The model environment is the core component of any spatially explicit agent-based model. It provides the spatio-temporal context within which simulated agents obtain information, make decisions, and behave accordingly. In ALMaSS, unlike most comparable systems, the model environment has been designed to serve every species model in the system simultaneously, from ground beetles a few millimetres long to roe deer capable of moving several kilometres per day. Pollinators rely on floral resources, pest-control species use open agricultural land, and hares and skylarks depend on meadows and grasslands. This breadth of ecological perspectives has forced the ALMaSS landscape representation to be sufficiently detailed and flexible to reflect the true complexity of the real world, a requirement that few other modelling systems have attempted to meet.

The newly described methodology, published in the journal Agricultural and Environmental Modelling, addresses a problem that is complementary to the software architecture of ALMaSS itself: how to acquire, integrate, and process the heterogeneous geospatial data needed to generate a high-resolution agricultural landscape map suitable for simulation. The challenge is far from trivial. Input datasets differ in format, coordinate reference system, spatial resolution, thematic scope, and provenance, and their combination inevitably produces artefacts and data gaps that must be resolved before a consistent, complete raster can be produced. The core methodological contribution is a multi-stage data-integration and quality-assurance algorithm for agricultural landscape map generation, including an improved procedure for the systematic removal of sliver artefacts and unclassified gaps, a significant advance on the approach first described for Danish landscapes nearly a decade ago.

The resulting landscape map is a detailed land-use and land-cover raster with a resolution of one square metre, typically covering an area of ten by ten kilometres. Each cell of the map references a homogeneous landscape polygon characterised by its land-use or land-cover type and associated attributes, including soil type, elevation, and farm type. The temporal component allows polygon properties such as vegetation height, biomass, floral resources, and crop management to change on a daily time step, capturing the dynamic character of agricultural landscapes throughout the year. This combination of fine spatial detail and daily temporal dynamics is what allows ALMaSS to represent the specific spatial and temporal heterogeneity that results from the way farmers organise cropping systems.

Building such a map requires drawing on a remarkable range of data sources. National or regional topographic object databases, typically corresponding to map scales of 1:5,000 to 1:25,000, provide the primary source of topography and land-cover information, capturing rivers, roads, railways, buildings, forests, hedgerows, and other natural or semi-natural features. Satellite-based products from the Copernicus Land Monitoring Service offer additional thematic layers at resolutions of ten metres or higher. For the agricultural component, which is particularly important for ALMaSS, the methodology relies on the Integrated Administration and Control System, the mechanism through which the European Union administers Common Agricultural Policy payments. Within this system, the Land Parcel Identification System identifies and stores geographical information on reference parcels, while direct payment registers provide information on the crops grown on those parcels, and agricultural holding registers allow individual parcels to be grouped into farm units.

By combining different components of the IACS database, the methodology subdivides agricultural land into individual fields and farm units and classifies farm units into types, principally by combining information on crops grown and animals reared per farm unit. The farm typology generally aligns with the types of farming recognised in the Farm Sustainability Data Network, the EU’s farm-accounting data collection system, but ALMaSS also distinguishes between farms specialising in field crops and those specialising in potato and beet production, owing to the particularly large number of agrotechnical treatments, including fertilisers and pesticides, applied to these crops. Standard farm types include arable, horticulture, permanent crops such as orchards, grazing livestock, granivores, mixed stock, and hobby farms, with distinctions between conventional and organic practices. Each farm unit is then linked to a crop rotation scheme, generated from the proportions of crops cultivated by farms of that type, so that the changing pattern of crops on any field precisely matches the overall crop distribution for that farm type over one hundred seasons.

The processing pipeline itself is a carefully choreographed sequence of geographic information system operations. Input datasets are harmonised to a common coordinate reference system, clipped to the study area with a one-kilometre buffer to minimise boundary effects, and converted to raster layers on a one-square-metre grid. Information layers are organised into themes, including natural, cultivable, built-up, cultural, water, and transport, and each layer is assigned a three-digit code that determines its spatial priority when layers are overlaid. The thematic layers are then stacked in sequence so that the most ecologically meaningful features appear at the top of the raw landscape map. Because input data from different sources rarely align perfectly, the raw map typically contains unclassified background areas and sliver artefacts, narrow gaps or overlaps between features that reflect data inconsistencies rather than reality. The multi-stage cleaning procedure applies geometry rules, neighbourhood rules, and land-cover rules to resolve these issues, reclassifying background areas as missing agricultural fields, field margins, riparian vegetation, yards, or wasteland, depending on their shape, size, and surroundings.

