Despite their high species diversity and endemism, island ecosystems are extremely vulnerable to invasive plant species (IPS). Reunion Island, located in the Indian Ocean, represents the largest area of intact vegetation in the Mascarene archipelago. However, the flora of Reunion Island is mostly threatened by IPS. Ecological modelling is crucial for studying IPS dynamics and for guiding management decisions on islands. To improve the capacity of IPS management in Reunion Island, this study aimed: (1) to predict the future spatio-temporal dynamics of IPS at island scale, (2) to better understand IPS spread factors. This work was conducted through an active “research-action” partnership. We designed a hybrid model to simulate the spatial spread over time of IPS on Reunion Island. We selected three species to be modelled: Anthoxanthum odoratum, Hiptage benghalensis and Solanum mauritianum. The hybrid model predicted an increase of 30 % of the current invaded surface by A. odoratum in Key Biodiversity Areas (KBAs); an increase of 20 % for S. mauritianum; and a spread outside of KBAs for H. benghalensis. The framework of the model, based on generic life-cycle mechanisms and species distribution models, can be adapted to other IPS of different plant types (such as lianas, shrubs and trees). The model framework is also adaptable and can be used with different climate change scenarios. With the implementation of IPS control actions also planned, the model could be used as a support tool by managers for analysing IPS trends spreading into KBAs and for taking appropriate actions.

The availability of crop type reference datasets for satellite image classification is very limited for complex agricultural systems as observed in developing and emerging countries. Indeed, agricultural land use is very dynamic, agricultural censuses are often poorly georeferenced and crop types are difficult to interpret directly from satellite imagery. In this paper, we present a database made of 24 datasets collected in a standardized manner over nine sites within the framework of the international JECAM (Joint Experiment for Crop Assessment and Monitoring) initiative; the sites were spread over seven countries of the tropical belt, and the number of data collection years depended on the site (from 1 to 7 years between 2013 and 2020). These quality-controlled datasets are distinguished by in situ data collected at the field scale by local experts, with precise geographic coordinates, and following a common protocol. Altogether, the datasets completed 27 074 polygons (20 257 crops and 6817 noncrops, ranging from 748 plots in 2013 (one site visited) to 5515 in 2015 (six sites visited)) documented by detailed keywords. These datasets can be used to produce and validate agricultural land use maps in the tropics. They can also be used to assess the performances and robustness of classification methods of cropland and crop types/practices in a large range of tropical farming systems. The dataset is available at https://doi.org/10.18167/DVN1/P7OLAP (Jolivot et al., 2021).

Food security is a crucial issue in the Sahel and could be endangered by climate change and demographic pressure during the 21st century. Higher temperatures and changes in rainfall induced by global warming are threatening rainfed agriculture in this region while the population is expected to increase approximately three-fold until 2050. Our study quantifies the impact of climate change on food security by combining climate modelling (16 models from CMIP5), crop yield (simulated by agronomic model, SARRA-O) and demographic evolution (provided by UN projection) under two future climatic scenarios. We simulate yield for the main crops in five countries in West Africa and estimate the population pressure on crop production to assess the number of available cereal production per capita. We found that, although uncertain, the African monsoon evolution leads to an increase of rainfall in Eastern Sahel and a decrease in Western Sahel under the RCP8.5 (Representative Concentration Pathway) scenario from IPCC, leading to the higher temperature increase by the end of the 21st century. With regard to the abundance of food for the inhabitants, all the scenarios in each country show that in 2050, local agricultural production will be below 50 kg per capita. This situation can have impact on crop import and regional migration.

Smallholder agriculture provides 90 % of primary food production in developing countries. Its mapping is thus a key element for national food security. Remote sensing is widely used for crop mapping, but it is underperforming for smallholder agriculture due to several constraints like small field size, fragmented landscape, highly variable cropping practices or cloudy conditions. In this study, we developed an original approach combining remote sensing and spatial modelling to improve crop type mapping in complex agricultural landscapes. The spatial dynamics are modelled using Ocelet, a domain-specific language based on interaction graphs. The method combines high spatial resolution satellite imagery (Spot 6/7, to characterize the landscape structure through image segmentation), high revisit frequency time series (Sentinel 2, Landsat 8, to monitor the land dynamic processes), and spatiotemporal rules (STrules, to express the strategies and practices of local farmers). The method includes three steps. First, each crop type is defined by a set of general STrules from which a model-based map of crop distribution probability is obtained. Second, a preliminary crop type map is produced using satellite image processing based on a combined Random Forest (RF) and Object Based Image Analysis (OBIA) classification scheme, after which each geographical object is labelled with the class membership probabilities. Finally, the STrules are applied in the model to identify objects with classes locally incompatible with known farming constraints and strategies. The result is a map of the spatial distribution of crop type mapping errors (omission or commission) that are subsequently corrected through the joint use of spatiotemporal rules and RF class membership probabilities. Combining remote sensing and spatial modelling thus provides a viable way to better characterize and monitor complex agricultural systems.

