Résumé
Dataset description This dataset consists of environmental variables describing the urban landscapes over metropolitan Montpellier, France. Two very high-resolution Pléiades satellite images, acquired on November 2 and 6, 2021, were used to derive these environmental variables. Orthorectified Pléiades images include a 50cm-resolution panchromatic band and four 2m-resolution multispectral bands (blue, red, green, near-infrared). A mosaic of the two images was made using Orfeo ToolBox (OTB) software (v. 8.1.2) to create a complete image of the Montpellier metropolitan area. Spectral indices : Spectral indice were obtained from multispectral Pléaides images by the following equations: Normalized Difference Vegetation Index (NDVI) and Brightness Index (BI) were computed using the formula NDVI = (NIR - RED) / (NIR + RED) and BI =√(RED2+NIR2), where "NIR" refers to the reflectance measured in the near-infrared band and "RED" corresponds to the reflectance measured in the red spectral band. A threshold was applied to the NDVI (> 0.2) to retain only areas with significant vegetation cover. For the BI, values above 500 were selected to obtain an urban layer. We used the NDVI vegetation layer to reclassify the areas classified as buildings by the BI but where there actually was vegetation. Processing was carried out using the QGIS software v. 3.16. Textural indices : Textural indices were extracted from the panchromatic Pléaides images and the spectral indices NDVI and BI using the FOTOTEX algorithm (Teillet et al., 2021, 2024) to characterize the structure of urban types within the study area. In FOTOTEX, Fourier transforms combined with principal component analyses (PCA) convert the textural information within the image into a frequency signal and reduce it into three principal components (PC1, PC2 and PC3), representing distinct texture indices. Processing was carried out using python v. 3.7. FOTOTEX was applied with the block method to the panchromatic image (PAN) with an analysis window size of 201 pixels and to spectral indices (NDVI and BI) with a window size of 101 pixels (i.e., 202 meters). Grid description : grid_MTP_complete_IS_IT.tif This grid of 202m is composed of environmental variables characterizing the Montpellier metropolitan area, made up of the localities of Baillargues, Beaulieu, Castelnau-le-Lez, Clapiers, Cournonsec, Cournonterral, Fabrègues, Grabels, Jacou, Juvignac, Lattes, Lavérune, Le Crès, Montpellier, Montaud, Montferrier-sur-Lez, Murviel-lès-Montpellier, Pérols, Pignan, Prades-le-Lez, Restinclières, Saint-Brès, Saint-Clément-de-Rivière, Saint-Drézéry, Saint-Geniès-des-Mourgues, Saint-Georges-d'Orques, Saint-Jean-de-Védas, Saussan, Sussargues, Vendargues, Villeneuve-lès-Maguelone. in Occitanie, France Fields: NDVI_mean: Mean of Normalized Difference Vegetation Index NDVI_stdev: Standard deviation of Normalized Difference Vegetation Index NDVI_max: Maximum of Normalized Difference Vegetation Index IB_mean: Mean of Brightness Index IB_stdev: Standard deviation of Brightness Index IB_max: Maximum of Brightness Index FOTO_IB_PC1: First principal component result from FOTOTEX algorithm applied to Brightness Index with the following parameters FOTO_IB_PC2: Second principal component result from FOTOTEX algorithm applied to Brightness Index FOTO_IB_PC3: Third principal component result from FOTOTEX algorithm applied to Brightness Index FOTO_NDVI_PC1: First principal component result from FOTOTEX algorithm applied to Normalized Difference Vegetation Index FOTO_NDVI_PC2: Second principal component result from FOTOTEX algorithm applied to Normalized Difference Vegetation Index FOTO_NDVI_PC3: Third principal component result from FOTOTEX algorithm applied to Normalized Difference Vegetation Index FOTO_PAN_PC1: First principal component result from FOTOTEX algorithm applied to the panchromatic band Pléiades with the following parameters FOTO_PAN_PC2: Second principal component result from FOTOTEX algorithm applied to the panchromatic band Pléiades with the following parameters FOTO_PAN_PC3: Third principal component result from FOTOTEX algorithm applied to the panchromatic band Pléiades with the following parameters Example of application of this dataset in Biomod2 : This grid of environmental layers describing urban landscapes was used in a species distribution model (SDMs) to calculate probabilities of larval presence Aedes albopictus in public spaces in Montpellier, France. We used in-situ observations of potential breeding sites from operational services of EID-MED (Entente Interdépartementale pour la Démoustication du littoral Méditerranéen) made exclusively in the public domain (i.e. storm drain, telecom cable chambers). SDMs were implemented using the RStats biomod2 package v. 4.2-5-2. Biomod2 allows building and evaluating individual SDMs and combining them into an ensemble model. Here are the results of SDMs ensemble model which allows to obtain probabilities of presence of Aedes albopictus in public spaces in the Montpellier metropolitan area. Grid result of biomod2 : proj_positif_070_075.tif Fields: positif_EMmeanbyTSS_mergedData_mergedAlgo: Mean of probabilities by True Skill Statistic (TSS) over the final selected models (i.e., artificial neural networks (ANNs), flexible discriminant analysis (FDA), generalized linear models (GLM), multiple adaptive regression splines (MARS), maximum entropy (MAXNET), eXtreme gradient boosting training (XGBOOST)) positif_EMcvbyTSS_mergedData_mergedAlgo: Coefficient of variation (sd / mean) of probabilities by True Skill Statistic (TSS) metrics over the selected models. If the CV gets a high evaluation score, it means that the uncertainty is high where the species is observed positif_EMmeanbyROC_mergedData_mergedAlgo: Mean of probabilities by Relative Operating Characteristic (ROC) over the selected models positif_EMcvbyROC_mergedData_mergedAlgo: Coefficient of variation (sd / mean) of probabilities by Relative Operating Characteristic (ROC) metrics over the selected models