In Europe, the 2003 summer heat wave damaged forested areas. This study aims to compare two approaches of NDVI time series analysis to monitor forest decline. Both methods analyze the trend of vegetation activity from 2000 to 2011. The first method is based on a phenometric related to spring vegetation activity, calculated for each year during the 2000-2011 period. In the second method (BFAST), the trend comes from the decomposition of the NDVI time series into three additive components: trend, seasonal and remainder. The two approaches gave similar results for estimated trends. The main advantage of BFAST is its ability to detect breakpoints in the linear trend. It allowed to highlight here the impact of exceptional events, like 2003 summer drought, on the development of forest stands. In the last part of our study, we implemented a validation based on in situ observations. Health status of silver fir stands was estimated analyzing the trees architecture. Significant relationships were highlighted between the indicator of spring vitality derived from remote sensing images and the observed status of forest stands.

No abstract

This paper examines the potential of MODIS-NDVI time series for detecting clear-cuts in a coniferous forest stand in the south of France. The proposed approach forms part of a survey monitoring the status of forest health and evaluating the forest decline phenomena observed over the last few decades. One of the prerequisites for this survey was that a rapid and easily reproducible method had to be developed that differentiates between forest clear-cuts and changes in forest health induced by environmental factors such as summer droughts. The proposed approach is based on analysis of the breakpoints detected within NDVI time series, using the "Break for Additive Seasonal and Trend" (BFAST) algorithm. To overcome difficulties detecting small areas on the study site, we chose a probabilistic approach based on the use of a conditional inference tree. For model calibration, clear-cut reference data were produced at MODIS resolution (250 m). According to the magnitude of the detected breakpoints, probability classes for the presence of clear-cuts were defined, from greater than 90% to less than 3% probability of a clear-cut. One of the advantages of the probabilistic model is that it allows end users to choose an acceptable level of uncertainty depending on the application. In addition, the use of BFAST allows events to be dated, thus OPEN ACCESSRemote Sens. 2015, 7 3589 making it possible to perform a retrospective analysis of decreases in forest vitality in the study area.

[1] Wildfires are a prevalent natural hazard in the south of France. Planners need a permanent fire danger assessment valid for several years over a territory as large and heterogeneous as Midi-Pyrénées region. To this end, we developed an expert knowledgebased index model adapted to the specific features of the study area. The fire danger depends on two complementary elements: spatial occurrence and fire intensity. Among the GIS layers identified as input variables for modeling, vegetation fire susceptibility is one of the most influent. However, the main difficulty at this scale is the scarcity or the lack of exhaustiveness of the data. In this respect, remote sensing imagery is capable of providing relevant information. We proposed to calculate an annual relative greenness index (annual RGRE) that reflects vegetation dryness in summer. We processed times series of Normalized Difference Vegetation Index (NDVI) from SPOT-VEGETATION images over the last six available years (1998 to 2003). The first step was to verify that these images characterize vegetation types and highlight intraannual and interannual response variability. It is then possible to identify phenological stages corresponding to the maximum NDVI (and therefore to maximum photosynthetic activity) during the growing season, the minimum NDVI at the end of the growing season and the minimum NDVI during winter period. These phenology metrics ground the annual RGRE calculation. Values obtained for each observation year show significant correlation (r 2 = 0.70) with the De Martonne aridity index calculated for the same period. A synthesis of yearly index was integrated in the model as a variable that expresses fire susceptibility.Citation: Chéret, V., and J. P. Denux (2007), Mapping wildfire danger at regional scale with an index model integrating coarse spatial resolution remote sensing data,

This study was conducted to assess fire susceptibility of Mediterranean vegetation by analyzing a time series of Moderate Resolution Imaging Spectroradiometer (MODIS) Terra images from 2000 to 2006. Synthetic indicators of vegetation status were defined based on analysis of annual variations of the Normalized Difference Vegetation Index (NDVI) and an understanding of phenological cycles. Spring and annual greenness indicators were calculated by combining NDVI values measured at different key phenological stages. The various fire susceptibility indicators were used to characterize fluctuations of vegetation activity related to changes in photosynthetic activity and fuel dryness. Susceptibility indicators were also mapped, and statistical relationships with meteorological conditions were identified.

