Search

In this study we used the PhenoRice algorithm to track recent variations of rice cultivation practices along the Senegal River Valley. Time series of MODIS imagery with 250 m spatial resolution and a nominal 8-days frequency were used as input for the algorithm to map the spatial and temporal variations of rice cultivated area and of several important phenological metrics (e.g., crop establishment and harvesting dates, length of season) for the 2003–2016 period in both the dry and the wet rice cultivation seasons. Comparison between PhenoRice results and ancillary and field data available for the Senegal part of the study area showed that the algorithm is able to track the interannual variations of rice cultivated area, despite the total detected rice area being consistently underestimated. PhenoRice estimates of crop establishment and harvesting dates resulted accurate when compared with field observations available for two sub-regions for a period of 10 years, and thus allow assessing interannual variability and tracking changes in agronomic practices. An analysis of interannual trends of rice growing practices based on PhenoRice results highlighted a clear shift of rice cultivation from the wet to the dry season starting approximately from 2008. The shift was found to be particularly evident in the delta part of the SRV. Additionally, a statistically significant trend was revealed starting 2006 towards a longer dry season (r2 = 0.81; Slope = 1.24 days y−1) and a shorter wet season (r2 = 0.65; Slope = 0.53 days y−1). These findings are in agreement with expert knowledge of changes ongoing in the area. In particular the shorter wet season is attributed to shortage of labor and equipment leading to a delay in completion of harvesting operations in the dry season, which led to the adoption of short-duration rice varieties by farmers in the wet season to avoid risk of yield losses due to climatic constraints. Aforementioned results highlight the usefulness of the PhenoRice algorithm for providing insights about recent variations in rice cultivation practices over large areas in developing countries, where high-quality up to date information about changes in agricultural practices are often lacking.

The paper proposes a fully automatic algorithm approach to map burned areas from remote sensing characterized by human interpretable mapping criteria and explainable results. This approach is partially knowledge-driven and partially data-driven. It exploits active fire points to train the fusion function of factors deemed influential in determining the evidence of burned conditions from reflectance values of multispectral Sentinel-2 (S2) data. The fusion function is used to compute a map of seeds (burned pixels) that are adaptively expanded by applying a Region Growing (RG) algorithm to generate the final burned area map. The fusion function is an Ordered Weighted Averaging (OWA) operator, learnt through the application of a machine learning (ML) algorithm from a set of highly reliable fire points. Its semantics are characterized by two measures, the degrees of pessimism/optimism and democracy/monarchy. The former allows the prediction of the results of the fusion as affected by more false positives (commission errors) than false negatives (omission errors) in the case of pessimism, or vice versa, the latter foresees if there are only a few highly influential factors or many low influential ones that determine the result. The prediction on the degree of pessimism/optimism allows the expansion of the seeds to be appropriately tuned by selecting the most suited growing layer for the RG algorithm thus adapting the algorithm to the context. The paper illustrates the application of the automatic method in four study areas in southern Europe to map burned areas for the 2017 fire season. Thematic accuracy at each site was assessed by comparison to reference perimeters to prove the adaptability of the approach to the context, estimated average accuracy metrics are omission error = 0.057, commission error = 0.068, Dice coefficient = 0.94 and relative bias = 0.0046.

In this paper, we provide an innovative contribution in the research domain dedicated to crop mapping by exploiting the of Sentinel-2 satellite images time series, with the specific aim to extract information on “where and when” crops are grown. The final goal is to set up a workflow able to reliably identify (classify) the different crops that are grown in a given area by exploiting an end-to-end (3+2)D convolutional neural network (CNN) for semantic segmentation. The method also has the ambition to provide information, at pixel level, regarding the period in which a given crop is cultivated during the season. To this end, we propose a solution called Class Activation Interval (CAI) which allows us to interpret, for each pixel, the reasoning made by CNN in the classification determining in which time interval, of the input time series, the class is likely to be present or not. Our experiments, using a public domain dataset, show that the approach is able to accurately detect crop classes with an overall accuracy of about 93% and that the network can detect discriminatory time intervals in which crop is cultivated. These results have twofold importance: (i) demonstrate the ability of the network to correctly interpret the investigated physical process (i.e., bare soil condition, plant growth, senescence and harvesting according to specific cultivated variety) and (ii) provide further information to the end-user (e.g., the presence of crops and its temporal dynamics).

