AgroTwin: A Pixel-Level Latent-State Dataset for Agricultural Digital Twin Modeling (2020–2025))

Abstract

This dataset provides three-dimensional latent-state representations (Z1, Z2, and Z3) derived from multisource remote sensing observations of 23 agricultural fields at the North Kazakhstan Agricultural Experimental Station (NKAES LLP), Kazakhstan, during the growing seasons (May–September) from 2020 to 2025. The latent representations were generated using a multilayer perceptron autoencoder (MLP-AE-3D) based on five agroecological variables: NDVI, NDRE, NDWI, land surface temperature (LST), and soil moisture (SM). These variables were derived from Sentinel-2, MODIS, and SMAP observations. Included files: AgroTwin_Pixel_Latent.zip: Pixel-level latent-state data representing agroecosystem conditions. NASA_POLYGON_CLIMATE_GLDAS_2020_2025.csv: Climate data associated with the study area for 2020–2025. The dataset supports research on agricultural digital twins, latent-state dynamics, temporal prediction, climate-related stress analysis, and sustainable agroecosystem management. Related publication: A Data-Driven Latent Digital Twin Framework for Sustainable Agroecosystem Management (Sustainability, 2026).

More by Aigul Mimenbayeva

• A Data-Driven Latent Digital Twin Framework for Sustainable Agroecosystem ManagementSustainability · 2026
• SPATIAL AND CLIMATE DATA ANALYSIS FOR MACHINE LEARNING-BASED CROP YIELD PREDICTION IN NORTHERN KAZAKHSTANNEWS OF THE NATIONAL ACADEMY OF SCIENCES OF THE REPUBLIC OF KAZAKHSTAN · 2026
• AgroTwin: A Pixel-Level Agricultural Dataset for Crop Monitoring, Anomaly Detection, and Machine Learning (2020–2025)Zenodo · 2026
• CLOUD-BASED SYSTEM FOR AGRICULTURAL LAND MONITORING USING REMOTE SENSING DATA IN NORTHERN KAZAKHSTANScientific Journal of Astana IT University · 2026