Estimating Leaf Area Index of Hyrcanian Forests Using Landsat-8 and Sentinel-2 Data

Abstract The Leaf Area Index (LAI) is a key ecological variable for monitoring land surface processes and plays an important role in understanding climate-sensitive forest ecosystems such as the Hyrcanian forests. This study aimed to estimate LAI using Linear Mixed-Effects (LME), M5 Model Tree (M5), Random Forest (RF), and Multiple Linear Regression (MLR) algorithms based on Landsat-8 and Sentinel-2 satellite data in the hardwood Hyrcanian forests located in Golestan Province, northern Iran. A total of 230 plots were established across five forest sites using a 100 × 100 m systematic grid. Specific Leaf Area (SLA) was measured for each tree species at each site, and LAI was subsequently estimated from these measurements. Results indicated that mean LAI values across the five sites ranged from 5.1 to 8.9. Among the algorithms, LME yielded the most accurate LAI estimates when applied to Landsat-8 data (RMSE = 22.32%, R 2 = 0.55). Also, for Sentinel-2 data, among the algorithms, LME yielded the most accurate LAI estimates (RMSE = 19.09%, R 2 = 0.60). Overall, Sentinel-2 improved model performance by reducing RMSE% by 1.29–8.04% and increasing R 2 by 0.06–0.18 compared to Landsat-8. Consequently, we conclude that the LME algorithm using Sentinel-2 data is highly suitable for accurately estimating LAI in the broad-leaved Hyrcanian forests.