Beyond tree cover: Characterizing southern China's forests using deep learning

Abstract

Mapping forests with satellite images at local to global scale is done on a routine basis, but to go beyond the mapping of forest cover and towards characterizing forests according to their types, species and use, requires a dense time‐series of images. This knowledge is important, because ecological and economic values differ between forests. A new generation of low cost very high spatial resolution satellite images and the advent of deep learning enables improved abilities for distinguishing objects based on their structure, which could potentially also be applied to map different forest classes related to type, species and use. Here we use GF‐1 images at 2 m resolution and map six forest classes including different planted species for the karst region in southwest China, covering 806,900 km2. We compare the results with field data and show that accuracies range between 78% and 90%. We show a dominance of plantations (15%) and secondary forests (70%), and only remnants of natural forests (6%). The possibility to map forest classes based on their crown structure derived from low cost very high‐resolution satellite imagery paves the road towards sustainable forest management and restoration activities, supporting the creation of connected habitats, increasing biodiversity and improved carbon storage. No temporal information is needed for our approach, which saves costs and leads to rapid results that can be updated at a high temporal frequency.

Estimation of forest above‐ground biomass in Guangxi, China, by integrating forest age and stack learning

“…Forest‐type data refers to the vegetation distribution results obtained by Li using a deep learning model (Li et al, 2022). The data divides Guangxi's forests into six types of cover, including artificial eucalyptus forests, artificial Pine and Cedar, old growth forests, intact forests, secondary forests and shrublands.…”

“…Forest-type data refers to the vegetation distribution results obtained by Li using a deep learning model (Li et al, 2022) Random forest is an integrated Bagging model with decision trees as the base model (Breiman, 2001), which is straightforward, simple to implement, computationally efficient and has strong performance when dealing with the classification and prediction of high-dimensional data (Breiman, 2001). The input feature variables are simultaneously given an important measure throughout the fitting process, and the bigger the value of the measurement, the more significant the input feature variables are.…”

A Social‐Ecological Framework to Enhance Sustainable Reforestation Under Geological Constraints

“…(Tong et al, 2020). High spatial resolution satellite images and machine learning accurately identify complex forest types under geological constraints (Li et al, 2023). 4.…”

“…(b) Remote sensing and machine learning: Time‐series of remote sensing images spanning 40 years unveil trends in reforestation (Tong et al., 2020). High spatial resolution satellite images and machine learning accurately identify complex forest types under geological constraints (Li et al., 2023). Ecological space optimization and sustainable reforestation in the next 100 years: (a) Continuous Evolution Analysis: Integrating human disturbances and forest evolution across history and recent decades provides a comprehensive understanding of the continuous evolution of “forest‐deforestation‐reforestation.” (b) Optimization strategies: Clarifying and optimizing ecological space involves reforestation, natural regeneration, grass plantation for herbivorous animal husbandry development, and nature reserves.…”

Differentiation characteristics of karst vegetation resilience and its response to climate and ecological restoration projects

“…Among them, south China karst is dominated by a low soil thickness (usually <10 cm); this, coupled with strong human development activities, severely damages the stability of forests, intensifies soil erosion, and leads to rocky desertification (Qiu et al, 2021). To this end, a series of ecological restoration projects, such as the Karst Rocky Desertification Comprehensive Control Project and Grain to Green program, has been implemented in this region in recent decades, and these projects have played an active role in improving regional vegetation coverage (Guo et al, 2019; Li et al, 2022), increasing carbon storage (Liu et al, 2015; Zhang et al, 2021), reducing soil erosion (Deng et al, 2012; Lan et al, 2021), enhancing ecosystem health (Liao et al, 2018), etc. That is, the vegetation productivity in south China karst has significantly increased, and the ecological environment has been significantly improved (Tong et al, 2018; Yue et al, 2020; Zhang et al, 2021).…”