Estimate Canopy Cover Research Articles

Characterizing canopy cover with ICESat-2: A case study of southern forests in Texas and Alabama, USA

The acquisition of elevation measurements from NASA's Ice, Cloud, and land Elevation Satellite-2 (ICESat-2) and availability of airborne lidar data over the US present an exceptional opportunity to understand vegetation structural estimates provided by ICESat-2. Although a critical forest biophysical attribute, canopy cover is not yet reported from ICESat-2. Thus, there is a need to better understand the application of ICESat-2 for providing canopy cover information. The overall goal of this study was to investigate methods to derive canopy cover and characterize the predictive capability, with ICESat-2. Given availability of dedicated vegetation products (ATL08) and custom noise filtering algorithms developed for ICESat-2 photon data, multiple datasets were evaluated in this study. With focus on two study sites, the Sam Houston National Forest (SHNF) in south-east Texas and the Solon Dixon Forestry Education Center (SDFEC) in southern Alabama, specific objectives were to: (1) Evaluate equations for estimating canopy cover using custom-processed geolocated photon data, ATL03, and ICESat-2's vegetation product data, ATL08, (2) Compare ICESat-2-derived estimates of canopy cover with airborne lidar canopy cover, and (3) Evaluate a modeling-based approach to improve predictions of canopy cover by leveraging available canopy metrics from ATL08 and custom-processed datasets. A total of six (6) potential measures of canopy cover were investigated in this study, based on 100-m ATL08 segments: (i) percentage of photons above 2 m, (ii) percentage of photons above 4.6 m, (iii) percent of canopy (inside canopy) and top-of-canopy photons of total canopy and ground photons, (iv) percent of canopy of total ground and canopy photons, (v) percent of top-of-canopy of total top-of-canopy and ground photons, and (vi) mean of (iii), (iv) and (v). While strongest correlations were produced from canopy cover computed with (ii) (r between 0.72 and 0.84), similar relationships were indicated from using (iii) and (v) (r values between 0.70 and 0.79). Considering a suite of available canopy parameters, the best R2 and RMSE values from canopy cover estimation models were produced from the use of custom-processed ICESat-2-derived metrics (R2 and RMSE values of 0.77 and 13%). Findings from this study highlight methods for estimating canopy cover with ICESat-2 that may be suitable to a range of cover types, in addition to vegetation settings where cut-offs based on tree heights (e.g., 2 m and 4.6 m) are not applicable. With multiple approaches to characterizing canopy cover, results serve to better understand the applicability of ICESat-2 as a data source for canopy cover information and ultimately inform ongoing efforts for deriving an updated gridded product.

Rethinking forest monitoring for more meaningful global forest landscape change assessments.

Forest ecosystems play an indispensable role in addressing various pressing sustainability and social-ecological challenges such as climate change, biodiversity loss, and ecosystem services deterioration, hence the monitoring of the world's forests is crucial. As part of the global forest assessment workflow, a forest is generally classified and mapped based on land use and/or using a tree canopy cover threshold. In this paper, we examine the limitations of this approach and argue that the use of a land use-based forest definition and tree canopy cover thresholds can overlook forest degradation and enhancement, disguise the actual status of forest landscapes, and misinform management planning. These limitations can delay the development and implementation of forest restoration and conservation measures. To help overcome these issues, we propose some enhancements to the global forest assessment workflow, including the sharing of spatial data and inclusion of tree canopy cover estimates in assessment reports. Such enhancements are needed to achieve more meaningful forest monitoring and reporting in the context of global environmental initiatives, such as those related to climate change mitigation and adaptation, forest restoration, biodiversity conservation, and ecosystem services monitoring.

CoveR: an R package for processing digital cover photography images to retrieve forest canopy attributes

Key messagecoveR is an R package for estimating canopy attributes from digital cover photography (DCP) images. The simplicity of the method and the open-accessibility of coveR can effectively extend the accessibility and applicability of DCP to a wider audience.Digital cover photography (DCP) is an increasingly popular tool for estimating canopy cover and leaf area index (LAI). However, existing solutions to process canopy images are predominantly tailored for hemispherical photography, whereas open-access tools for DCP are lacking. We developed an R package (coveR) to support the whole processing of DCP images in an automated, fast, and reproducible way. The package functions, which are designed for step-by-step single-image analysis, can be performed sequentially in a pipeline while ensuring quality-checking of each processing step. A wrapper function ‘coveR()’ is also created to perform all the image processing workflow in a single function. A case study is presented to demonstrate the reliability of canopy attributes derived from coveR in pure beech (Fagus sylvatica L.) stands with variable canopy density and structure. Estimates of gap fraction and effective LAI from DCP were validated against reference measurements obtained from terrestrial laser scanning. By providing a simple, transparent, and flexible image processing procedure, coveR supported the use of DCP for routine measurements and monitoring of forest canopy attributes. This, combined with the possibility to implement DCP in many devices, including smartphones, micro-cameras, and remote trail cameras, can greatly expand the accessibility of the method also by non-experts.

