Key points are not available for this paper at this time.
Evapotranspiration (ETo), vital for agricultural and environmental management, faces challenges from climate change and spatial variability. Accurate Land Use-specific ETo estimates are essential for sectors like agriculture, forestry, and water management. Leveraging remote sensing technology, particularly optical remote sensing, holds promise in overcoming limitations posed by scarce weather station data and cloud cover issues. The study encompasses a wide array of meteorological parameters, including Solar Radiation (SR), Temperature (T), Relative Humidity (RH), Wind Speed (WS), and Rainfall (RF), gathered from the archive of Public Works Department archives for the period 2016-2017. Employing the FAO Penman-Monteith method, we calculated reference ETo, representing ETo under standard conditions. This involved intricate steps, such as determining mean T, vapor pressure, the slope of the vapor pressure curve, psychrometric constant, net radiation, and, ultimately, ETo. To enhance our understanding, we employed Partial Least Squares Regression (PLSR) to model the relationship between predictor variables (VV and VH Polarized sigma naught values from Sentinel-1A) and ETo. We generated equations for both monthly mean datasets and overall study period mean, offering insights into short-term fluctuations and long-term trends. Comparative analyses across land cover types unveiled intriguing patterns. Urban transportation areas exhibited stability, while deciduous forests and wetlands showcased temporal variations. In the ETo comparative analysis, each land cover category exhibited distinctive patterns, providing valuable insights into the dynamics of ETo. Among the land cover parameters, ETo was significantly impacted by relative humidity (RH) (70.80% to 89.89%), and temperature (T). Urban vegetated areas had stable T values (29.37C), while forests showed dynamic variations in T (24.24C to 28.94C). The VH polarization captured a diverse range of climatic influences, resulting in a broader range of dynamic ETo values (7.38 to 10.76 mm/day) compared to the VV polarization (6.74 to 9.34 mm/day). The performance of the VH sensor varied; moderate accuracy was observed in October 2016 (R 2 = 0.50) with slight underestimation (Bias = -0.08), whereas exceptional accuracy was seen in December 2017 (R 2 = 1.00) with positive bias (0.57) and excellent agreement (KGE = 0.92). The VV sensors in October 2016 had a firm fit (R 2 = 0.55), moderate underestimation (Bias = -0.87), and December 2017 showed a good fit (R 2 = 0.57), slight overestimation (Bias = 0.44), and good agreement (KGE = 0.44). Thus, integrating machine learning and satellite imagery improves ETo accuracy for real-time monitoring in adaptive management amid climate change, showcasing sensor-specific variations. For precise estimation of ETo, future research should integrate multi-source satellite data and machine learning, which is crucial for adaptive environmental management.
Ramalingam et al. (Sat,) studied this question.