Crop area extent estimates and crop type maps provide crucial information for agricultural monitoring and management. Remote sensing imagery in general and, more specifically, high temporal and high ...spatial resolution data as the ones which will be available with upcoming systems, such as Sentinel-2, constitute a major asset for this kind of application. The goal of this paper is to assess to what extent state-of-the-art supervised classification methods can be applied to high resolution multi-temporal optical imagery to produce accurate crop type maps at the global scale. Five concurrent strategies for automatic crop type map production have been selected and benchmarked using SPOT4 (Take5) and Landsat 8 data over 12 test sites spread all over the globe (four in Europe, four in Africa, two in America and two in Asia). This variety of tests sites allows one to draw conclusions applicable to a wide variety of landscapes and crop systems. The results show that a random forest classifier operating on linearly temporally gap-filled images can achieve overall accuracies above 80% for most sites. Only two sites showed low performances: Madagascar due to the presence of fields smaller than the pixel size and Burkina Faso due to a mix of trees and crops in the fields. The approach is based on supervised machine learning techniques, which need in situ data collection for the training step, but the map production is fully automatic.
Supervised classification systems used for land cover mapping require accurate reference databases. These reference data come generally from different sources such as field measurements, thematic ...maps, or aerial photographs. Due to misregistration, update delay, or land cover complexity, they may contain class label noise, i.e., a wrong label assignment. This study aims at evaluating the impact of mislabeled training data on classification performances for land cover mapping. Particularly, it addresses the random and systematic label noise problem for the classification of high resolution satellite image time series. Experiments are carried out on synthetic and real datasets with two traditional classifiers: Support Vector Machines (SVM) and Random Forests (RF). A synthetic dataset has been designed for this study, simulating vegetation profiles over one year. The real dataset is composed of Landsat-8 and SPOT-4 images acquired during one year in the south of France. The results show that both classifiers are little influenced for low random noise levels up to 25%-30%, but their performances drop down for higher noise levels. Different classification configurations are tested by increasing the number of classes, using different input feature vectors, and changing the number of training instances. Algorithm complexities are also analyzed. The RF classifier achieves high robustness to random and systematic label noise for all the tested configurations; whereas the SVM classifier is more sensitive to the kernel choice and to the input feature vectors. Finally, this work reveals that the cross-validation procedure is impacted by the presence of class label noise.
Long term flux measurements of different crop species are necessary to improve our understanding of management and climate effects on carbon flux variability as well as cropland potential in ...terrestrial carbon sequestration. The main objectives of this study were to analyse the seasonal dynamics of CO
2 fluxes and to establish the effects of climate and cropland management on the annual carbon balance.
CO
2 fluxes were measured by means of the eddy correlation (EC) method over two cropland sites, Auradé and Lamasquère, in South West France for a succession of three crops: rapeseed, winter wheat and sunflower at Auradé, and triticale, maize and winter wheat at Lamasquère. The net ecosystem exchange (NEE) was partitioned into gross ecosystem production (GEP) and ecosystem respiration (R
E) and was integrated over the year to compute net ecosystem production (NEP). Different methodologies tested for NEP computation are discussed and a methodology for estimating NEP uncertainty is presented.
NEP values ranged between −369
±
33
g
C
m
−2
y
−1 for winter wheat at Lamasquère in 2007 and 28
±
18
g
C
m
−2
y
−1 for sunflower at Auradé in 2007. These values were in good agreement with NEP values reported in the literature, except for maize which exhibited a low development compared to the literature. NEP was strongly influenced by the length of the net carbon assimilation period and by interannual climate variability. The warm 2007 winter stimulated early growth of winter wheat, causing large differences in GEP, R
E and NEE dynamics for winter wheat when compared to 2006. Management had a strong impact on CO
2 flux dynamics and on NEP. Ploughing interrupted net assimilation during voluntary re-growth periods, but it had a negligible short term effect when it occurred on bare soil. Re-growth events after harvest appeared to limit carbon loss: at Lamasquère in 2005 re-growth contributed to store up to 50
g
C
m
−2. Differences in NEE response to climatic variables (VPD, light quality) and vegetation index were addressed and discussed.
Net biome production (NBP) was calculated yearly based on NEP and considering carbon input through organic fertilizer and carbon output through harvest. For the three crops, the mean NBP at Auradé indicated a nearly carbon balanced ecosystem, whereas Lamasquère lost about 100
g
C
m
−2
y
−1; therefore, the ecosystem behaved as a carbon source despite the fact that carbon was imported through organic fertilizer. Carbon exportation through harvest was the main cause of this difference between the two sites, and it was explained by the farm production type. Lamasquère is a cattle breeding farm, exporting most of the aboveground biomass for cattle bedding and feeding, whereas Auradé is a cereal production farm, exporting only seeds.
