Unveiling historical agroecological patterns through artificial intelligence (AI) and Geographic Information Systems (GIS)

Abstract. Tracing the evolution of regional agroecological specifics is crucial for comprehending the successes and setbacks of historical human intervention in land use, significantly impacting the current and future suitability of agricultural land. Despite diligent efforts to standardize diverse data sources and quantitatively reconstruct data from various periods, researchers grapple with ongoing questions about the reliability of historical agroecological data. This underscores the imperative for novel methodologies aimed at enhancing the quality and quantity of available historical agroecological data. Recognizing the pivotal role of these historical sources, this paper unveils the preliminary outcomes of the AgroecoDecipher project—dedicated to tracing geographic land patterns through historical agricultural records and artificial intelligence. The initial phase involves gathering and harmonizing data through the digitization, georeferencing, and storage of historical surveys for each of Portuguese municipality (n = 277). Employing an exploratory methodology grounded in artificial intelligence (AI) and Geographic Information Systems (GIS), the projected solutions aim to extract a comprehensive database from textual records and map files, facilitating their accessibility for geospatial analysis. The overarching results have contributed to the development of open science and collaborative solutions, embedded within enduring tools for agroecological analysis. This includes the establishment of routines based on open-source AI tools for optical character recognition (OCR), coupled with the formulation of guidelines for text parsing. These endeavors not only preserve the historical information contained in these sources but also establish an invaluable resource for researchers and future studies.