Remote sensing data, crop modelling, and statistical methods are combined in an original method to overcome current limitations of crop yield estimation. It is then tested for timely estimation of maize grain yields and their year-on-year variability in Burkina Faso. Outputs from the SARRA-O crop model were used as a proxy for observed data for calibration. The final remote sensing-based yield model was constructed on the interaction between aboveground biomass at flowering (AGB-F) and crop water stress (Cstr) over the reproductive and maturation phases. Various vegetation and drought-related indices were derived from different spectral domains and tested. Model performance was evaluated by cross-validation against (a) simulated yields and (b) independent yields from ground surveys aggregated at village level. The results showed that the RF (Random Forest) model outperformed the MLR (Multiple Linear Regression) model for year-on-year yield estimation at the end of the season when compared to simulated yields. Surface soil moisture (SSM) information, as a proxy for soil water available for plant growth, together with information on the temperature of the canopy cover, helped to improve the RF maize yield model, impacting more particularly the estimation of crop water stress. Lastly, two months before harvest the RF model predicted 46% of the observed end-of-season maize grain yield variability. The combined remote sensing, crop model and machine learning method is thus an effective approach for estimating and forecasting inter-annual maize crop yields in environments where field data are scarce, such as in most parts of the African continent. However, more research is needed to better retrieve the spatial variability of yields in order to strengthen current agricultural monitoring systems, and to address societal challenges, such as declining food security.

Mosquitoes are vectors of major pathogen agents worldwide. Population dynamics models are useful tools to understand and predict mosquito abundances in space and time. To be used as forecasting tools over large areas, such models could benefit from integrating remote sensing data that describe the meteorological and environmental conditions driving mosquito population dynamics. The main objective of this study is to assess a process-based modeling framework for mosquito population dynamics using satellite-derived meteorological estimates as input variables. A generic weather-driven model of mosquito population dynamics was applied to Rift Valley fever vector species in northern Senegal, with rainfall, temperature, and humidity as inputs. The model outputs using meteorological data from ground weather station vs satellite-based estimates are compared, using longitudinal mosquito trapping data for validation at local scale in three different ecosystems. Model predictions were consistent with field entomological data on adult abundance, with a better fit between predicted and observed abundances for the Sahelian Ferlo ecosystem, and for the models using in-situ weather data as input. Based on satellite-derived rainfall and temperature data, dynamic maps of three potential Rift Valley fever vector species were then produced at regional scale on a weekly basis. When direct weather measurements are sparse, these resulting maps should be used to support policy-makers in optimizing surveillance and control interventions of Rift Valley fever in Senegal.

This paper introduces dynamic configurational changes for patchy mosaics, parameterizable dilation and/or erosion processes of units in a vectorial landscape. Patchy-based models are rare although this conceptual framework could yield insights to the functioning and dynamics of agricultural and natural landscapes. Compared to common raster-based models, they are more parsimonious and intuitive, but their algorithmic computations are challenging. The aim of this study is to implement polygon dilation throughout Minkowski sum on polygons, along with their associated formal grammar, in the DYPAL modeling platform. This computation is a challenging open-problem and this paper provides a first working approach along with the methodology for a generic implementation framework. As an example, this paper illustrates configurational changes on a complex forest-savannah dynamics in a tropical biodiversity hotspot, New Caledonia. Finally, perspectives for configurational changes in landscape modeling are discussed. (C) 2017 Elsevier B.V. All rights reserved.