Seeding rate, fertiliser and herbicide effects on canopy growth and productivity of direct-seeded rice (DSR) under different management practices

A low seeding rate with seed drill is a good management practice to promote good canopy growth and increase yield of direct seeding rice (DSR) in lowland rice fields. Many studies did not evaluate seeding rates in an integrated way under different management practices in smallholder farmers’ fields. Estimating canopy cover from Canopeo and normalised vegetation difference index (NDVI) may offer a non-destructive, rapid way to assess crop growth and productivity under different field conditions. On-farm field experiments were conducted comparing the yield and canopy growth in response to seed drilled seeding rates of 20, 40, 60 and 80 kg ha −1 and hand-broadcast seeding rates of 180 kg ha −1 (farmers’ practice) grown under different management conditions in Cambodia. In the first experiment, different seeding rates were compared in two villages, Kork Ton Loab (KTL) and Svay Cheat (SVC), in the 2018 growing season. In the second experiment, seeding rates and fertiliser application rates were compared at Kork Ton Loab location in the 2019 cropping season. The third experiment compared seeding rates and pre-emergence herbicide options at Oroung village in the 2018 and 2020 cropping seasons. In experiment 1, sowing at 20 and 40 kg ha −1 consistently decreased mean canopy cover with Canopeo index at Kork Ton Loab and Svay Cheat locations. In experiment 2, sowing at 80 kg ha −1 significantly increased the mean canopy cover compared with 20 and 40 kg ha −1 seeding rates with Canopeo index at 60 and 75 DAS at the Kork Ton Loab location. The low seeding rates of 60, 80 and 180 kg ha −1 increased in canopy cover at 60 DAS with NDVI compared with the lower seeding rates at Kork Ton Loab and Svay Cheat locations. Seeding rates had significant effects on grain yield and gross margin only for experiment 1 and experiment 3 in the 2020 cropping season. For experiment 1, the seed drill rates of 60 and 80 kg ha −1 did not significantly differ in grain yield or gross margin compared with broadcast of 180 kg ha −1 in both locations. Sowing at 20, 40, 60 and 80 kg ha −1 produced similar yield and gross margin at Kork Ton Loab except for the 20 kg ha −1 seeding rate that significantly reduced yield and gross margin at Svay Cheat. For experiment 3, the seed drill of 80 kg ha −1 and 180 kg ha −1 by broadcast (Farmer’s Practice) significantly increased yield up to 3.5 t ha −1 compared with 40 and 60 seeding rates in the 2020 cropping season. Increasing fertiliser rates (high fertiliser; N-100, P-80, and K-50 kg ha −1 ) by two times increased grain yield by 12.8% compared with the low fertiliser rate of N-50, P-40, and K-30 kg ha −1 . Pre-emergence herbicide application significantly increased mean grain yield by 16% and 15% in the 2018 and 2020 cropping seasons, respectively. Therefore, the seed drill of 80 kg ha −1 , high fertiliser rate and pre-emergence herbicide interventions improved uniformity in canopy growth and increased grain yield of DSR. This information can be incorporated into integrated crop and weed management programs for the productivity of direct-seeded rice in the lowland ecosystem. • Low seeding rates (60–80 kg ha -1 ) with seed drill provide similar yield and gross margin as hand-broadcast of 180 kg ha -1 . • The use of vegetation-derived information from reflectance sensors estimates canopy growth in dynamic field conditions. • Sowing at 60 and 80 kg ha -1 with seed drill reduced undesirable agronomic traits compared with hand-broadcast of 180 kg ha -1 .

