Farm practice typologies as a strategy for management-relevant land use and land cover mapping in the Great Lakes Region

Abstract The Great Lakes region (GLR) is dominated by agricultural land use, characterized by diverse land use and land cover (LULC), land management practices (LMPs), and, in turn, diverse land processes. Conventional LULC classification systems often rely on coarse categorizations, summarized into spatial units that are either unmeaningful to management or lack the fidelity to capture human-driven heterogeneity relevant to environmental outcomes. Furthermore, most LULC systems are temporal snapshots, agnostic to the spatial dependencies of temporal land-use transitions and fail to provide an inventory of the diversity of land use and management within regions. Hence, we developed a multi-year integrated typology framework by combining LULC class transitions (2008–2023) with trends in LMPs, to characterize these differences in the landscape, using inductively developed, threshold-based classification rules at the farm-parcel level.​ Our multi-year integrated typology combined LULC and LMP classes to yield a total of 135 unique land types for the GLR, capturing sufficient variation in agricultural management and landscape heterogeneity that is not reflected in conventional LULC products. Staple Crop Rotation – Tile Drained (StRot-TD) and Staple Crop Rotation – Conventional (StRot-Con) emerged as the dominant multi-year integrated typologies across the GLR. Across three threshold scenarios in sensitivity analysis, 84.27%, 99.01% and 83.43% of parcels retained LULC transition, LMPs trend and multi-year integrated typology classes respectively. The diversity of annual integrated typology increased over time due to the complexity of land use and management practices, even though it decreased for conventional LULC. To demonstrate the practicality of our approach, we used linear mixed-effects models, which compare the performance of our annual integrated typology relative to conventional LULC classifications in predicting water-quality parameters (phosphorus, nitrite, and Kjeldahl nitrogen) in streams draining agricultural watersheds. The annual integrated typology models achieved lower RMSE and higher R 2 than conventional LULC models in the validation for phosphorus, nitrite, and Kjeldahl nitrogen, supporting the predictive value of combining the management attributes in the case study. The framework offers a flexible, interpretable tool for landscape characterization, agricultural planning, conservation implementation, and informed decision-making, and is replicable across regions.