Modelling tree biomass using direct and additive methods with point cloud deep learning in a temperate mixed forest

Science of Remote Sensing, Год журнала: 2023, Номер 8, С. 100110 - 100110

Опубликована: Ноя. 18, 2023

Airborne laser scanning (ALS) data has been widely used for total aboveground tree biomass (AGB) modelling, however, there is less research focusing on estimating specific components (wood, branches, bark, and foliage). Knowledge about these essential carbon accounting, understanding forest nutrient cycling, other applications. In this study, we compare additive AGB estimation (sum of estimated components) with direct using deep neural network (DNN) random (RF) models. We utilise two point cloud DNNs: point-based Dynamic Graph Convolutional Neural Network (DGCNN) Octree-based (OCNN). DNN RF models were trained a dataset comprised 2336 sample plots from mixed temperate in New Brunswick, Canada. Results indicate that perform similarly to terms coefficient determination (R2) root-mean square error (RMSE), reduced the mean absolute percentage (MAPE) by 22% average. Compared RF, DNNs provided small improvement performance, OCNN explaining 5% more variation (R2 = 0.76) reducing MAPE 20% Overall, study showcases effectiveness highlights potential enhanced estimation. To further improve recommend larger training datasets, implementing hyperparameter optimization, incorporating additional such as multispectral imagery.

Язык: Английский

---

Accuracy evaluation and effect factor analysis of GEDI aboveground biomass product for temperate forests in the conterminous United States

GIScience & Remote Sensing, Год журнала: 2023, Номер 61(1)

Опубликована: Дек. 12, 2023

The Global Ecosystem Dynamics Investigation (GEDI) is expected to revolutionize the quantification of aboveground carbon at continental scales, through its unprecedented dense vertical observations vegetation structure. As primary task, GEDI recently introduced L4A, 25 m near-global footprint biomass density (AGBD) product. a global mission with significant policy and management applications, it urgent conduct comprehensive evaluation L4A analyze factors affecting product's performance. In this study, accuracy assessed using co-registered airborne Lidar surveys collected during 2018 ~ 2019 corresponding AGBD plots 19 sites National Ecological Observatory Network (NEON). analysis included 11 forest types spanned 17 eco-climatic domains across conterminous United States ensure representativeness comprehensiveness result. interplay nine quantified, including simulated waveform strategy deviation (SWSD) used in canopy characteristics (tree height, crown size, cover), heterogeneity (crown size standard deviation, tree height density), other (forest type topographic slope). Results show that compared NEON observations, generally underestimates (Bias: −31.65 Mg/ha), moderate relative error exhibited 14 (%RMSE ranging from 19% 50%). For half types, threshold lowest requirement products set by GCOS was met or close being met. Broadleaf forests high values had %RMSE (less than 35%), while coniferous low highest (over Among different considered, SWSD contributed most L4A's accuracy, importance 56.63%, manifested indirect impacts characteristics. (32.40%) second after SWSD; also much higher (3.99%). These results indicate limitation only heights as predictors due limited representation horizontal structure complexity within footprint. findings study are step forward appropriate application provide perspectives aid improvement.

Язык: Английский

---

Design and performance of the Climate Change Initiative Biomass global retrieval algorithm

Science of Remote Sensing, Год журнала: 2024, Номер 10, С. 100169 - 100169

Опубликована: Сен. 30, 2024

Язык: Английский

---

Combining Landsat time series and GEDI data for improved characterization of fuel types and canopy metrics in wildfire simulation

Journal of Environmental Management, Год журнала: 2023, Номер 345, С. 118736 - 118736

Опубликована: Авг. 3, 2023

Wildfires in the Chornobyl Exclusion Zone (CEZ) and other radioactively contaminated areas threaten human health well-being with potential to resuspend radionuclides. Wildfire behavior simulation is a necessary tool examine efficiency of fuel treatments CEZ, but it requires systematically updated maps types canopy metrics. The objective this study was demonstrate an effective approach for mapping types, height (CH), cover (CC) territories by radionuclides using Landsat time series (LTS) Global Ecosystem Dynamics Investigation (GEDI) LiDAR observations. We combined LTS GEDI data map metrics used wildfire simulations within CEZ. Our classification model showed adequate overall accuracy (75%) land covers associated types. phenology extracted from reliably distinguished spectrally similar vegetation (such as grasslands croplands) which exhibit different flammability through year. also predicted suite relative heights CC at 30-m pixel level (R2 = 0.23-0.26) nearest neighbor technique. imputed adequately captured dynamics CH CEZ after recent large wildfires occurred 2015, 2020, 2022. Thus, we illustrate processing produce wall-to-wall characteristics that are important simulations. conclude continuous updating crucial ensure relevant fire management landscapes support local decision-making.

Язык: Английский

---

Forest degradation contributes more to carbon loss than forest cover loss in North American boreal forests

International Journal of Applied Earth Observation and Geoinformation, Год журнала: 2024, Номер 128, С. 103729 - 103729

Опубликована: Март 6, 2024

The carbon sinks of North American boreal forests have been threatened by global warming and forest disturbances in recent decades, but knowledge about the balance these years remains unknown. We tracked annual aboveground (AGC) changes from 2016 to 2021 across regions NASA's Arctic Boreal Vulnerability Experiment (ABoVE) core study domain, using Vegetation Optical Depth derived low-frequency passive microwave observations. results showed that a net AGC increase + 28.49 Tg C/yr during period, with total gains 219.34 counteracting losses −190.86 C/yr. Forest degradation (-162.21 C/yr), defined as reduction capacity provide goods services, contributes 5 times more loss than cover (-28.65 complete removal tree cover. This indicates has dominated region.

Язык: Английский

---