Combining post-disturbance land cover and tree canopy cover from Landsat time series data for mapping deforestation, forest degradation, and recovery across Cambodia

ABSTRACT Mapping of deforestation, forest degradation, and recovery is essential to characterize country-level forest change and formulate mitigation actions. Previous studies have mainly used a simple forest/non-forest classification after forest disturbance to identify deforestation and forest degradation. However, a more flexible approach that is applicable to different forest conditions is desirable. In this study, we examined an approach for mapping deforestation, forest degradation, and recovery using disturbance types and tree canopy cover estimates from annual Landsat time-series data from 1988 to 2020 across Cambodia. We developed models to estimate both disturbance types and tree canopy cover based on a random forest algorithm using predictor variables derived from a trajectory-based temporal segmentation approach. The estimated disturbance types and canopy cover in each year were then used in a rule-based classification of deforestation, forest degradation, and recovery. The producer’s and user’s accuracies ranged from 59.1% to 72.9% and 60.8% to 91.6%, respectively, for the forest change classes of mapping deforestation, forest degradation, and recovery. The approach developed here can be adjusted for different definitions of deforestation, forest degradation, and recovery according to research objectives and thus has the potential to be applied to other study areas.

A Comparison of the Performances of Unmanned-Aerial-Vehicle (UAV) and Terrestrial Laser Scanning for Forest Plot Canopy Cover Estimation in Pinus massoniana Forests

Canopy cover is an important indicator and commonly used in forest management applications. Unmanned-Aerial-Vehicle (UAV)—Borne Laser Scanning (ULS) has drawn increasing attention as a new alternative source for forest field inventory due to its spatial resolution comparable to that of Terrestrial Laser Scanning (TLS). In this study, the performance of plot canopy cover estimations from ULS and TLS is investigated. The experiment was conducted in 16 plots from two Pinus massoniana forests with different stand conditions in Guangxi, China. Both the Canopy Height Model (CHM)-based and Individual Tree Delineation (ITD)-based methods were used to estimate the canopy cover. The influence of CHM pixel sizes on the estimations was also analyzed. Our results demonstrated that the accuracies of ULS (R2: 0.992–0.996, RMSE: 0.591–0.820%) were better than those of TLS (R2: 0.541–0.846, RMSE: 3.642–6.297%) when compared against the reference. The average difference between the ULS and TLS estimations was 6.91%, and the disagreement increased as the forest complexity increased. The reasonable CHM pixel sizes for the canopy cover estimations were 0.07–1.2 m for ULS and 0.07–1.5 m for TLS. This study can provide useful information for the selection of data sources and estimation methods in plot canopy cover mapping.

Tracking the prevalence of a fungal pathogen, Batrachochytrium dendrobatidis (chytrid fungus), using environmental DNA

AbstractChytridiomycosis, a primary disease driving widespread and unprecedented amphibian declines, is caused by the fungal pathogen Batrachochytrium dendrobatidis (Bd). Tracking Bd through space and time requires monitoring protocols that efficiently and reliably assess pathogen prevalence and intensity, which in turn requires an understanding of environment–pathogen dynamics. Environmental DNA (eDNA) was used to track Bd prevalence and intensity in 95 waterbodies in southern Ontario, Canada, and assess zoospore counts relative to biotic, abiotic, and geographic factors. Bd was also monitored on a semi‐weekly basis in 10 waterbodies to better understand patterns of temporal variability. Bd showed variable prevalence, with 47% and 29% of waterbodies having zoospores detected in May and July, respectively. Patterns of prevalence were markedly variable both within and across waterbodies, indicating high spatio‐temporal heterogeneity. Bd prevalence was not related to environmental factors, geographic variables, or amphibian species richness, but intensity was negatively related to estimated canopy cover. In intensively sampled waterbodies, Bd counts were highly variable through time, with some sites switching from detection to non‐detection (and vice versa) across 2‐week intervals. We conclude that eDNA can be a useful tool for monitoring Bd zoospores in wetlands but emphasize the need for additional research into environmental and methodological factors affecting zoospore detection and abundance before this method should be widely adopted.

Wheat Canopy Cover Estimation by Optimized Random Forest and UAV Multispectral imagery

Canopy cover estimation is widely applied to reflect crop status in agriculture research and management. In particular, an accurate CC estimation is beneficial for crop model calibration, providing high-accuracy observations. Recent solutions on CC are drawn by experimental regression or basic machine learning classifier because CC estimation can be treated as a wheat/non-wheat segmentation task. However, the appearance of hyperparameters in such machine learning algorithms impairs the segmentation performance. In this paper, by the means of UAV multispectral imagery, Bayesian optimization based Random Forest approach is selected to tune the uncertain hyperparameters accurately and robustly, providing a novel way on CC estimation. Experimental results collected in Yangling experiment field by the RedEdge camera on DJI M100 UAV are to evaluate the proposed method. Comparative studies show that the overall accuracy can reach up to 99.9%, promoting 0.2% in comparison with basic Random Forest. Therefore, integrating optimized Random Forest and UAV multispectral imagery can be applied in CC estimation at farmland scales.