Due to the low efficiency of nitrogen fertilizers in flooded rice paddies, there is a rising demand for tools able to detect crop nitrogen status in space and time to allow farmers to use the technical novelties of precision agriculture to improve fertilizer management in extensive fields. This work sets up an operational approach to increase nitrogen use efficiency of top-dressing fertilization by supporting variable rate fertilization in rice cropping systems. The procedure exploits (i) crop modelling to identify best periods for fertilization (When), (ii) Sentinel-2 imagery to draw management zones (MZ) and lead field scouting (Where), and (iii) smartphone app to measure nitrogen nutritional index (NNI) (How much). Automatically generated MZ from Sentinel-2 data were able to identify within field patches with different nutritional status and NNI data well described the crop temporal dynamic in relation to crop development and nutritional needs. The workflow was implemented to provide farmers with timely information on plant nutritional status during the 2018 growing season to define site-specific fertilization strategies implemented with variable rate technology (VRT). Tests conducted on 6 fields over 30 ha in 3 farms showed the feasibility of the proposed workflow in real farming conditions allowing a reduction of applied fertilizer up to 25% in the areas with sufficient nutritional status. Demonstration revealed that VRT based on geospatial information from integrated in-field and satellite data can provide agronomic and environmental benefits compared with standard fertilization resulting in promising outcomes both in terms of yield (increase in the range 0.2–0.5 t ha−1) and nitrogen use efficiency (increase up to 7.8%).

The paper proposes a scalable fuzzy approach for mapping the status of the environment integrating several distinct models exploiting geo big data. The process is structured into two phases: the first one can exploit products yielded by distinct models of remote sensing image interpretation defined in the scientific literature, and knowledge of domain experts, possibly ill-defined, for computing partial evidence of a phenomenon. The second phase integrates the partial evidence maps through a learning mechanism exploiting ground truth to compute a synthetic Environmental Status Indicator (ESI) map. The proposal resembles an ensemble approach with the difference that the aggregation is not necessarily consensual but can model a distinct decision attitude in between pessimistic and optimistic. It is scalable and can be implemented in a distributed processing framework, so as to make feasible ESI mapping in near real time to support land monitoring. It is exemplified to map the presence of standing water areas, indicator of water resources, agro-practices or natural hazard from remote sensing by considering different models.

sen2r is a scalable and flexible R package to enable downloading and preprocessing of Sentinel-2 satellite imagery via an accessible and easy to install interface. It allows the execution of several preprocessing steps which are commonly performed by Sentinel-2 users: searching the Sentinel-2 archive for datasets available over a spatial area of interest and in a defined time window, downloading them, applying the Sen2Cor atmospheric correction algorithm to compute surface reflectances, merging adjacent tiles, performing geometric transformations, applying a cloud mask, computing spectral indices and colour images. The package is designed to be accessible to a range of users, from beginners to skilled R users. It comes with a Graphical User Interface, which can be used to set the processing parameters and launch processing operations: this feature makes sen2r accessible also for novices with limited programming experience. High-level R functions, which enable customised image processing workflows and control over intermediate steps, can be useful to experienced remote sensing researchers. Thanks to those functions it is possible to easily schedule automatic processing chains, so to manage massive processing operations. This paper describes the main characteristics, functionalities and performance of the package and highlights its usefulness as the operational back-end of service-oriented architectures, as illustrated by the Saturno project.

Information on when and where rice is planted and harvested is important for crop management under a changing climate and for monitoring crop production for early warning and market information systems. The diversity of plant genetic, crop management, and environmental conditions leads to a wide variation in the number of rice crops per year and the dates of crop establishment and harvesting across Asia. Asia-wide rice crop calendars exist (e.g., RiceAtlas) but are based on heterogeneous data sources with varying levels of detail and are challenging to update. Earth observations can contribute to consistent and replicable crop calendars. Here we demonstrate and validate a method for generating a rice crop calendar across Asia. Our analysis at administrative unit-level is based on pixel-level analysis with the PhenoRice algorithm using MODIS imagery (2003–16) to estimate start of season (SoS) and end of season (EoS) dates. PhenoRice outputs were post-processed to generate representative statistics on the number of rice crop seasons per year and their SoS/EoS dates per administrative unit across Asia, called RICA (a RIce crop Calendar for Asia). RICA SoS and EoS dates across all seasons correlated strongly with RiceAtlas crop establishment and harvesting dates (R2 of 0.88 and 0.82 respectively, n = 1,186). The mean absolute errors were around 26 and 33 days for SoS and EoS, respectively. A detailed assessment in the Philippines where data in RiceAtlas are particularly accurate had even better results (R2 of 0.93 and 0.85 respectively, n = 131). Comparisons to other published rice calendars also suggested that RICA captured rice cropping season dates well. Our study results in a unique and validated method to estimate rice crop calendar information on continental scale from remote sensing data.