The convergence of new EO data flows, new methodological developments and cloud computing infrastructure calls for a paradigm shift in operational agriculture monitoring. The Copernicus Sentinel-2 ...mission providing a systematic 5-day revisit cycle and free data access opens a completely new avenue for near real-time crop specific monitoring at parcel level over large countries. This research investigated the feasibility to propose methods and to develop an open source system able to generate, at national scale, cloud-free composites, dynamic cropland masks, crop type maps and vegetation status indicators suitable for most cropping systems. The so-called Sen2-Agri system automatically ingests and processes Sentinel-2 and Landsat 8 time series in a seamless way to derive these four products, thanks to streamlined processes based on machine learning algorithms and quality controlled in situ data. It embeds a set of key principles proposed to address the new challenges arising from countrywide 10 m resolution agriculture monitoring. The full-scale demonstration of this system for three entire countries (Ukraine, Mali, South Africa) and five local sites distributed across the world was a major challenge met successfully despite the availability of only one Sentinel-2 satellite in orbit. In situ data were collected for calibration and validation in a timely manner allowing the production of the four Sen2-Agri products over all the demonstration sites. The independent validation of the monthly cropland masks provided for most sites overall accuracy values higher than 90%, and already higher than 80% as early as the mid-season. The crop type maps depicting the 5 main crops for the considered study sites were also successfully validated: overall accuracy values higher than 80% and F1 Scores of the different crop type classes were most often higher than 0.65. These respective results pave the way for countrywide crop specific monitoring system at parcel level bridging the gap between parcel visits and national scale assessment. These full-scale demonstration results clearly highlight the operational agriculture monitoring capacity of the Sen2-Agri system to exploit in near real-time the observation acquired by the Sentinel-2 mission over very large areas. Scaling this open source system on cloud computing infrastructure becomes instrumental to support market transparency while building national monitoring capacity as requested by the AMIS and GEOGLAM G-20 initiatives.
Display omitted
•First ever national crop mapping at 10 m for Mali, Ukraine and South-Africa•Near real time agriculture monitoring at parcel level made operational nationwide•Demonstration across the world of multi-sensor EO exploitation for crop monitoring•Sentinel-2 time series mapping crop type at 10 m resolution along the growing season•Sen2-Agri: an innovative system to monitor crops in any country around the globe
The exploitation of new high revisit frequency satellite observations is an important opportunity for agricultural applications. The Sentinel-2 for Agriculture project S2Agri ...(http://www.esa-sen2agri.org/SitePages/Home.aspx) is designed to develop, demonstrate and facilitate the Sentinel-2 time series contribution to the satellite EO component of agriculture monitoring for many agricultural systems across the globe. In the framework of this project, this article studies the construction of a dynamic cropland mask. This mask consists of a binary "annual-cropland/no-annual-cropland" map produced several times during the season to serve as a mask for monitoring crop growing conditions over the growing season. The construction of the mask relies on two classical pattern recognition techniques: feature extraction and classification. One pixel- and two object-based strategies are proposed and compared. A set of 12 test sites are used to benchmark the methods and algorithms with regard to the diversity of the agro-ecological context, landscape patterns, agricultural practices and actual satellite observation conditions. The classification results yield promising accuracies of around 90% at the end of the agricultural season. Efforts will be made to transition this research into operational products once Sentinel-2 data become available.
Satellite and airborne optical sensors are increasingly used by scientists, and policy makers, and managers for studying and managing forests, agriculture crops, and urban areas. Their data acquired ...with given instrumental specifications (spectral resolution, viewing direction, sensor field-of-view, etc.) and for a specific experimental configuration (surface and atmosphere conditions, sun direction, etc.) are commonly translated into qualitative and quantitative Earth surface parameters. However, atmosphere properties and Earth surface 3D architecture often confound their interpretation. Radiative transfer models capable of simulating the Earth and atmosphere complexity are, therefore, ideal tools for linking remotely sensed data to the surface parameters. Still, many existing models are oversimplifying the Earth-atmosphere system interactions and their parameterization of sensor specifications is often neglected or poorly considered. The Discrete Anisotropic Radiative Transfer (DART) model is one of the most comprehensive physically based 3D models simulating the Earth-atmosphere radiation interaction from visible to thermal infrared wavelengths. It has been developed since 1992. It models optical signals at the entrance of imaging radiometers and laser scanners on board of satellites and airplanes, as well as the 3D radiative budget, of urban and natural landscapes for any experimental configuration and instrumental specification. It is freely distributed for research and teaching activities. This paper presents DART physical bases and its latest functionality for simulating imaging spectroscopy of natural and urban landscapes with atmosphere, including the perspective projection of airborne acquisitions and LIght Detection And Ranging (LIDAR) waveform and photon counting signals.
Flavescence dorée is a grapevine disease affecting European vineyards which has severe economic consequences and containing its spread is therefore considered as a major challenge for viticulture. ...Flavescence dorée is subject to mandatory pest control including removal of the infected vines and, in this context, automatic detection of Flavescence dorée symptomatic vines by unmanned aerial vehicle (UAV) remote sensing could constitute a key diagnosis instrument for growers. The objective of this paper is to evaluate the feasibility of discriminating the Flavescence dorée symptoms in red and white cultivars from healthy vine vegetation using UAV multispectral imagery. Exhaustive ground truth data and UAV multispectral imagery (visible and near-infrared domain) have been acquired in September 2015 over four selected vineyards in Southwest France. Spectral signatures of healthy and symptomatic plants were studied with a set of 20 variables computed from the UAV images (spectral bands, vegetation indices and biophysical parameters) using univariate and multivariate classification approaches. Best results were achieved with red cultivars (both using univariate and multivariate approaches). For white cultivars, results were not satisfactory either for the univariate or the multivariate. Nevertheless, external accuracy assessment show that despite problems of Flavescence dorée and healthy pixel misclassification, an operational Flavescence dorée mapping technique using UAV-based imagery can still be proposed.
