Design of Forestry Carbon Sink Measurement and Prediction System Based on Convolutional Neural Network

To release oxygen and sequester carbon, forests are essential. For Chinese cities to attain carbon neutrality and carbon peak, a precise evaluation of forest carbon storage is important. This work used a prediction system based on convolutional neural networks in the design of a forestry carbon sink measurement and prediction system. Using eastern China's 2020 forest field inventory data as an experimental sample, we investigated the capabilities of deep-learning algorithms such as Convolutional Neural Network (CNN). The findings demonstrated that the main factors influencing the estimation of forest carbon density were coherence from Sentinel-l, backscatter and textural features from ALOS-2, and vegetation indices from Sentinel-2. It was also discovered that the CNN model outperformed conventional models. Combining the optical and radar data effectively confirmed the improvements through the results of the forest carbon-density estimation. Deep learning provides a better chance of correctly calculating forest carbon density from multisource remote-sensing data than conventional regression techniques.