Dynamic Monitoring Method for Aboveground Biomass of Maize Based on Multi-source Data

The aboveground biomass of maize is regarded as a crucial indicator for monitoring maize growth. However, it’s challenging to meet the practical needs for dynamic remote sensing of aboveground biomass during early growth stages. Field experiments are conducted at National Field Scientific Observation and Research Station of Agricultural Meteorology at Gucheng, Hebei Province. Based on data from maize staged sowing experiments conducted from 2020 to 2024, observations from 2020 to 2021 are utilized for model construction, while data from 2022 to 2024 are employed for model application. 3 modeling approaches, including multiple linear regression, partial least squares regression, and random forest are used to develop models for aboveground biomass at different growth stages, and models are compared by adopting a strategy combining vegetation indices with meteorological factors. Results show that spectral bands most strongly correlated with aboveground biomass mainly concentrate in 720-1300 nm range, with correlation coefficients ranging from 0.50 to 0.86 (passing the test of 0.05 level). When vegetation indices are combined with meteorological factors, such as effective accumulated temperature and total radiation, the performance of 3 models significantly improves compared to using vegetation indices, respectively. In various stages of individual development, the random forest method demonstrates the best performance in biomass simulation. In the test set, the corresponding coefficient of determination values are 0.59, 0.85, 0.85, and 0.81, respectively. Moreover, determination coefficients of aboveground biomass model applications at all growth stages are above 0.40. The study indicates that integrating vegetation indices with meteorological factors can significantly enhance the accuracy of aboveground biomass estimation models at different growth stages. The excellent performance of the random forest method highlights its potential for accurately estimating biomass dynamics. This integrated approach effectively addresses the challenge of monitoring maize during its early growth stages and provides a more accurate method for estimating aboveground biomass. In summary, combining vegetation indices with meteorological factors for aboveground biomass modeling is an effective strategy to improve the accuracy of maize growth monitoring. Research findings provide technical support for precise crop growth monitoring and refined agricultural management.