Species diversity and soil quality are deteriorating due to continuous disturbances in ecosystems caused by human interference. However, agroforestry is considered a good approach to minimizing environmental problems. Therefore, the objective of this study was to determine the impacts of participatory agroforestry on restoring soil fertility, farm productivity and biodiversity in the degraded Madhupur Sal forest of Bangladesh. The study purposefully selected 40 common agroforestry programs in Madhupur Sal forest for the collection of soil and plant data from 2020 to 2023. Agroforestry programs have improved soil organic matter, soil carbon, pH, and available N, P and K content to a substantial degree and protected soil degradation, enhancing yield. The soil improvement index represents the potentiality of agroforestry in restoring soil nutrients and carbon in the form of organic matter, which is an important indicator for carbon sequestration and mitigating the impacts of climate change. The resultant cost–benefit and land equivalent ratios were steadily higher, which corroborates the greater productivity and profitability of agroforestry compared to monoculture systems. In contrast, agroforestry restored 31 plant species, opening up opportunities for restoring plant species in the threatened forest ecosystem. Therefore, this study recommended selecting appropriate site-specific species for managing agroforestry and restoring ecosystems.

Crop statistics are crucial for developing a demand-based export and import strategy to ensure a country’s sustainable food security. Remote sensing efficiently generates essential crop statistics, while ground-based supplementary sensor data offers sufficient information for crop delineation. This study explored the multispectral satellite imagery using in-situ ground-based hyperspectral reflectance phenology information as training data to delineate wheat from other competitive winter crops in Northwestern Bangladesh as a case study. Wheat spectral signatures were primarily obtained through a hand-held Spectroradiometer at various phenological stages, aligned with Sentinel-2 data availability. Five vegetation indices (VIs), namely Normalized Difference Vegetation Index (NDVI), Red-edge NDVI (RENDVI), Enhanced Vegetation Index (EVI), Greenness Chromatic Coordinate (GCC) and Soil-Adjusted Vegetation Index (SAVI), were derived from Spectroradiometer-data across six wheat growth stages: seedling, tillering, booting, flowering, grain development, and maturity. Maximum and minimum threshold values for the VIs at those six growth stages were determined from regression analysis of the values collected from Spectroradiometer and Sentinel-2. A rule-based classification technique was then used to categorize Sentinel-2 for wheat crop delineation based on those threshold values. The results revealed that maps based on NDVI, EVI, and SAVI showed overall accuracies of 83.33%, 85.18%, and 81.48%, respectively. These accuracies were found to be statistically acceptable (p < 0.05) outcomes. A positive agreement was observed when comparing the remotely sensed area at the union (4th tier administrative level) with the officially reported data of Bangladesh. This innovative method has the potential to be extended for developing phenology and area delineation for other major crops locally and globally.