Estimación de variables forestales a partir sensores lidar y ópticos e inteligencia artificial

Mihai Tanase; Juan Pablo Martini; Pablo Miranda; Daniel Garcia Garcia; Victoria Wilke; Jaime Diez; Sergio Natal; Daniel San Martin

doi:10.18172/cig.6767

Authors

Mihai Tanase undefined
Juan Pablo Martini undefined
Pablo Miranda undefined
Daniel Garcia Garcia undefined
Victoria Wilke undefined
Jaime Diez undefined
Sergio Natal Predictia Inteligent Data Solutions
Daniel San Martin Predictia Inteligent Data Solutions

DOI:

https://doi.org/10.18172/cig.6767

Keywords:

forest variables, remote sensing, time series, artificial intelligence

Abstract

Accurate and continuous characterization of forest fuel properties is essential for assessing wildfire risk and predicting fire behavior. This study addresses the limitations posed by the scarce temporal availability of lidar data by integrating temporally consistent optical imagery (Landsat) with occasional lidar acquisitions to estimate structural forest variables such as canopy cover fraction (FCC) and canopy height (H) in two regions of peninsular Spain: Madrid and the Basque Country. These variables are widely recognized as proxies for fuel properties characterization, reflecting both the amount and spatial continuity of forest fuels. Machine Learning (ML) models (Random Forest and Extreme Gradient Boosting) were compared with Deep Learning architectures, (DL) including transformer-based models with self-attention mechanisms (NeNeT).

The results show that DL models significantly improve accuracy, with an average reduction in root mean square error (RMSE) of 30% compared to traditional methods. NeNeT, in particular, demonstrated strong performance in capturing complex spatial relationships, improving height estimates in dense forests of the Basque Country (RMSE was reduced from 7.0 to 4.0 m). In contrast, differences were smaller in the more open Mediterranean forests of Madrid, suggesting that less computationally demanding methods like XGB may be suitable in certain contexts. Despite their advantages, DL models present operational limitations, particularly due to high computational demands. For instance, producing historical maps for the peninsular Spain would require up to four years of processing with NeNeT versus seven months with XGB, assuming no parallelization. Moreover, DL models tend to learn spurious patterns related to image acquisition (e.g., number of observations or dates), which can introduce biases if not properly controlled.

In conclusion, combining lidar and optical sensors with advanced artificial intelligence models enables highly accurate estimation of key variables for wildfire management. However, model choice should balance achieved precision with available computational resources, taking into account the ecosystem type and specific application needs.

Downloads

Download data is not yet available.

References

Aragoneses, E., García, M., Salis, M., Ribeiro, L.M., Chuvieco, E. (2023). Classification and mapping of European fuels using a hierarchical, multipurpose fuel classification system. Earth System Science Data, 15, 1287-1315. https://doi.org/10.5194/essd-15-1287-2023 DOI: https://doi.org/10.5194/essd-15-1287-2023

Bolton, D.K., Tompalski, P., Coops, N.C., White, J.C., Wulder, M.A., Hermosilla, T., Queinnec, M., Luther, J.E., van Lier, O.R., Fournier, R.A., Woods, M., Treitz, P.M., van Ewijk, K.Y., Graham, G., Quist, L. (2020). Optimizing Landsat time series length for regional mapping of lidar-derived forest structure. Remote Sensing of Environment, 239, 111645. https://doi.org/10.1016/j.rse.2020.111645 DOI: https://doi.org/10.1016/j.rse.2020.111645

Breiman, L. (2001). Random Forests. Machine Learning, 45, 5-32. https://doi.org/10.1023/A:1010933404324 DOI: https://doi.org/10.1023/A:1010933404324

Cascio, W.E. (2018). Wildland fire smoke and human health. Science of the Total Environment, 624, 586-595. https://doi.org/10.1016/j.scitotenv.2017.12.086 DOI: https://doi.org/10.1016/j.scitotenv.2017.12.086

Caughlin, T.T., Barber, C., Asner, G.P., Glenn, N.F., Bohlman, S.A., Wilson, C.H. (2021). Monitoring tropical forest succession at landscape scales despite uncertainty in Landsat time series. Ecological Applications, 31, e02208. https://doi.org/10.1002/eap.2208 DOI: https://doi.org/10.1002/eap.2208

Chen, S., Woodcock, C.E., Bullock, E.L., Arévalo, P., Torchinava, P., Peng, S., Olofsson, P. (2021). Monitoring temperate forest degradation on Google Earth Engine using Landsat time series analysis. Remote Sensing of Environment, 265, 112648. https://doi.org/10.1016/j.rse.2021.112648 DOI: https://doi.org/10.1016/j.rse.2021.112648