The methodology has been operationalised across a wide range of European agricultural landscapes. ALMaSS landscape models have been developed or are being developed under various Horizon 2020 projects, including B-GOOD, EcoStack, and PoshBee, the Horizon Europe project PollinERA, and national projects with the Netherlands Environmental Assessment Agency and the German Environmental Agency. A fully operational landscape model exists for regions or countries where the complete set of external input files can be generated and used in simulation runs, and several more are currently being updated. The entire workflow is implemented in three Python scripts, publicly available in a GitLab repository, supporting batch processing across multiple landscape windows and enabling scaling to regional or national extents on mainframe computers or clusters.

The value of these detailed landscape maps extends well beyond the ALMaSS modelling framework. The generated land-use and land-cover rasters can describe structural and farmland heterogeneity through a set of landscape- and class-level metrics, including landscape diversity, landscape shape index, number of fields, mean field size, and field boundary density. Researchers have used these metrics to simulate the population dynamics of the ground beetle Bembidion lampros in landscapes with varying degrees of heterogeneity and different farming systems, finding that beetle populations were larger and distributed over larger areas in more diverse landscapes with a higher proportion of herbaceous semi-natural areas. The maps have also been used to select fieldwork study sites meeting specific landscape conditions, for instance in studies of the ground beetle Poecilus cupreus and the red mason bee Osmia bicornis, where the location of traps and nests was based on detailed analysis of landscape structure and crop characteristics derived from the generated maps.

ALMaSS is regularly employed to assess the impacts of agricultural and environmental policies on biodiversity and ecosystem services across European landscapes. By working with actual landscape structures that capture the full complexity of farming systems, land ownership patterns, and ecological features, the system enables the evaluation of multiple policy instruments simultaneously within the same landscape context. This capability allows policymakers to identify synergies between various measures, such as agri-environmental schemes, biodiversity conservation programmes, and pesticide regulations, while revealing potential conflicts that might otherwise remain hidden. Such integrated assessments are particularly valuable given the increasing complexity of agricultural and environmental policy frameworks in the EU, where multiple directives and regulations must work coherently to achieve sustainable agricultural landscapes.

Nevertheless, the methodology faces important limitations. Access to IACS data, which is crucial for developing ALMaSS landscape models for new areas, can be challenging, as field ownership information is typically provided only upon request and often requires additional agreements with data providers. Satellite-based crop classification products can provide unified crop information over extensive areas, but they often exhibit relatively low thematic resolution and do not facilitate the delineation of farm units. European-wide products such as those from the Copernicus Land Monitoring Service are designed to map land cover rather than land use, a distinction that matters greatly, since natural meadows, managed mown grasslands, and amenity grass differ significantly in the resources they provide but are not differentiated within the Copernicus grassland layer. The reliance on the ArcPy library within the proprietary ESRI ArcGIS environment also affects reproducibility and accessibility, and efforts are underway to develop an open-source implementation using GDAL and OGR that would enable execution on Linux-based high-performance computing systems.

Looking ahead, the researchers envision a Data Linkage and Integration module that would facilitate automated generation of landscape inputs for any selected location within the EU, drawing on the best available national and European datasets. The primary challenge lies in maintaining dynamically updated data flows and developing algorithms for flexible, quality-aware data selection across heterogeneous sources. Another crucial step in future development is incorporating the ability to modify landscape structure during simulation, accounting for land-use and land-cover changes outside arable fields, potentially by integrating ALMaSS with agent-based models of farm cessation, expansion, and diversification. As these capabilities mature, the digital farmland environments that ALMaSS species inhabit will become ever more faithful reflections of the living landscapes they represent, strengthening the evidence base on which Europe’s agricultural future is being decided.

Subject of Research: Development of a spatio-temporal representation of agricultural landscapes as the modelling environment for spatially explicit agent-based models in the Animal Landscape and Man Simulation System ( ALMaSS )

Article Title: Development of a spatio-temporal representation of agricultural landscapes as the modelling environment for spatially explicit agent-based models in the Animal Landscape and Man Simulation System ( ALMaSS )

Article References: Ziółkowska, E., Jaśkowiec, B., Groom, G. B., & Topping, C. J. (2026). Development of a spatio-temporal representation of agricultural landscapes as the modelling environment for spatially explicit agent-based models in the Animal Landscape and Man Simulation System (ALMaSS). Agricultural and Environmental Modelling, 8, Article e167439. https://doi.org/10.3897/aem.8.167439

Image Credits: AI Generated

DOI: 10.3897/aem.8.167439

Keywords: Development, spatio-temporal, representation, agricultural, landscapes, modelling, environment, spatially, explicit, agent-based, models, Animal