Temporally Integrated MODIS NDVI and Simulated Cumulative Biomass in Mediterranean Grasslands

Mediterranean grasslands exhibit strong seasonality, complicating comparisons between satellite-derived vegetation greenness and cumulative biomass production. This study examined how temporal integration and segmentation influence the association between the MODIS Normalized Difference Vegetation Index (NDVI) and cumulative biomass production simulated by the SIMPAST ecohydrological model. MODIS NDVI time series (MOD09Q1; eight-day composites; 250 m) spanning 2000–2025 were analysed at three grassland sites in central Spain. Non-accumulated NDVI and two cumulative metrics, based on simple and conditional accumulation, were compared with simulated production under conventional seasons and five fixed eco-phenological periods. Simulations used six original meteorological forcing trajectories and an additional forcing dataset based on observed weather. The temporally matched reference relationship between non-accumulated NDVI and simulated eight-day biomass production yielded mean R2 values of 0.188, 0.069, and 0.073 at Piñuecar, Colmenar Viejo, and Tielmes, respectively. Pooled relationships between cumulative NDVI and cumulative simulated biomass were substantially stronger (R2 = 0.714–0.746 for simple accumulation and 0.600–0.723 for conditional accumulation), although these higher values partly reflected their shared within-year temporal progression. Observed meteorological forcing showed the same general pattern, with interval-level reference R2 values of 0.122–0.263 and substantially stronger pooled relationships for the cumulative metrics. However, first differencing reduced mean R2 for simple accumulation to 0.180 at Piñuecar, 0.004 at Colmenar Viejo, and 0.097 at Tielmes, indicating that the strong cumulative associations depended partly on shared within-year progression. Conventional seasons generally yielded the strongest associations in autumn or spring, whereas eco-phenological segmentation generally concentrated them in winter-to-early spring growth. Neither eco-phenological segmentation nor conditional accumulation consistently increased R2 relative to its alternative. These findings distinguish correspondence between cumulative seasonal trajectories from associations at individual observation intervals. Temporal segmentation provides a complementary interpretative framework, but high cumulative R2 values do not establish predictive performance. Operational applications require independent field evaluation of whether these indicators capture biomass production variability beyond shared seasonal progression.