The paper proposes a transparent approach for mapping the status of environmental phenomena from multisource information based on both soft computing and machine learning. It is transparent, intended as human understandable as far as the employed criteria, and both knowledge and data-driven. It exploits remote sensing experts&rsquo, interpretations to define the contributing factors from which partial evidence of the environmental status are computed by processing multispectral images. Furthermore, it computes an environmental status indicator (ESI) map by aggregating the partial evidence degrees through a learning mechanism, exploiting volunteered geographic information (VGI). The approach is capable of capturing the specificities of local context, as well as to cope with the subjectivity of experts&rsquo, interpretations. The proposal is applied to map the status of standing water areas (i.e., water bodies and rivers and human-driven or natural hazard flooding) using multispectral optical images by ESA Sentinel-2 sources. VGI comprises georeferenced observations created both in situ by agronomists using a mobile application and by photointerpreters interacting with a geographic information system (GIS) using several information layers. Results of the validation experiments were performed in three areas of Northern Italy characterized by distinct ecosystems. The proposal showed better performances than traditional methods based on single spectral indexes.

In this work we propose an approach for mapping flooded areas from Sentinel-2 MSI (Multispectral Instrument) data based on soft fuzzy integration of evidence scores derived from both band combinations (i.e. Spectral Indices - SIs) and components of the Hue, Saturation and Value (HSV) colour transformation. Evidence scores are integrated with Ordered Weighted Averaging (OWA) operators, which model user’s decision attitude varying smoothly between optimistic and pessimistic approach. Output is a map of global evidence degree showing the plausibility of being flooded for each pixel of the input Sentinel-2 (S2) image. Algorithm set up and validation were carried out with data over three sites in Italy where water surfaces are extracted from stable water bodies (lakes and rivers), natural hazard flooding, and irrigated paddy rice fields. Validation showed more than satisfactory accuracy for the OR-like OWA operators (F-score > 0.90) with performance slightly decreased (F-score

The present short paper describes the functionalities of a desktop tool enabling non-expert users to develop downstream services for the periodic download and processing of data from Sentinel sources.

Using crop models as supporting tools for analyzing the interaction between genotype and environment represents an opportunity to identify priorities within breeding programs. This study represents the first attempt to use simulation models to define rice ideotypes improved for their resistance to biotic stressors (i.e., diseases); moreover, it extends approaches for evaluating the impact of changes in traits for tolerance to abiotic constraints (temperature shocks inducing sterility). The analysis—targeting the improvement of 34 varieties in six Italian rice districts—was focused on the impact of blast disease, and of pre-flowering cold- and heat-induced spikelet sterility. In silico ideotypes were tested at 5-km spatial resolution under current conditions and climate change scenarios centered on 2020, 2050, and 2085, derived according to the projections of two general circulation models–Hadley and NCAR–for two IPCC emission scenarios–A1B and B1. The study was performed using a dedicated simulation platform, i.e., ISIde, explicitly developed for ideotyping studies. The ideotypes improved for blast resistance obtained clear yield increases for all the combinations GCM × emission scenario × time horizon, i.e., 12.1 % average yield increase under current climate, although slightly decreasing for time windows approaching the end of the century and with a marked spatial heterogeneity in responses across districts. Concerning abiotic stressors, increasing tolerance to cold-induced sterility would lead to a substantial yield increase (+9.8 %) only for Indica-type varieties under current climate, whereas no increases are expected under future conditions and, in general, for Japonica-type varieties. Given the process-based logic behind the models used—supporting coherence of model responses under future scenarios—this study provides useful information for rice breeding programs to be realized in the medium-long term.