Temperate forests are under climatic and economic pressures. Public bodies, NGOs and the wood industry are looking for accurate, current and affordable data driven solutions to intensify wood ...production while maintaining or improving long term sustainability of the production, biodiversity, and carbon sequestration. Free tools and open access data have already been exploited to produce accurate quantitative forest parameters maps suitable for policy and operational purposes. These efforts have relied on different data sources, tools, and methods that are tailored for specific forest types and climatic conditions. We hypothesized we could build on these efforts in order to produce a generic method suitable to perform as well or better in a larger range of settings. In this study we focus on building a generic approach to create forest parameters maps and confirm its performance on a test site: a maritime pine (Pinus pinaster) forest located in south west of France. We investigated and assessed options related with the integration of multiple data sources (SAR L- and C-band, optical indexes and spatial texture indexes from Sentinel-1, Sentinel-2 and ALOS-PALSAR-2), feature extraction, feature selection and machine learning techniques. On our test case, we found that the combination of multiple open access data sources has synergistic benefits on the forest parameters estimates. The sensibility analysis shows that all the data participate to the improvements, that reach up to 13.7% when compared to single source estimates. Accuracy of the estimates is as follows: aboveground biomass (AGB) 28% relative RMSE, basal area (BA) 27%, diameter at breast height (DBH) 20%, age 17%, tree density 24%, and height 13%. Forward feature selection and SVR provided the best estimates. Future work will focus on validating this generic approach in different settings. It may prove beneficial to package the method, the tools, and the integration of open access data in order to make spatially accurate and regularly updated forest structure parameters maps effortlessly available to national bodies and forest organizations.
The VENμS mission launched in 2017 provides multispectral optical images of the land surface with a 2-day revisit time at 5 m resolution for over 100 selected sites. A few sites are subject to ...seasonal snow accumulation, which gives the opportunity to monitor the variations of the snow cover area at unprecedented spatial and temporal resolution. However, the 12 spectral bands of VENμS only cover the visible and near-infrared region of the spectra while existing snow detection algorithms typically make use of a shortwave infrared band to determine the presence of snow. Here, we evaluate two alternative snow detection algorithms. The first one is based on a normalized difference index between the near-infrared and the visible bands, and the second one is based on a machine learning approach using the Theia Sentinel-2 snow products as training data. Both approaches are tested using Sentinel-2 data (as surrogate of VENμS data) as well as actual VENμS in the Pyrenees and the High Atlas. The results confirm the possibility of retrieving snow cover without SWIR with a slight loss in performance. As expected, the results confirm that the machine learning method provides better results than the index-based approach (e.g., an RMSE equal to the learning method 1.35% and for the index-based method 10.80% in the High Atlas.). The improvement is more evident in the Pyrenees probably due to the presence of vegetation which complicates the spectral signature of the snow cover area in VENμS images.
Among many indirect approaches to retrieve effective leaf area index (LAI), hemispherical photography is now widely used by the scientific community in forestry applications. A recent software ...(CAN_EYE) is used to estimate effective and true LAI from unidirectional gap fractions measured in crops. The effective LAI is computed with the Poisson law whereas the true LAI is estimated introducing a clumping index in the Poisson law. The clumping index estimation is based on the Lang and Xiang averaging method. CAN_EYE includes an automatic image classification and allows the processing of series of photographs which is mandatory to sample the spatial variability of the canopy. The objective of this study is to determine if the use of the clumping index in the gap fraction formulation improves seasonal LAI estimates of crops. Hemispherical photographs were taken throughout two growing seasons over wheat, sunflower and maize canopies. CAN_EYE LAI estimates were then compared to destructive LAI. The conditions under which photographs were acquired and processed are discussed. For the three crops studied here, the minimum distance required between camera and canopy is 1
m. When feasible, there is a clear advantage in acquiring the images from above canopies and on overcast days to facilitate the image classification. For wheat and sunflower, the best LAI estimates are assessed with effective LAI (RMSE of 0.15,
y
=
0.9540
x for wheat and RMSE of 0.38,
y
=
0.8427
x for sunflower). For maize, the best LAI estimates are obtained using the clumping index (RMSE of 0.39 and
y
=
0.9010
x). Despite good fits between CAN_EYE and destructive LAI estimates, compensation effects between leaf area index and leaf angle distribution may occur during the inversion procedure. Moreover, values of clumping index given by CAN_EYE are in certain cases correlated with the size of the cells used to divide photographs. The Lang and Xiang averaging method introduced into CAN-EYE should be improved.