Freund, Y., Schapire, R.E. (1997). A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting. Journal of Computer and System Sciences, 55, 119-139. https://doi.org/10.1006/jcss.1997.1504 DOI: https://doi.org/10.1006/jcss.1997.1504

Gale, M.G., Cary, G.J., Van Dijk, A.I.J.M., Yebra, M. (2021). Forest fire fuel through the lens of remote sensing: Review of approaches, challenges and future directions in the remote sensing of biotic determinants of fire behaviour. Remote Sensing of Environment, 255, 112282. https://doi.org/10.1016/j.rse.2020.112282 DOI: https://doi.org/10.1016/j.rse.2020.112282

Giglio, L., Randerson, J., Van der Werf, G., Kasibhatla, P., Collatz, G., Morton, D., DeFries, R. (2010). Assessing variability and long-term trends in burned area by merging multiple satellite fire products. Biogeosciences, 7, 1171-1186. https://doi.org/10.5194/bg-7-1171-2010 DOI: https://doi.org/10.5194/bg-7-1171-2010

Hudak, A.T., Fekety, P.A., Kane, V.R., Kennedy, R.E., Filippelli, S.K., Falkowski, M.J., Tinkham, W.T., Smith, A.M.S., Crookston, N.L., Domke, G.M., Corrao, M.V., Bright, B.C., Churchill, D.J., Gould, P.J., McGaughey, R.J., Kane, J.T., Dong, J. (2020). A carbon monitoring system for mapping regional, annual aboveground biomass across the northwestern USA. Environmental Research Letters, 15, 095003. http://doi.org/10.1088/1748-9326/ab93f9 DOI: https://doi.org/10.1088/1748-9326/ab93f9

Jerome, H.F. (2001). Greedy function approximation: A gradient boosting machine. The Annals of Statistics, 29, 1189-1232. http://doi.org/10.1214/aos/1013203451 DOI: https://doi.org/10.1214/aos/1013203451

Lucas‐Borja, M.E., Delgado‐Baquerizo, M., Muñoz‐Rojas, M., Plaza‐Álvarez, P.A., Gómez‐Sanchez, M.E., González‐Romero, J., Peña‐Molina, E., Moya, D., de las Heras, J. (2021). Changes in ecosystem properties after post‐fire management strategies in wildfire‐affected Mediterranean forests. Journal of Applied Ecology, 58, 836-846. https://doi.org/10.1111/1365-2664.13819 DOI: https://doi.org/10.1111/1365-2664.13819

Luo, X., Lin, F., Chen, Y., Zhu, S., Xu, Z., Huo, Z., Yu, M., Peng, J. (2019). Coupling logistic model tree and random subspace to predict the landslide susceptibility areas with considering the uncertainty of environmental features. Scientific Reports, 9, 15369. https://doi.org/10.1038/s41598-019-51941-z DOI: https://doi.org/10.1038/s41598-019-51941-z

Matasci, G., Hermosilla, T., Wulder, M.A., White, J.C., Coops, N.C., Hobart, G.W., Bolton, D.K., Tompalski, P., Bater, C.W. (2018). Three decades of forest structural dynamics over Canada's forested ecosystems using Landsat time-series and lidar plots. Remote Sensing of Environment, 216, 697-714. https://doi.org/10.1016/j.rse.2018.07.024 DOI: https://doi.org/10.1016/j.rse.2018.07.024

Merghadi, A., Yunus, A.P., Dou, J., Whiteley, J., ThaiPham, B., Bui, D.T., Avtar, R., Abderrahmane, B. (2020). Machine learning methods for landslide susceptibility studies: A comparative overview of algorithm performance. Earth-Science Reviews, 207, 103225. https://doi.org/10.1016/j.earscirev.2020.103225 DOI: https://doi.org/10.1016/j.earscirev.2020.103225

Miller, C., Urban, D.L. (2000). Connectivity of forest fuels and surface fire regimes. Landscape Ecology, 15, 145-154. https://doi.org/10.1023/A:1008181313360 DOI: https://doi.org/10.1023/A:1008181313360

Plucinski, M.P., Anderson, W.R., Bradstock, R.A., Gill, A.M. (2010). The initiation of fire spread in shrubland fuels recreated in the laboratory. International Journal of Wildland Fire, 19, 512-520. http://dx.doi.org/10.1071/WF09038 DOI: https://doi.org/10.1071/WF09038