Models are increasingly being used to investigate agro-ecosystems dynamics, although processes interacting at different scales remain difficult to consider. When upscaled or downscaled based on aggregation or disaggregation methods, information is generally distorted. This study explores agro-ecosystem modelling using an interaction graph-based modelling approach that explicitly link elements at different scales without up or downscaling. The study area/time frame is the cotton region of West Burkina Faso over the last fifteen years. Field, plot, farm and climate entities are linked in graphs that evolve according to functions computed along different time steps. Three main processes and their interrelations are simulated, occurring at different spatial and temporal scales: crop area expansion, crop rotation and crop production. Three simulation examples are presented to illustrate the analytical possibilities allowed by the approach. These examples test i) the geographical distribution of plots as a means to face climatic risks, ii) the effect of fallowing practice in a spatially constrained cotton dominated landscape and iii) the consequences of reduced access to credit for farmers to buy fertilizers. Model outputs enable quantifying and mapping the respective effects of processes at different scales. Results show that modelling across scales is achievable without resorting to methods of aggregation or disaggregation, which opens new perspectives in multi-scalar analyses of agro-ecosystems that link land production and land use and land cover.

Rural areas of West Burkina Faso have seen notable transformations these last two decades due to high population growth and farming systems evolution. Satellite images acquired frequently and covering large areas are essential for detecting such landscape changes and long term trends. However, these images generally have coarse spatial resolutions and can only provide information about changes in the main vegetation patterns. The factors causing these changes are more difficult to determine, although there are essential for monitoring landscape evolution. We hereby present a method based on multi-scalar modelling of past landscape dynamics crossed with changes in vegetation trends identified from coarse resolution satellite images. The aim of our presentation is to use the model to simulate and illustrate how land cover and land use changes may impact vegetation response by improving the qualification and understanding of the observed trends. The cropping systems dynamics of the study area, the Tuy province of West Burkina Faso, were modelled with the Ocelet Modelling Platform over the last fifteen years through a multi-scalar model. The model was validated at local scale with information derived from high resolution images. At the same time, vegetation trends were analysed using Ordinary Least Square regressions based on MODIS NDVI time series. Simulated cropland change maps were then used to decompose the remote sensing-based trends. This allowed the spatial identification of factors responsible for the vegetation changes. The original approach we proposed here opens new opportunities for the understanding and monitoring of landscape changes using time series of coarse resolution satellite images.

La modelisation et la simulation de dynamiques spatiales, en particulier pour l'etude de l'evolution de paysages ou de problematiques environnementales pose la question de l'integration des differentes formes de representation de l'espace au sein d'un meme modele. Ocelet est une approche de modelisation de dynamiques spatiales basee sur le concept original de graphe d'interaction. Le graphe porte a la fois la structure d'une relation entre entites d'un modele et la semantique decrivant son evolution. Les relations entre entites spatiales sont ici traduites en graphes d'interactions et ce sont ces graphes que l'on fait evoluer lors d'une simulation. Les concepts a la base d'Ocelet peuvent potentiellement manipuler les deux formes de representation spatiale connues, celle aux contours definis (format vecteur) ou la discretisation en grille reguliere (format raster). Le format vecteur est deja integre dans la premiere version d'Ocelet. L'integration du format raster et la combinaison des deux restaient a etudier et a realiser. L'objectif de la these est d'abord etudier les problematiques liees a l'integration des champs continus et leur representation discretisee en pavage regulier, a la fois dans le langage Ocelet et dans les concepts sur lesquels il repose. Il a fallu notamment prendre en compte les aspects dynamiques de cette integration, et etudier les transitions entre donnees geographiques de differentes formes et graphe d'interactions a l'aide de concepts formalises. Il s'est agi ensuite de realiser l'implementation de ces concepts dans la plateforme de modelisation Ocelet, en adaptant a la fois son compilateur et son moteur d'execution. En n, ces nouveaux concepts et outils ont ete mis a l'epreuve dans trois cas d'application tres differents : deux modeles sur l'ile de la Reunion, le premier simulant le ruissellement dans le bassin versant de la Ravine Saint Gilles s'ecoulant vers la Cote Ouest de l'ile, l'autre simulant la diffusion de plantes invasives dans les plaines des hauts a l'interieur du Parc National de La Reunion. Le dernier cas decrit la spatialisation d'un modele de culture et est applique ici pour simuler les rendements de cultures cerealieres sur l'ensemble de l'Afrique de l'Ouest, dans le contexte d'un systeme d'alerte precoce de suivi des cultures a l'echelle regionale.