Agricultural monitoring systems require spatio-temporal information on widely cultivated staple crops like rice. More emphasis has been made on area estimation and crop detection than on the temporal aspects of crop cultivation, but seasonal and temporal information such as i) crop duration, ii) date of crop establishment and iii) cropping intensity are as important as area for understanding crop production. Rice cropping systems are diverse because genetic, environmental and management factors (G × E × M combinations) influence the spatio-temporal patterns of cultivation. We present a rule based algorithm called PhenoRice for automatic extraction of temporal information on the rice crop using moderate resolution hypertemporal optical imagery from MODIS. Performance of PhenoRice against spatially and temporally explicit reference information was tested in three diverse sites: rice-fallow (Italy), rice-other crop (India) and rice-rice (Philippines) systems. Regional product accuracy assessments showed that PhenoRice made a conservative, spatially representative and robust detection of rice cultivation in all sites (r2 between 0.75 and 0.92) and crop establishment dates were in close agreement with the reference data (r2 = 0.98, Mean Error = 4.07 days, Mean Absolute Error = 9.95 days, p < 0.01). Variability in algorithm performance in different conditions in each site (irrigated vs rainfed, direct seeding vs transplanting, fragmented vs clustered rice landscapes and the impact of cloud contamination) was analysed and discussed. Analysis of the maps revealed that cropping intensity and season length per site matched well with local information on agro-practices and cultivated varieties. The results show that PhenoRice is robust for deriving essential temporal descriptions of rice systems in both temperate and tropical regions at a level of spatial and temporal detail that is suitable for regional crop monitoring on a seasonal basis.

Crop simulation models are commonly used to forecast the performance of cropping systems under different hypotheses of change. Their use on a regional scale is generally constrained, however, by a lack of information on the spatial and temporal variability of environment-related input variables (e.g., soil) and agricultural practices (e.g., sowing dates) that influence crop yields. Satellite remote sensing data can shed light on such variability by providing timely information on crop dynamics and conditions over large areas. This paper proposes a method for analyzing time series of MODIS satellite data in order to estimate the inter-annual variability of winter wheat sowing dates. A rule-based method was developed to automatically identify a reliable sample of winter wheat field time series, and to infer the corresponding sowing dates. The method was designed for a case study in the Camargue region (France), where winter wheat is characterized by vernalization, as in other temperate regions. The detection criteria were chosen on the grounds of agronomic expertise and by analyzing high-confidence time-series vegetation index profiles for winter wheat. This automatic method identified the target crop on more than 56% (four-year average) of the cultivated areas, with low commission errors (11%). It also captured the seasonal variability in sowing dates with errors of ±8 and ±16 days in 46% and 66% of cases, respectively. Extending the analysis to the years 2002–2012 showed that sowing in the Camargue was usually done on or around November 1st (±4 days). Comparing inter-annual sowing date variability with the main local agro-climatic drivers showed that the type of preceding crop and the weather conditions during the summer season before the wheat sowing had a prominent role in influencing winter wheat sowing dates.

Growing non-food crops in marginal lands has been proposed as a solution to avoid land competition with food production. Mapping marginal agricultural lands is therefore fundamental for the sustainable development of rural landscapes. This study proposes a method based on remote sensing data to identify marginal agricultural lands for the production of wood biomass, and analyse potential trade-offs and synergies between the new wood crops, food production, and Ecosystem Services (ES) provided by vegetation. The province of Rovigo (northern Italy) was chosen as a representative case study. Three classes of marginal agricultural lands were mapped through the use of the Soil Adjusted Vegetation Index (SAVI): i) abandoned or unused agricultural lands, ii) potentially poorly or non-managed croplands, and iii) potentially low productivity croplands. Results showed that marginal agricultural lands cover 1.7% of the agricultural areas of the province, and approximately 13,642 MWh yr−1 of Second-Generation (2G) bioenergy can be produced in marginal agricultural areas while enhancing ES provided by vegetation, and avoiding any trade-off with food production. Since this energy potential covers just 8.4% of the total potential authorized in the province, the enhancement of ES could provide a suitable argument to support the conversion of marginal agricultural lands and increase the multifunctionality of the agricultural landscape.

