.9 % (SE 1.2 %) and 88.8 % (SE 2 %) respectively. Class classification stability was high across all three districts and the individual classes that matter for defining key interfaces between human habitation, forests, crop, and plantation cultivation, were generally well separated. A comparison with the 300 m global ESA CCI land cover map highlights lower ESA CCI class accuracies and the importance of increased spatial resolution when dealing with complex landscape mosaics. A comparison with the 30 m Global Forest Change product reveals an accurate mapping of forest loss and different dynamics between districts (i.e., Forests lost to Built-up versus Forests lost to Plantations), demonstrating an interesting complementarity between our maps and the % tree cover Global Forest Change product. When studying infectious disease responses to land use change in tropical forest ecosystems, we recommend using bespoke land cover/use classifications reflecting functional resource use by relevant vectors, reservoirs, and people. Alternatively, global products should be carefully validated with ground reference points representing locally relevant habitats.

Evaluation of Spectral Indices for Mapping Burned Areas using Unsupervised Classification in Different Ecosystems using Spectral Indices from Sentinel-2 images

Fire is one of the natural agents that has the greatest impact on the terrestrial ecosystem and plays an important ecological function in a huge portion of the Earth's surface. Remote sensing is an important source for mapping and monitoring forests as well as environmental damage to the landscape caused by fire.To estimate the severity of the fire, the temporal distinction between the spectral indices before and after the fire is critical. This study examines the performance of spectral indices derived from Sentinel-2 Multi Spectral Instrument (MSI) bands for the detection of fire-affected areas in forest fires in regions with different ecosystems located in Brazil, the United States, and Portugal. Separability Index (M) and the Reed-Xiaoli Detection (RXD) Anomaly automatic classifier allowed us to assess the spectral separability and the thematic accuracy of the burned area for the different spectral indices. The analysis parameters were based on spatial dispersion with validation data, Commission Errors (CE), Omission Errors (OE), and Sorensen–Dice Coefficient (DC). The results indicated that the indices based exclusively on the longer shortwave infrared (SWIR1), shorter SWIR (SWIR2), and red-edge bands showed a high degree of separability and were more suitable for detecting the burned areas, although it was observed that the diversity of land use affects the performance of the indices. The evaluation of the performance of the spectral indices based on Sentinel-2 data was important in the analyzes of potentialities and limitations in the detection of burned areas in the face of the different global biomes.

Análisis de la cobertura boscosa del Parque Nacional Tingo María (Perú) utilizando el algoritmo random forest

The establishment of natural protected areas is one of the most effective strategies to conserve forests and their biodiversity; however, the uncontrolled advance of deforestation resulting from the change of use to expand the agricultural frontier has become a threat to these intangible areas. This research aimed to analyze the dynamics of forest cover in Parque Nacional Tingo María (PNTM) and its buffer zone (ZA) located in the high jungle of the Huánuco region of Peru. The main input was Sentinel-2 images that were classified using the Random Forest algorithm. As a result, coverage maps were obtained for the study area corresponding to the years 2017, 2019, 2021 and 2023, achieving considerable thematic accuracy. During the evaluation periods, the rates of change from forest to non-forest within the PNTM presented low values -0.26% (2017 - 2019); -1.24% (2019 - 2021) and -0.02% (2021 - 2023). While the forests in the ZA have undergone a dynamic transition, with rates of change of -2.97%; -4.39% and -1.15% derived from land use change. The landscape metrics suggest that the forests of the PNTM are moderately fragmented, and the forests of the ZA are strongly fragmented, which leads to the conclusion that the protected natural area has fulfilled its objective of maintaining vegetation cover.

The impact of selection of reference samples and DEM on the accuracy of land cover classification based on Sentinel-2 data

Up-to-date and reliable information on land cover and land use status is important in many aspects of human activities. Knowledge about the reference dataset, its coverage, nomenclature, thematic and geometric accuracy, spatial resolution is crucial for appropriate selection of reference samples used in the classification process. In this study, we examined the impact of the selection and pre-processing of reference samples for the classification accuracy. The classification based on Random Forest algorithm was performed using firstly the automatically selected reference samples derived directly from the national databases, and secondly using the pre-processed and verified reference samples. The verification procedures involved the iterative analysis of histogram of spectral features derived from the Sentinel-2 data for individual land cover classes. The verification of the reference samples improved the accuracy of delineation of all land cover classes. The highest improvement was achieved for the woodland broadleaved and non- and sparce vegetation classes, with the overall accuracy increasing from 51% to 73%, and from 33% to 74%, respectively. The second objective of this study was to derive the best possible land cover classification over the mountain area in Norway, therefore we examined whether the use of the Digital Elevation Model (DEM) can improve the classification results. Classifications were carried out based on Sentinel-2 data and a combination of Sentinel-2 and DEM. Using the DEM the accuracy for nine out of ten land cover classes was improved. The highest improvement was achieved for classes located at higher altitudes: low vegetation and non- and sparse vegetation.

