Optimize the Estimation of Maize Height Using Sentinel-1: A Case Study in Umbria, Italy

Height is an important parameter to evaluate the crops growing state, being strongly correlated to plant other biophysical features. An accurate of this variable can allow farmers to carry out interventions aimed at optimizing crop development, which can be translated into an economic profit at the end of the cycle. The aim of the research is to explore key factors affecting the accuracy of remote sensing models in determining summer maize (Zea mays L.) height using Sentinel-1 SAR data. The study was conducted in 2023 on four fields in Umbria, Italy. The chosen approach is based on the construction of eight random forest (RF) models using different combinations of predictor variables, number of satellite images and speckle filters. In-situ maize height data and corresponding satellite images were acquired between 14th July and 4th September 2023. The results show that models constructed using four pre-full canopy coverage satellite image data (~150 cm) exhibited results with error ranging from 33.9 to 39.2 cm, while those built using all images yielded an error of 75.4 to 89.6 cm, thus demonstrating that maize height could be monitored effectively until to the point of complete canopy cover using SAR data. Future investigations are necessary could focus on refining model parameters further and exploring additional factors to enhance the robustness of the models.