Florian Hartig

← Publications

Using synthetic data to evaluate the benefits of large field plots for forest biomass estimation with LiDAR

Fabian Ewald Fassnacht, Hooman Latifi & Florian Hartig

Remote Sensing of Environment, 213, 115–128 (2018)
Cite this
@article{fassnacht2018using,
  author = {Fabian Ewald Fassnacht and Hooman Latifi and Florian Hartig},
  title = {Using synthetic data to evaluate the benefits of large field plots for forest biomass estimation with LiDAR},
  journal = {Remote Sensing of Environment},
  volume = {213},
  pages = {115--128},
  year = {2018},
  doi = {10.1016/j.rse.2018.05.007},
}

DOI: 10.1016/j.rse.2018.05.007
Citation statistics: 61 (Google Scholar) · 45 (OpenAlex), as of 29 October 2026

Google Scholar citations per year, as of 20 October 2026 (yearly counts are fetched separately from the total)

Funding

This study was supported through funding by Fulbright visiting scholar program (F.E.F., research stay).

View article (DOI)

Abstract

With the maturation of methods for estimating aboveground forest biomass by remote sensing, researchers increasingly need test data, particularly ground reference data, that are large enough to fine-tune existing approaches and test their robustness under diverse conditions. In this context, realistic synthetic datasets present an interesting alternative to costly and limited field data. Here, we present a new approach to simulate realistic canopy height and cover type data by combining an individual-tree forest simulator with real LiDAR point clouds of individual trees. We demonstrate the utility of our approach by re-examining the influence of field plot size on the predictive power of remote-sensing models for biomass estimation. Our approach with a complete (wall-to-wall) field reference dataset and matching synthetic remote sensing data allowed us to not only perform internal cross-validations with field plots that were used to fit the model (as in studies with real data), but to also consider the quality of model predictions to a standardized spatial grid or to the entire region. Our results confirm earlier reports of smaller predictive errors with increased field plot sizes under internal model validation (RMSE of 125 t/ha at 10 m field plot size and RMSE of 40 t/ha for 50 m field plots). However, we show that this is mainly an artifact of comparing the models with the same data they were fit, thus with validation data of different scales. When validating on a grid with standardized scale, smaller field plots performed almost equally well as larger field plots (small RMSE decrease between 4 t/ha and 7 t/ha when going from 10 m to 50 m plots), and even outperformed them if we assumed that increasing the plot size means that fewer field plots can be obtained (small RMSE increase between 9 t/ha and 12 t/ha when going from 10 m to 50 m plots). We conclude that synthetic remote sensing datasets are a useful tool for method testing. The suggested approach may be used to reexamine our current methodological understanding, which is often based on tests with real data of very limited sizes, as well as to optimize workflows, model choices and data collection. A wider use of synthetic data could be instrumental in improving remote sensing methodology.

What the paper shows and why it matters

(AI-generated)

Maps of forest biomass (the aboveground mass of trees per hectare) are often made from airborne laser scanning (LiDAR), with a statistical model calibrated on field plots measured on the ground. Such plots are costly: real studies often have fewer than 100, and ordinary forest inventories were not designed for this purpose. Advice on field campaigns therefore rests on small datasets in which the true biomass is never fully known. Simulating the laser signal could help, but ray-tracing simulators are computationally expensive and have mostly used simple crown shapes. The gap the paper addresses is that large, realistic test datasets with a complete, known reference are missing.

The paper shows that the reported advantage of larger field plots for LiDAR-based forest biomass models largely disappears when all models are tested on the same 50 m grid, and that synthetic data make such a re-test possible. The study places real LiDAR point clouds of single beech and pine trees at the tree positions produced by an individual-tree forest growth simulator, which gives a complete biomass reference for the whole forest. When each model is validated on plots of its own size the prediction error falls from about 125 t/ha for 10 m plots to about 40 t/ha for 50 m plots. When every model is instead evaluated on the same 50 m grid, the gain from larger plots shrinks to 4–7 t/ha. For a fixed total sampled area, larger plots mean fewer plots; then small plots do slightly better (by 9–12 t/ha), chiefly because they give more plots. The study attributes the earlier benefit mainly to unequal validation, presumably because small plots are more variable and so harder to predict.

In the simulated setting (two tree species, no positional mismatch between field and laser data), small to medium plots can be as good as or better than large ones at a fixed sampled area (a stand-in for a fixed field budget), contrary to earlier work favouring large plots, including a meta-analysis of 48 LiDAR-based studies (Zolkos et al. 2013). The study notes that real forests may differ and that plot shape, model type and sensor settings remain to be tested. It also argues that synthetic datasets with a complete reference help to re-examine methodological beliefs, and it provides the workflow as R scripts.

References

  • Zolkos et al. (2013). A meta-analysis of terrestrial aboveground biomass estimation using lidar remote sensing. Remote Sensing of Environment 128, 289–298. doi:10.1016/j.rse.2012.10.017