Quan, X., Li, Y., He, B., Cary, G.J., Lai, G. (2021). Application of Landsat ETM+ and OLI Data for Foliage Fuel Load Monitoring Using Radiative Transfer Model and Machine Learning Method. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 14, 5100-5110. https://doi.org/10.1109/JSTARS.2021.3062073 DOI: https://doi.org/10.1109/JSTARS.2021.3062073

Sexton, J.O., Bax, T., Siqueira, P., Swenson, J.J., Hensley, S. (2009). A comparison of lidar, radar, and field measurements of canopy height in pine and hardwood forests of southeastern North America. Forest Ecology and Management, 257, 1136-1147. http://dx.doi.org/10.1016/j.foreco.2008.11.022 DOI: https://doi.org/10.1016/j.foreco.2008.11.022

Soenen, S.A., Peddle, D.R., Coburn, C.A. (2005). SCS+C: A modified sun-canopy-sensor topographic correction in forested terrain. IEEE Transactions on Geoscience and Remote Sensing, 43(9), 2148-2159. https://doi.org/10.1109/TGRS.2005.852480 DOI: https://doi.org/10.1109/TGRS.2005.852480

Souza, C.M., Roberts, D.A., Cochrane, M.A. (2005). Combining spectral and spatial information to map canopy damage from selective logging and forest fires. Remote Sensing of Environment, 98, 329-343. https://doi.org/10.1016/j.rse.2005.07.013 DOI: https://doi.org/10.1016/j.rse.2005.07.013

Souza, C.M., Siqueira, J.V., Sales, M.H., Fonseca, A.V., Ribeiro, J.G., Numata, I., Cochrane, M.A., Barber, C.P., Roberts, D.A., Barlow, J. (2013). Ten-Year Landsat Classification of Deforestation and Forest Degradation in the Brazilian Amazon. Remote Sensing, 5(11), 5493-5513. https://doi.org/10.3390/rs5115493 DOI: https://doi.org/10.3390/rs5115493

Tanase, M.A., de la Riva, J., Santoro, M., Pérez-Cabello, F., Kasischke, E. (2011). Sensitivity of SAR data to post-fire forest regrowth in Mediterranean and boreal forests. Remote Sensing of Environment, 115, 2075-2085. https://doi.org/10.1016/j.rse.2011.04.009 DOI: https://doi.org/10.1016/j.rse.2011.04.009

Tanase, M.A., Mihai, M.C., Miguel, S., Cantero, A., Tijerin, J., Ruiz-Benito, P., Domingo, D., Garcia-Martin, A., Aponte, C., Lamelas, M.T. (2024). Long-term annual estimation of forest above ground biomass, canopy cover, and height from airborne and spaceborne sensors synergies in the Iberian Peninsula. Environmental Research, 259, 119432. https://doi.org/10.1016/j.envres.2024.119432 DOI: https://doi.org/10.1016/j.envres.2024.119432

Tanase, M.A., Villard, L., Pitar, D., Apostol, B., Petrila, M., Chivulescu, S., Leca, S., Borlaf-Mena, I., Pascu, I.-S., Dobre, A.-C., Pitar, D., Guiman, G., Lorent, A., Anghelus, C., Ciceu, A., Nedea, G., Stanculeanu, R., Popescu, F., Aponte, C., Badea, O. (2019). Synthetic aperture radar sensitivity to forest changes: A simulations-based study for the Romanian forests. Science of the Total Environment, 689, 1104-1114. https://doi.org/10.1016/j.scitotenv.2019.06.494 DOI: https://doi.org/10.1016/j.scitotenv.2019.06.494

Wilson, J. (2013). Remote Sensing Technologies and Global Markets. In, Market Reserch. Instrumentation And Sensors BBC Research LLC.

Young, N.E., Anderson, R.S., Chignell, S.M., Vorster, A.G., Lawrence, R., Evangelista, P.H. (2017). A survival guide to Landsat preprocessing. Ecology, 98, 920-932. https://doi.org/10.1002/ecy.1730 DOI: https://doi.org/10.1002/ecy.1730

Zhang, L., Shao, Z., Liu, J., Cheng, Q. (2019). Deep Learning Based Retrieval of Forest Aboveground Biomass from Combined LiDAR and Landsat 8 Data. Remote Sensing, 11, 1459. https://doi.org/10.3390/rs11121459 DOI: https://doi.org/10.3390/rs11121459