Wine growing needs to adapt to confront climate change. In fact, the lack of water becomes more and more important in many regions. Whereas vineyards have been located in dry areas for decades, so they need special resilient varieties and/or a sufficient water supply at key development stages in case of severe drought. With climate change and the decrease of water availability, some vineyard regions face difficulties because of unsuitable variety, wrong vine management or due to the limited water access. Decision support tools are therefore required to optimize water use or to adapt agronomic practices. This study aimed at monitoring vine water status at a large scale with Sentinel-2 images. The goal was to provide a solution that would give spatialized and temporal information throughout the season on the water status of the vines. For this purpose, thirty six plots were monitored in total over three years (2018, 2019 and 2020). Vine water status was measured with stem water potential in field measurements from pea size to ripening stage. Simultaneously Sentinel-2 images were downloaded and processed to extract band reflectance values and compute vegetation indices. In our study, we tested five supervised regression machine learning algorithms to find possible relationships between stem water potential and data acquired from Sentinel-2 images (bands reflectance values and vegetation indices). Regression model using Red, NIR, Red-Edge and SWIR bands gave promising result to predict stem water potential (R2=0.40, RMSE=0.26).

Reliable estimates of poplar plantations area are not available at the French national scale due to the unsuitability and low update rate of existing forest databases for this short-rotation species. While supervised classification methods have been shown to be highly accurate in mapping forest cover from remotely sensed images, their performance depends to a great extent on the labelled samples used to build the models. In addition to their high acquisition cost, such samples are often scarce and not fully representative of the variability in class distributions. Consequently, when classification models are applied to large areas with high intra-class variance, they generally yield poor accuracies. In this paper, we propose the use of active learning (AL) to efficiently adapt a classifier trained on a source image to spatially distinct target images with minimal labelling effort and without sacrificing classification performance. The adaptation consists in actively adding to the initial local model, new relevant training samples from other areas, in a cascade that iteratively improves the generalisation capabilities of the classifier, leading to a global model tailored to different areas. This active selection relies on uncertainty sampling to directly focus on the most informative pixels for which the algorithm is the least certain of their class labels. Experiments conducted on Sentinel-2 time series showed that when the same number of training samples was used, active learning outperformed passive learning (random sampling) by up to 5% of overall accuracy and up to 12% of class F-score. In addition, and depending on the class considered, the random sampling required up to 50% more samples to achieve the same performance of an active learning-based model. Moreover, the results demonstrate the suitability of the derived global model to accurately map poplar plantations among other tree species with overall accuracy values up to 14% higher than those obtained with local models. The proposed approach paves the way for national-scale mapping in an operational context.

Detailed forest-cover mapping at a regional scale by supervised classification is technically limited by various factors. This study evaluates the ability of a landscape stratification method to improve classification accuracy. An object-based segmentation technique (OBIA) was performed to delineate radiometrically homogeneous regions into the study area, used as strata for the classification of a time series of Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index (NDVI) data. As a reduction of the spatial variability of the signatures of the vegetation classes is expected, Maximum Likelihood Classifier (MLC) was used to analyse potential effects on classification accuracy. Accuracy assessment was based on the calculation of kappa coefficient (.) and reject fraction (RF). The values obtained with and without stratification were compared, to assess their global and per-stratum influence on the quality of a detailed forest-cover map (20 different classes). To study the influence of topographical and landscape stratum characteristics on classification accuracy, eight indicators were calculated. Their correlation with. and RF differences due to stratification was analysed. Our study showed that stratification improved global and per-stratum classification accuracy and in parallel caused an RF increase. Both these evolutions are not conditioned by the stratum topographical and landscape characteristics but strongly influenced by stratum and classified vegetation area