Monitoring Forest Cover Dynamics Using Orthophotos and Satellite Imagery

The assessment of changes in forest coverage is crucial for managing protected forest areas, particularly in the face of climate change. This study monitored forest cover dynamics in a 6535 ha mountain area located in north-west Romania as part of the Apuseni Natural Park from 2003 to 2019. Two approaches were used: vectorization from orthophotos and Google Earth images (in 2003, 2005, 2009, 2012, 2014, 2016, 2017, and 2019) and satellite imagery (Landsat 5 TM, 7 ETM, and 8 OLI) pre-processed to Surface Reflectance (SR) format from the same years. We employed four standard classifiers: Support Vector Machine (SVM), Random Forest (RF), Maximum Likelihood Classification (MLC), Spectral Angle Mapper (SAM), and three combined methods: Linear Spectral Unmixing (LSU) with Natural Breaks (NB), Otsu Method (OM) and SVM, to extract and classify forest areas. Our study had two objectives: 1) to accurately assess changes in forest cover over a 17-year period and 2) to determine the most efficient methods for extracting and classifying forest areas. We validated the results using performance metrics that quantify both thematic and spatial accuracy. Our results indicate a 9% loss of forest cover in the study area, representing 577 ha with an average decrease ratio of 33.9 ha/year−1. Of all the methods used, SVM produced the best results (with an average score of 88% for Overall Quality (OQ)), followed by RF (with a mean value of 86% for OQ).

A Reference-Free Method for the Thematic Accuracy Estimation of Global Land Cover Products Based on the Triple Collocation Approach

Global land cover (GLC) data are an indispensable resource for understanding the relationship between human activities and the natural environment. Estimating their classification accuracy is significant for studying environmental change and sustainable development. With the rapid emergence of various GLC products, the lack of high-quality reference data poses a severe risk to traditional accuracy estimation methods, in which reference data are always required. Thus, meeting the needs of large-scale, fast evaluation for GLC products becomes challenging. The triple collocation approach (TCCA) is originally applied to assess classification accuracy in earthquake damage mapping when ground truth is unavailable. TCCA can provide unbiased accuracy estimation of three classification systems when their errors are conditionally independent. In this study, we extend the idea of TCCA and test its performance in the accuracy estimation of GLC data without ground reference data. Firstly, to generate two additional classification systems besides the original GLC data, a k-order neighbourhood is defined for each assessment unit (i.e., geographic tiles), and a local classification strategy is implemented to train two classifiers based on local samples and features from remote sensing images. Secondly, to reduce the uncertainty from complex classification schemes, the multi-class problem in GLC is transformed into multiple binary-class problems when estimating the accuracy of each land class. Building upon over 15 million sample points with remote sensing features retrieved from Google Earth Engine, we demonstrate the performance of our method on WorldCover 2020, and the experiment shows that screening reliable sample points during training local classifiers can significantly improve the overall estimation with a relative error of less than 4% at the continent level. This study proves the feasibility of estimating GLC accuracy using the existing land information and remote sensing data, reducing the demand for costly reference data in GLC assessment and enriching the assessment approaches for large-scale land cover data.

Analysis of Spectral Separability for Detecting Burned Areas Using Landsat-8 OLI/TIRS Images under Different Biomes in Brazil and Portugal

