Remote Sensing-Based Estimation of Sorghum LAI Using NDVI from Sentinel-2 and Regression Analysis in GEE

Abstract

Abstract The rapid advancement of digital technologies in agriculture has transformed crop monitoring, particularly through the use of remote sensing and Geographic Information Systems (GIS). These tools have proven indispensable in assessing crop growth dynamics, especially in resource-constrained, semi-arid regions. Among climate-resilient crops, sorghum plays a pivotal role in ensuring food and nutritional security under erratic rainfall and limited irrigation conditions. Monitoring its growth using satellite-derived vegetation indices enables real-time, large-scale assessments that are critical for informed decision-making. This study employed the Google Earth Engine (GEE) platform for the retrieval and processing of Sentinel-2 NDVI data to evaluate its effectiveness in estimating Leaf Area Index (LAI), a key biophysical parameter closely linked to crop vigor and productivity. The primary objective was to determine the most appropriate regression model to establish the relationship between NDVI and LAI. A total of 160 field-observed LAI measurements were collected across two districts of Maharashtra, India—Solapur and Ahmednagar—representing varied agro-ecological conditions. Regression analysis revealed that second-order polynomial models outperformed linear, logarithmic, exponential, and power models, with higher R² values (> 0.40) and lower RMSE (0.83–0.89) in district-level analysis. The combined dataset showed moderate performance (R² >0.25, RMSE = 1.20), reflecting the influence of spatial variability. NDVI-based crop area classification showed high accuracy, with Kappa coefficients exceeding 0.70, and sorghum areas were estimated at 2,58,925 ha in Solapur and 1,48,475 ha in Ahmednagar, aligning within 5% of official government statistics. These results highlight the value of integrating NDVI and LAI through polynomial regression models for accurate, real-time crop monitoring, supporting climate-smart agriculture, precision farming, and policy-level planning in semi-arid regions.

Plant phenotyping relevance

Sentinel-2 NDVIとGEEを用いてソルガムのLAIという植物形質を推定し、回帰モデルを比較・検証しているため、形質取得手法が中心である。

Code and data availability

The preprint describes 160 field LAI measurements, GPS ground-truth points, Sentinel-2 imagery processed in GEE, and regression models, but contains no data availability statement, no public dataset deposit, and no author code/scripts/workflow URL. All URLs in the text are references to cited prior work, not paper-phen

No evidence-backed public reproduction asset is currently recorded.