Quantifying actual evapotranspiration (ETa) in semiarid Mediterranean-climate shrublands is limited by meteorological data scarcity, thus restricting the calculation of vegetation coefficients (Kc). Here, we integrated micrometeorological data, soil moisture observations and satellite-derived spectral indices (red–near-infrared and red-edge bands) to calculate leaf Area indices (LAIs) using three empirical models within the FAO-56 framework, which were subsequently validated using measurements from an eddy covariance (EC) tower within a chaparral ecosystem in the Guadalupe Valley, Baja California, Mexico, from June 2015 to September 2016. The spectral indices captured phenological seasonality and showed strong associations with measured ETa (ETEC) (R = 0.751–0.795). A normalised difference red-edge (NDRE) index, based on RapidEye red-edge band (690–730 nm), achieved the highest predictive accuracy. The structural overestimation in FAO-56 parameterisation was corrected via calibration of an ecosystem-specific scaling factor for the basal crop coefficient (β = 0.33, 95% CI: 0.238–0.425). The estimated ETa (ETK) ranged from 0.01 to 0.90 mm day−1, closely aligning with the ETEC of 0.11 to 1.03 mm day−1. Leave-one-out cross-validation (LOOCV, n = 17) yielded a Nash–Sutcliffe efficiency (NSE) of 0.66 and a percent bias (PBIAS) of +6.45%. However, including soil evaporation (Ke) degraded model performance substantially (NSE < −14 for all index-LAI combinations), indicating that a
transpiration-only formulation is more appropriate for this chaparral, consistent with its relatively closed canopy and limited bare-soil fraction. Overall, integrating high-resolution remote sensing, specifically NDRE, with climate and soil data enables a robust characterisation of water balances in seasonal dryland ecosystems.