Fire is one of the natural agents with the greatest impact on the terrestrial ecosystem and plays an important ecological role in a large part of the terrestrial surface. Remote sensing is an important technique applied in mapping and monitoring changes in forest landscapes affected by fires. This study presents a spectral separability analysis for the detection of burned areas using Landsat-8 OLI/TIRS images in the context of fires that occurred in different biomes of Brazil (dry ecosystem) and Portugal (temperate forest). The research is based on a fusion of spectral indices and automatic classification algorithms scientifically proven to be effective with as little human interaction as possible. The separability index (M) and the Reed–Xiaoli automatic anomaly detection classifier (RXD) allowed the evaluation of the spectral separability and the thematic accuracy of the burned areas for the different spectral indices tested (Burn Area Index (BAI), Normalized Burn Ratio (NBR), Mid-Infrared Burn Index (MIRBI), Normalized Burn Ratio 2 (NBR2), Normalized Burned Index (NBI), and Normalized Burn Ratio Thermal (NBRT)). The analysis parameters were based on spatial dispersion with validation data, commission error (CE), omission error (OE), and the Sørensen–Dice coefficient (DC). The results indicated that the indices based exclusively on the SWIR1 and SWIR2 bands showed a high degree of separability and were more suitable for detecting burned areas, although it was observed that the characteristics of the soil affected the performance of the indices. The classification method based on bitemporal anomalous changes using the RXD anomaly proved to be effective in increasing the burned area in terms of temporal alteration and performing unsupervised detection without relying on the ground truth. On the other hand, the main limitations of RXD were observed in non-abrupt changes, which is very common in fires with low spectral signal, especially in the context of using Landsat-8 images with a 16-day revisit period. The results obtained in this work were able to provide critical information for fire mapping algorithms and for an accurate post-fire spatial estimation in dry ecosystems and temperate forests. The study presents a new comparative approach to classify burned areas in dry ecosystems and temperate forests with the least possible human interference, thus helping investigations when there is little available data on fires in addition to favoring a reduction in fieldwork and gross errors in the classification of burned areas.

Thematic accuracy assessment of the NLCD 2019 land cover for the conterminous United States

ABSTRACT The National Land Cover Database (NLCD), a product suite produced through the MultiResolution Land Characteristics (MRLC) consortium, is an operational land cover monitoring program. Starting from a base year of 2001, NLCD releases a land cover database every 2–3-years. The recent release of NLCD2019 extends the database to 18 years. We implemented a stratified random sample to collect land cover reference data for the 2016 and 2019 components of the NLCD2019 database at Level II and Level I of the classification hierarchy. For both dates, Level II land cover overall accuracies (OA) were 77.5% ± 1% (± value is the standard error) when agreement was defined as a match between the map label and primary reference label only, and increased to 87.1% ± 0.7% when agreement was defined as a match between the map label and either the primary or alternate reference label. At Level I of the classification hierarchy, land cover OA was 83.1% ± 0.9% for both 2016 and 2019 when agreement was defined as a match between the map label and primary reference label only, and increased to 90.3% ± 0.7% when agreement also included the alternate reference label. The Level II and Level I OA for the 2016 land cover in the NLCD2019 database were 5% higher compared to the 2016 land cover component of the NLCD2016 database when agreement was defined as a match between the map label and primary reference label only. No improvement was realized by the NLCD2019 database when agreement also included the alternate reference label. User’s accuracies (UA) for forest loss and grass gain were>70% when agreement included either the primary or alternate label, and UA was generally<50% for all other change themes. Producer’s accuracies (PA) were>70% for grass loss and gain and water gain and generally<50% for the other change themes. We conducted a post-analysis review for map-reference agreement to identify patterns of disagreement, and these findings are discussed in the context of potential adjustments to mapping and reference data collection procedures that may lead to improved map accuracy going forward.

Smaller is better? Unduly nice accuracy assessments in roof detection using remote sensing data with machine learning and k-fold cross-validation

Deriving the thematic accuracy of models is a fundamental part of image classification analyses. K-fold cross-validation (KCV), as an accuracy assessment technique, can be biased because existing built-in algorithms of software solutions do not handle the high autocorrelation of remotely sensed images, leading to overestimation of accuracies. We aimed to quantify the magnitude of the overestimation of KCV-based accuracies and propose a method to overcome this problem with the example of rooftops using a WorldView-2 (WV2) satellite image, and two orthophotos. Random split to training/testing subsets, independent testing and different types of repeated KCV sampling strategies were used to generate input datasets for classification. Results revealed that applying the random splitting of reference data to training/testing subsets and KCV methods had significantly biased the accuracies by up to 17%; overall accuracies (OAs) can incorrectly reach >99%. We found that repeated KCV can provide similar results to independent testing when spatial sampling is applied with a sufficiently large distance threshold (in our case 10 m). Coarser resolution of WV2 ensured more reliable results (up to a 5–9% increase in OA) than orthophotos. Object-based pixel purity of buildings showed that when using a majority filter for at least of 50% of objects the final accuracy approached 100% with each sampling method. The final conclusion is that KCV-based modelling ensures better accuracy than single models (with better pixel purity on the object level), but the accuracy metrics without spatially filtered sampling are not reliable.