Spectroradiometric field surveys, especially when addressed at heterogeneous targets and within a complex environmental context, require careful organization and structuring. This work focuses on a proposal of an operational workflow that holds together different aspects of in situ spectral data acquisition, from planning to data analysis, and is ancillary to environmental analysis based on remote sensing. The main objective is the enhancement of the information coming from acquired spectra, which is accomplished through a set of established guidelines for: campaign planning, measurement collection, and data management. The feasibility and advantages of such approach are demonstrated through a real case scenario, covering two spectroradiometric campaigns aimed at geolithology studies and performed in September 2010 and 2011 in the Dolomites (Northern Italy), where complex environmental and variable atmospheric conditions, as well as, large target variability and spectral heterogeneity occur. The workflow here adopted allowed us to face all the challenges in order to usefully accomplish our campaigns. Issues commonly arising in spectroradiometric field activities are easily foreseen, tackled and mitigated, if necessary, while preserving a high degree of flexibility and generalization for exploiting the workflow over heterogeneous areas and application fields.

Spatial assessment of environmental phenomena at regional/global scale involves the analysis and fusion of multiple, complex, multidisciplinary, and large-scale information. Since very often reliable models of such phenomena are lacking, the ''syndrome approach'' has been adapted to this purpose. In this context, there is a strong need for frameworks capable of handling data from heterogeneous sources in order to fuse them into synthetic indicators by modelling the uncertain and incomplete knowledge of the phenomenon. The approach here proposed models a syndrome by soft revision of bipolar information having heterogeneous role: a set of contextual conditions constraining the flourishing of the syndrome (negative information), identified on the basis of the expert's knowledge, and a typical pattern of notable symptoms (positive information) that are indeed proxies of observations of the syndrome occurrence. Specifically, three soft revision strategies are defined in the paper to combine negative and positive information whose overall aim is to define indicators of occurrence of an environmental syndrome with distinct objectives. The main concern of the first two revisions is to reduce the incoherence of bipolar information by taking into account the information reliability: the first strategy models a cautious decision attitude by eliminating incoherence and by modelling priorities of either negative or positive information and their partial trust, while the second strategy models a cautious and precautionary decision attitude giving credit to observations by taking into account False Negatives and False Positives. Finally, the third strategy models decision attitudes characterized by distinct level of risk. The proposed methodology is exemplified by a real case study whose objective is to define an indicator of the Rural Exodus syndrome for the Mediterranean region. In this case the method fuses biophysical related variables derived from Earth Observation (EO) data with geophysical and socioeconomic conditions based on the proposed strategies.

This work presents a simple, cost-effective, and operational approach to monitor crop water requirements at the regional scale for water management and monitoring purposes. The recommended Food and Agricultural Organization of the United Nations methodology FAO-56 calculates crop evapotranspiration using crop-specific coefficients Kc, which vary according to the crop type, health, and phenological stage. This approach, though widely applied for irrigation planning, cannot always match the appropriate crop coefficient with the actual crop phenological stage and health condition, especially in anomalous situations. Previous research demonstrated that crop coefficients and spectral vegetation indexes are correlated. Recent studies have used this relationship with high-resolution satellite data from different sensors to provide information to irrigation advisory services. However, high-resolution data are not feasible for an operational and routine monitoring of water consumption and needs. This paper tests the usefulness of time series of coarse resolution satellite data such as those collected by the moderate-resolution imaging spectroradiometer MODIS sensor, to monitor crop coefficients temporal and spatial variability and therefore crop water needs at the regional scale taking advantage of the peculiar characteristics offered by MODIS in terms of high temporal resolution and preprocessed products availability. The outlined methodology takes into account the actual growing stage of the crops and nearly real-time vegetation variations, overcoming some limitations of the traditional FAO approach while preserving the maximum operability. The analysis was carried out in the South Milan agricultural area on data referring to 2003 and 2004. The results agreed with those of other studies and proved to be able to account for the anomalous conditions of the summer in 2003. These results were then compared with those obtained using the traditional FAO crop coefficient curves built with data collected during field campaigns in the same years in rice fields. Constraints, limitations, and possible uses are discussed.