Linear Feature-Based Triangulation for Large-Scale Orthophoto Generation Over Mechanized Agricultural Fields

Unmanned aerial vehicles (UAVs) equipped with imaging and ranging sensors have become an effective remote sensing data acquisition tool for digital agriculture. Among potential products derived from UAVs, high-resolution orthophotos play an important role in several phenotyping activities, such as canopy cover estimation and flowering date identification. Current structure from motion (SfM) tools for image-based 3-D reconstruction and orthophoto generation cannot perform well when working with large-scale imagery over mechanized agricultural fields. This failure is mainly due to their inability to identify enough conjugate points among overlapping images captured at low altitudes. This study addresses such limitation through a new strategy that uses plant row segments as linear features in the triangulation process. The linear features are derived in two steps. First, an automated approach is implemented to extract plant row segments from the LiDAR data which are then back-projected to the imagery using available trajectory and system calibration parameters. In the second step, a machine-assisted strategy is used to adjust the line segments in image space for deriving accurate linear features. In the proposed framework, the triangulation process is conducted by investigating two mathematical models—referred to as object-space and image-space coplanarity constraints—for incorporating linear features in the bundle adjustment (BA). The orthophoto is generated using the refined trajectory and system calibration parameters derived from the BA process. Several experimental results over an agricultural filed show that the proposed framework outperforms commonly used SfM tools, e.g., Pix4D Mapper Pro and Agisoft Metashape in terms of generating orthophotos with high visual quality and geolocation accuracy. Also, results indicate that the object-space coplanarity constraint is more robust against potential noise in line measurements when compared to the image-space coplanarity model. However, both models lead to high absolute accuracy in the range of ±2–4 cm when the noise level in the image measurements of points along the line is reasonable, i.e., ~5–10 pixels.

Assessment of cotton and sorghum stand establishment using UAV‐based multispectral and DSLR‐based RGB imagery

AbstractPlant density and canopy cover are key agronomic traits for cotton (Gossypium hirsutum L.) and sorghum [Sorghum bicolor (L.) Moench] phenotypic evaluation. The objective of this study was to evaluate utility of broadband red–green–blue (RGB) and narrowband green, red, red‐edge, and near‐infrared spectral data taken by an unmanned aerial vehicle (UAV), and RGB taken by a digital single‐lens reflex camera for assessing the cotton and sorghum stands. Support Vector Machine was used to analyze UAV images, whereas ImageJ was used for RGB images. Fifteen vegetation indices (VIs) were evaluated for their accuracy, predictability, and residual yield. All VIs had Cohen's k > .65, F score > .63, and User and Producer accuracy of more than 71 and 69%, respectively. Soil‐adjusted vegetation indices (SAVIs) among narrowband VIs and excess green minus excess red (ExG–ExR) among broadband VIs provided more agreeable estimates of cotton and sorghum density than the remaining VIs with R2 and index of agreement (IoA) up to .79 and .92, respectively. The estimated canopy cover explained up to 83 and 82% variability in leaf area index (LAI) of cotton and sorghum, respectively. The ImageJ produced R2 from .79 to .90 and .83 to .86 and IoA .89 to .97 and ∼.91 between estimated and observed cotton and sorghum density, respectively. ImageJ explained up to 82 and 79% variability in cotton and sorghum LAI, respectively. Although ImageJ can give close estimates of crop density and cover, UAV‐based narrowband VIs still can provide an agreeable, reliable, and time‐efficient estimate of these attributes.

Rainfall interception by mountain big sagebrush (Artemisia tridentataspp.vaseyana): Dryland shrub canopy cover affects net precipitation

AbstractCanopy interception loss is an important component of the water budget for many ecosystems, and may be particularly influential in semi‐arid shrublands where water is limiting. In this experiment, we quantified interception loss by mountain big sagebrush (Artemisia tridentata spp. vaseyana)using simulated rainfall events in the field. Various levels of canopy cover and rainfall intensities were tested to measure their effects on net precipitation and interception loss. Additionally, the comparisons were made between three methods of measuring shrub cover to enhance the scalability of results. Stands of sagebrush from 20% to 50% canopy cover intercepted 18.5 ± 12.5% of incoming precipitation. Percent interception loss differed by 13.7% between the high and low cover classes, indicating that net precipitation may be significantly reduced under mature, dense stands of sagebrush. Hemispherical photography was a viable method for estimating canopy cover in this vegetation type. Interception loss by sagebrush was 14% lower on average than pinyon and juniper measured in the same watershed. Results from this study quantified the effects of sagebrush canopy on rainfall interception and improved understanding of vegetation dynamics in the sagebrush steppe.