Pl@ntNet Crops: merging citizen science observations and structured survey data to improve crop recognition for agri-food-environment applications

We present a new application to recognize 218 species of cultivated crops on geo-tagged photos, ‘Pl@ntNet Crops’. The application and underlying algorithms are developed using more than 750k photos voluntarily collected by Pl@ntNet users. The app is then enriched by data and photos coming from the European Union’s (EU) Land Use and Coverage Area frame Survey (LUCAS). During five tri-annual LUCAS campaigns from 2006 to 2018, 242 476 close-up ‘cover’ photos of crops were collected. The survey protocol for these photos specified that ‘the picture should be taken at a close distance, so that the structure of leaves can be clearly seen, as well as flowers or fruits’. This unique labelled data provides an opportunity to further generalize the Pl@ntNet computer vision algorithms to recognize crops and enlarge their geographic representivity across the EU. To include LUCAS cover photos, we semantically match Pl@ntNet species and LUCAS legends, predict the species on LUCAS cover photos with the existing Pl@ntNet algorithm, and consider the accuracy of the classification and the number of species enriched by the photos. By setting a threshold of 0.5 on the Pl@ntNet prediction probabilities, 70 170 LUCAS photos representing 101 species classified with an accuracy of 0.9 were added to the ‘Crops’ app. The thematic accuracy of the legacy LUCAS data was improved by distinguishing 218 species, opposed to the original 36 LUCAS levels. Official and publicly financed LUCAS datastreams can now be improved because of Pl@ntNet citizen science, photo collection, and deep learning model development. Further use of the app and policy-relevant workflows in the agri-food-environment domain are discussed.

Optimal Site Selection Using Geographical Information System (GIS) Based Multi-criteria Decision Analysis (MCDA): A Case Study To Concentrated Solar Power Plants (CSP) In Brazil

Optimal site selection using multi-criteria decision analysis (MCDA) is an important step to support decision makers to locate places that benefit the maximum potential of technology's preconditions. As a new, emerging and renewable technology in Brazil, concentrated solar power (CSP) plays an important role in power generation mix, and it’s crucial to indicate the viability of Brazilian regions to CSP power plants. To achieve this goal, a detailed workflow of multi-criteria analysis based on geographic information system (GIS based AHP) is set in free software, in which criteria are selected from literature, acquired as geospatial data, weighted with parity matrices through online questionnaires filled by experts, processed on QGIS with weighted linear combination (WLC), and the results validated by its AHP consistency, thematic accuracy and comparative analysis with VIKOR and TOPSIS methodologies. As a product, sites are mapped according to viability indices, with higher values in most of Northeast, Central-west and Southeast regions of Brazil, showing good stability by its validation. Thereby, the workflow with free software allows the methodology to be replicated to support decision makers in locating viable and restrictive places for technologies.

Evaluation of the Ability of SLSTR (Sentinel-3B) and MODIS (Terra) Images to Detect Burned Areas Using Spatial-Temporal Attributes and SVM Classification

Forest fires are considered one of the major dangers and environmental issues across the world. In the Cerrado biome (Brazilian savannas), forest fires have several consequences, including increased temperature, decreased rainfall, genetic depletion of natural species, and increased risk of respiratory diseases. This study presents a methodology that uses data from the Sea and Land Surface Temperature Radiometer (SLSTR) sensor of the Sentinel-3B satellite and the Moderate Resolution Imaging Spectroradiometer (MODIS) of the Terra satellite to analyze the thematic accuracy of burned area maps and their sensitivity under different spectral resolutions in a large area of 32,000 km2 in the Cerrado biome from 2019 to 2021. The methodology used training and the Support Vector Machine (SVM) classifier. To analyze the spectral peculiarities of each orbital platform, the Transformed Divergence (TD) index separability statistic was used. The results showed that for both sensors, the near-infrared (NIR) band has an essential role in the detection of the burned areas, presenting high separability. Overall, it was possible to observe that the spectral mixing problems, registration date, and the spatial resolution of 500 m were the main factors that led to commission errors ranging between 15% and 72% and omission errors between 51% and 86% for both sensors. This study showed the importance of multispectral sensors for monitoring forest fires. It was found, however, that the spectral resolution and burning date may gradually interfere with the detection process.