International audience; Plant diseases are responsible for major economic losses in the agricultural industry worldwide. Monitoring plant health and detecting pathogen early are essential to reduce disease spread and facilitate effective management practices. DNA-based and serological methods now provide essential tools for accurate plant disease diagnosis, in addition to the traditional visual scouting for symptoms. Although DNA-based and serological methods have revolutionized plant disease detection, they are not very reliable at asymptomatic stage, especially in case of pathogen with systemic diffusion. They need at least 1–2 days for sample harvest, processing, and analysis. Here, we describe modern methods based on nucleic acid and protein analysis. Then, we review innovative approaches currently under development. Our main findings are the following: (1) novel sensors based on the analysis of host responses, e.g., differential mobility spectrometer and lateral flow devices, deliver instantaneous results and can effectively detect early infections directly in the field; (2) biosensors based on phage display and biophotonics can also detect instantaneously infections although they can be integrated with other systems; and (3) remote sensing techniques coupled with spectroscopy-based methods allow high spatialization of results, these techniques may be very useful as a rapid preliminary identification of primary infections. We explain how these tools will help plant disease management and complement serological and DNA-based methods. While serological and PCR-based methods are the most available and effective to confirm disease diagnosis, volatile and biophotonic sensors provide instantaneous results and may be used to identify infections at asymptomatic stages. Remote sensing technologies will be extremely helpful to greatly spatialize diagnostic results. These innovative techniques represent unprecedented tools to render agriculture more sustainable and safe, avoiding expensive use of pesticides in crop protection.

The drought episode that struck Middle East countries in 2007-2009 was the worst one hitting the region in more than 60 years. An analysis of rainfall and vegetation dynamics over the temporal range covering 11 hydrological seasons from September 2002 to August 2013 has been carried out by using time series of satellite data from TRMM and MODIS platforms, over the study area covering the so called Fertile Crescent: spanning from southern Turkey, Syria, and Jordan, to Iraq and western Iran. The results of this analysis, along with supportive information about surface water reserves drawn from satellite radar altimetry suggest that the intensification of activities in new agricultural areas in southern Turkey, steadily developed during the last decade, contributed to the sensible deplenishment of water resources available over the basin.

This paper describes a mapping project carried out using both optical and SAR data on an agricultural area in northern Italy where the main crops are corn, rice and wheat. Temporal trends of backscatter and reflectance, given by the variations in vegetation growth, soil conditions and agricultural practices were analyzed and interpreted thanks to the ground-measured data. Information extracted from both optical and SAR data (vegetation indices, backscatter and texture features) were used to create training sets for implementing three different classification approaches. The work aimed at comparing early crop maps with maps derived at the end of the season. Results show that the classification accuracy obtained using only multispectral optical data is higher than the one reached using only SAR as input. Integrating both optical and SAR multitemporal features provides some advantages in terms of a more reliable crop map, especially during an early temporal stage scenario. Among the supervised algorithms tested, Maximum Likelihood shows the best overall accuracy performances at each thematic level, time step and using both optical and SAR input data.

International audience; In this contribution we analyze the problem of the fusion of time series of heterogeneous remote sensing images to serve classification and monitoring activities which can aid farming applications such as crop classification, change detection and monitoring. We propose several soft fusion operators that are based on different assumptions and model distinct desired properties. Conducted experiments on various geographic regions have been carried out and illustrate the effectiveness of our proposal.; Dans cette contribution, nous analysons le problème de la fusion de série d'images hétérogènes de télédétection au service de classification et les activités de surveillance qui peuvent aider les applications agricoles telles que la classification des cultures, la détection des changements et de la surveillance. Nous proposons plusieurs opérateurs de fusion mous qui sont basés sur diverses hypothèses et modèle distinct désiré. Les expériences menées sur diverses régions géographiques ont été réalisées afin d'illustrer l'efficacité de notre proposition.

Rice farming, one of the most important agricultural activities in the world producing staple food for nearly one-fifth of the global population, covers 153 MHa every year corresponding to a production of more than 670 Mton. Retrieve updated information on actual rice cultivated areas and on key phenological stages occurrence is fundamental to support policy makers, rice farmers and consumers providing the necessary information to increase food security and control market prices. In particular, remote sensing is very important to retrieve spatial distributed information on large scale fundamental to set up operational agro-ecosystem monitoring tool. The present work wants to assess the reliability of automatic image processing algorithm for the identification of rice cultivated areas. A method, originally tested for Asian tropical rice areas, was applied on temperate European Mediterranean environment. Modifications of the method have been evaluated to adapt the original algorithm to the different experimental conditions. Finally, a novel approach based on phenological detection analysis has been tested on Northern Italy rice district. Rice detection was conducted using times series of Vegetation Indices derived by MODIS MOD09A1 products for the year 2006 and the accuracy of the maps was assessed using available thematic cartography. Error matrix analysis shows that the new proposed method, applied in a fully automatic way, is comparable to the results of the original approach when it is customized and adapted for the specific study area. The new algorithm minimizes the use of external data and provides also spatial distributed information on crop phenological stages.