This paper introduces the vectorial Kappa (κv) that one can utilize to assess congruence between two vectorial mosaics. The vectorial Kappa extends for vectorial mosaics the approach of the so-called Cohen's Kappa index, commonly used to compare raster mosaics. By comparing both approaches, we aim to demonstrate how efficient and convenient a vector-based congruence may be when working on vectorial mosaics.

Résumé. L’étude de l’évolution d’un système complexe spatialisé devient critique dans de nombreux domaines. Par exemple, pour comprendre l’influence de l’homme sur le ruissellement dans les entités spatiales et selon leurs types d’occupation du sol (culture, urbanisation, habitation...) de manière à permettre aux décideurs d’anticiper les politiques d’urbanisme et d’agriculture. Dans cette démonstration, nous nous intéressons plus particulièrement à la modélisation des compétences de l’expert pour simuler les dynamiques spatiales de systèmes paysager complexes.

Modelling land use and cover changes (LUCC) at local and landscape scales simultaneously, in terms of composition and configuration, remains today highly challenging. Agricultural landscapes offer an illustrative context for this purpose. This article presents a modelling platform (DYPAL) able to simulate LUCC at both local (agricultural parcel, farm) and landscape levels by combining LUCC processes with an optimization algorithm. The efficiency of this approach is assessed by comparing it with an approach applying the same LUCC processes without optimization. Simulations have been developed for two representative case studies of temperate intensive agricultural mosaics: (1) neutral landscapes with simple and theoretical rules and (2) observed landscapes with realistic crop successions rules. Results show that this modelling platform improves the simulation of LUCC achieved at fine resolution, although not systematically. Improvements are observed when compared to theoretical farming practices. But, when compared with an observed landscape, it is true for one type of (arable) farms only. Several hypotheses are discussed, such as the fact that farmers possibly do not follow optimized rules. Finally, this study illustrates that the use of several indices is crucial to assess whether a simulated landscape is realistic or not, because it does not rely to the assessment of the predictive power of the model.

Patchy landscapes driven by human decisions and/or natural forces are still a challenge to be understood and modelled. No attempt has been made up to now to describe them by a coherent framework and to formalize landscape changing rules. Overcoming this lacuna was our first objective here, and this was largely based on the notion of Rewriting Systems, also called Formal Grammars. We used complicated scenarios of agricultural dynamics to model landscapes and to write their corresponding driving rule equations. Our second objective was to illustrate the relevance of this landscape language concept for landscape modelling through various grassland managements, with the final aim to assess their respective impacts on biological conservation. For this purpose, we made the assumptions that a higher grassland appearance frequency and higher land cover connectivity are favourable to species conservation. Ecological results revealed that dairy and beef livestock production systems are more favourable to wild species than is hog farming, although in different ways. Methodological results allowed us to efficiently model and formalize these landscape dynamics. This study demonstrates the applicability of the Rewriting System framework to the modelling of agricultural landscapes and, hopefully, to other patchy landscapes. The newly defined grammar is able to explain changes that are neither necessarily local nor Markovian, and opens a way to analytical modelling of landscape dynamics.

In West Africa, environmental conditions that are highly variable in space and time, in particular due to climate, impose heavily constrained choices on agricultural practices. At large scale, this variability is expressed by an annual rainfall distribution that ranges from more than 1200 mm in the Guinean Zone with double rainy season, to almost no rainfall in a single rainy season near the Sahara. This large spatial variability is also expressed at finer scales due to the stormy character of rainfall events. The length of the rainy seasons, the distribution of rainfall events and their intensity within the season all show that the variability is also temporally very large. During the course of the season, it is essential to take into account differences in agricultural practices while assessing potential productions at different spatial scales. The integration of the SARRA-H crop model in the spatial dynamics modelling platform Ocelet offers new opportunities for this assessment. Processes can be modelled at different spatial and temporal scales and multiple data sources, including vector and raster formats, can be managed efficiently. A working prototype has been developed and is being tested during the 2016 crop season at the AGRHYMET Regional Centre in Niamey, in the framework of its Food Security Early Warning System. The objective of this presentation is to show the crop monitoring capabilities of the spatialized version of the SARRA-H crop model that integrates common agricultural practices at the scale of West Africa.