A HIGH-PRECISION EXPLICIT FOREST CARBON STOCK MODEL BASED ON REMOTE SENSING

Na Zhu, Bisheng Yang, Wenyin Gong, Shen Ying, Wei Dai, Zhen Dong2023 · DOI

Abstract

The current status as well as the potential absorption capability of the forest carbon sinks in terrestrial ecosystems are in urgent need of further studies. The ground observation-based methods are labor-intensive, and the resulting statistics from samples are difficult to evaluate, while inversion methods based on remote sensing data lack theoretical explanation and universality. This paper proposes a pixel-level, multi-scale, high-precision Explicit Forest carbon stock Model (EFM) that is universal and adaptive. First, four key variables were used in the construction of the EFM: remote sensing image resolution, forest canopy density, terrain slope, and forest height; Second, simulated forest scenes were generated based on the growth characteristics of individual trees, and EFM parameters were solved and analyzed by these simulated pixels; Third, the EFM was tested at various scales and forest saturation levels to verify its accuracy, robustness, and applicability. Simulation and real-life experiments show that the correlation coefficient is greater than 0.98 and the relative error is about 20%. The EFM solves the problem that existing methods lack theoretical interpretation and universal applicability, thus can be used to map forest carbon stocks at high resolution and large scale and even monitor forest carbon dynamics at a global scale.

Introduction

Accurately measuring forest carbon storage is vital for climate change research. However, current methodologies face significant challenges. Ground-based surveys, while accurate, are labor-intensive and difficult to scale globally. Conversely, remote sensing methods often rely on "black-box" machine learning or deep learning models that lack theoretical transparency and universality. These models frequently struggle with "saturation" in dense forests and fail when applied across different spatial resolutions.

To address these issues, this research introduces the Explicit Forest carbon stock Model (EFM). The EFM is designed to be both theoretically interpretable and adaptive across varying spatial scales, providing a robust framework for high-precision carbon stock estimation.

Methodology

The EFM moves away from forced regressions, instead utilizing a power function based on the biological growth characteristics of trees. The model incorporates four primary variables:

  1. Resolution ($P$): The size of the image pixel.
  2. Forest Canopy Density ($F$): The proportion of ground covered by canopy.
  3. Terrain Slope ($\theta$): Accounting for surface area variations on sloped terrain.
  4. Forest Height ($H$): The vertical structure of the stand.

The resulting formula is expressed as: $C = a \cdot (P^2 F / \cos \theta)^b \cdot H^c$.

To determine the parameters ($a, b, c$), the researchers generated simulated forest scenes. These simulations were grounded in known growth equations for specific species, such as Chinese fir, pine, and eucalyptus, relating height ($H$) to diameter ($D$).

Growth characteristics (Height vs. Diameter) for Fir, Pine, and Eucalyptus species used to calibrate the EFM.

By "growing" thousands of virtual trees and measuring them at different pixel resolutions (from 10m to 100m), the model's parameters were solved using least squares adjustment. This approach bypasses the "data hunger" of traditional models by leveraging theoretical biological growth patterns.

Results and Discussion

The EFM demonstrated exceptional performance in both simulated and real-world environments. In simulation experiments across 91 different spatial scales, the model achieved a correlation coefficient ($R^2$) of 1.000 for training data and remained above 0.99 for testing.

Regression analysis showing the high correlation between true and fitted carbon stock values across multiple spatial scales (10m to 50m).

Key findings include:

  • Scale Invariance: By explicitly including pixel resolution ($P$) in the formula, the EFM maintains accuracy across different satellite products without retraining.
  • Saturation Resistance: Unlike traditional indices that saturate in dense forests, the EFM remains sensitive by incorporating height ($H$) and canopy density ($F$) as distinct physical variables.
  • Real-World Validation: Testing at two mountainous sites in Guangxi, China, yielded a relative error of 17.6% to 22.3%. While higher than simulations, this represents high precision for large-scale mapping in complex terrain.

Conclusion

The Explicit Forest carbon stock Model (EFM) provides a universal, high-precision solution for forest carbon monitoring. By grounding the model in physical and biological variables, it overcomes the limitations of "black-box" approaches. Future work will focus on expanding the library of EFM parameters for diverse tree species and regions, enabling global-scale, high-resolution carbon dynamics monitoring.

Summary updated

Abstract

The current status as well as the potential absorption capability of the forest carbon sinks in terrestrial ecosystems are in urgent need of further studies. The ground observation-based methods are labor-intensive, and the resulting statistics from samples are difficult to evaluate, while inversion methods based on remote sensing data lack theoretical explanation and universality. This paper proposes a pixel-level, multi-scale, high-precision Explicit Forest carbon stock Model (EFM) that is universal and adaptive. First, four key variables were used in the construction of the EFM: remote sensing image resolution, forest canopy density, terrain slope, and forest height; Second, simulated forest scene were generated based on the growth characteristics of individual trees, and EFM parameters were solved and analyzed by these simulated pixels; Third, the EFM was tested at various scales and forest saturation levels to verify its accuracy, robustness, and applicability, the simulation and real-life experiments show that the correlation coefficient is greater than 0.98 and the relative error is about 20%. The EFM solves the problem that the existing methods are lack in theoretical interpretation and universal applicability, thus can be used to map forest carbon stocks at high resolution and large scale and even monitor forest carbon dynamics at global scale.
BibTeX
@inproceedings{zhu2023high,
  author = {Na Zhu and Bisheng Yang and Wenyin Gong and Shen Ying and Wei Dai and Zhen Dong},
  title = {{A HIGH-PRECISION EXPLICIT FOREST CARBON STOCK MODEL BASED ON REMOTE SENSING}},
  booktitle = {˜The},
  year = {2023},
  doi = {10.5194/isprs-archives-xlviii-1-w2-2023-1831-2023},
}
Figure