In this paper, we propose a novel method which involves neural adaptive techniques for identifying salient features and for classifying high dimensionality data. In particular a network pruning algorithm acting on MultiLayer Perceptron topology is the foundation of the feature selection strategy. Feature selection is implemented within the back-propagation learning process and based on a measure of saliency derived from bell functions positioned between input and hidden layers and adaptively varied in shape and position during learning. Performances were evaluated experimentally within a Remote Sensing study, aimed to classify hyperspectral data. A comparison analysis was conducted with Support Vector Machine and conventional statistical and neural techniques. As seen in the experimental context, the adaptive neural classifier showed a competitive behavior with respect to the other classifiers considered; it performed a selection of the most relevant features and showed a robust behavior operating under minimal training and noisy situations.

Hyperspectral imaging is becoming an important analytical tool for generating land-use map. High dimensionality in hyperspectral remote sensing data, on one hand, provides us with more potential discrimination power for classification tasks. On the other hand, the classification performance improves up to a limited point as additional features are added, and then deteriorates due to the limited number of training samples. Proceeding from these considerations, the present work is aimed to systematically evaluate the robustness of novel classification techniques in classifying hyperspectral data under the twofold condition of high dimensionality and minimal training. We consider in the study a neural adaptive model based on Multi Layer Perceptron (MLP). Accuracy has been evaluated experimentally, classifying MIVIS Hyperspectral data to identify different typology of vegetation in Ticino Regional Park. A performance analysis has been conducted comparing the novel approach with Support Vector Machine and conventional statistical and neural techniques. The adaptive model shows advantages especially when mixed data are presented to the classifiers in combination with minimal training conditions.

In this paper, we propose a method able to fuse spectral information with spatial contextual information in order to solve “operationally” classification problem. The salient aspect of the method is the integration of heterogeneous data within a Multi-Layer Perceptron model. Spatial and spectral relationships are not explicitly formalized in an attempt to limit design and computational complexity; raw data are instead presented directly as input to the neural network classifier. The method in particular addresses new open problems in processing hyperspectral and high resolution data finding solution for multisource analysis. Experimental results in real domain show this fusing approach is able to produce accurate classification. The method in fact is able to handle the problem of a volumetric mixture typical of natural forest ecosystems identifying the different surfaces present under the tree canopy. The understory map, produced by the neural classification method, was used as input to the inversion of radiative transfer models that show a significant increase in the retrieval of important biophysical vegetation parameter.

In pipeline management the accurate prediction of weak displacements is a crucial factor in drawing up a prevention policy since the accumulation of these displacements over a period of several years can lead to situations of high risk. This work addresses the specific problem related to the prediction of displacements induced by rainfall in unstable areas, of known geology, and crossed by underground pipelines. A neural model has been configured which learns of displacements from instrumented sites (where inclinometric measurements are available) and is able to generalise to other sites not equipped with inclinometers.

The monitoring of the environment's status at continental scale involves the integration of information derived by the analysis of multiple, complex, multidisciplinary, and large-scale phenomena. Thus, there is a need to define synthetic Environmental Indicators (EIs) that concisely represent these phenomena in a manner suitable for decision-making. This research proposes a flexible system to define EIs based on a soft fusion of contributing environmental factors derived from multi-source spatial data (mainly Earth Observation data). The flexibility is twofold: the EI can be customized based on the available data, and the system is able to cope with a lack of expert knowledge. The proposal allows a soft quantifier-guided fusion strategy to be defined, as specified by the user through a linguistic quantifier such as 'most of'. The linguistic quantifiers are implemented as Ordered Weighted Averaging operators. The proposed approach is applied in a case study to demonstrate the periodical computation of anomaly indicators of the environmental status of Africa, based on a 7-year time series of dekadal Earth Observation datasets. Different experiments have been carried out on the same data to demonstrate the flexibility and robustness of the proposed method.