Using canopy height model derived from UAV imagery as an auxiliary for spectral data to estimate the canopy cover of mixed broadleaf forests.

Canopy cover is an important structural trait that is frequently used in forest inventories to assess sustainability as well as many other important aspects of forest stands. Remote sensing data is more effective and suitable for canopy cover estimating than traditional field measurements such as sample plots, especially at broad scales. Measurement and mapping this attribute in fine-scale is a difficult task. Aerial imagery using unmanned aerial vehicle (UAV) has been recognized as an excellent tool to estimate canopy attributes. In this study, we compared the potential of using digital hemispherical photography (DHP), digital cover photography (DCP), UAV RGB data, and canopy height model (CHM) for estimation of canopy cover of mix broad-leaved forest on seven different stands. The canopy cover was measured from two digital canopy photographic methods, including DHP (as the reference method) and DCP. The stand orthophotos were segmented using a multi-resolution image segmentation method. Afterward, the classification in two classes of the canopy cover and the non-canopy cover was conducted using minimum distance classification to estimate canopy cover. The CHM layer was generated based on the SfM algorithm and utilized in the canopy cover estimation in each stand as auxiliary data. The results showed a slight improvement when we used the CHM as auxiliary data. Overall, the results showed that the efficiency of the ground digital canopy photographic methods (zenith view) in multi-storied and dense forests is the lowest. In return, our method for digital aerial canopy photography (object-based canopy segmentation and classification) is simple, quick, efficient, and cost-effective.

Ground layer Cerrado plants sustain higher maximum photosynthetic rates after medium-term fire events

Fire is one of the most important factors driving community assembly and ecosystem functioning in tropical savannas. However, few studies have evaluated the physiological responses of ground layer plant communities to fire disturbance. Here we used different fire regimes to investigate possible changes in leaf maximum gas exchange (Amax and gs) and leaf nutritional content (N, P, K, Ca and Mg) among different plant growth forms in savanna ground layer communities. We compared responses of ground layer plant communities under two different fire regimes: (1) no recent fire occurrence; and (2) two recurrent fire events in the last 20 years. We estimated canopy cover, soil chemical properties and species abundance on burned and unburned plots in order to calculate abundance-weighted species average trait values for gas exchange and leaf nutrient content. We found that burned plots exhibited lower canopy cover and soil organic matter content, and an overall higher soil macronutrients availability compared to unburned plots. These environmental differences clearly influenced the ground layer plant communities, which depicted higher Amax and gs in burned areas regardless of growth form. We found no significant differences among leaf nutrient traits, except for a lower Mg concentration in the burned site species. Our results support the hypothesis that distinct fire regimes select for a different set of leaf functional traits, with fire occurrence acting as an important driver increasing the maximum photosynthetic rate on the ground layer. While nutrient use seems not to be affected by medium-term recurrent fires, physiological plasticity on carbon and water use processes in response to changes in resource availability can promote the persistence of savanna species under frequent fire.

A High Resolution Image Based Approach for Estimating the Canopy Cover of a Semi-Deciduous Brazilian Atlantic Forest Fragment

The Atlantic Forest in Brazil has been suffering natural and anthropic perturbations over the years, which has been impacting, qualitatively and quantitatively, with different degrees and intensities, its canopy cover. These perturbations can be caused by the selective cut of species, burning, natural death of trees, among several other factors, which act directly on the composition and the floristic diversity of fragments of the forest. The main motivation for conducting this work was to alert about the relevance of the Atlantic forest conservation and the need for a constant monitoring system to preserve what has been left of it. This can be partially achieved by permanently estimating its canopy cover. This paper describes the implementation and evaluation of the descriptors: energy, entropy, homogeneity, contrast and the sum of the high-frequency Discrete Fourier Transform (DFT), for estimating canopy cover based on High Resolution Camera (HRC) satellite images. The analyses carried out based on the results of the experiments showed that the energy descriptor was the best descriptor among those used and, comparatively to the others, obtained the higher correlation (r), determination (R) and significance (p). Based on the experiments, the energy descriptor presented an expressive potential to be used in processes for estimating the canopy cover based on satellite images, in large areas of the semi-deciduous forest.

Predicting canopy cover of diverse forest types from individual tree measurements

Quantifying tree canopy cover is fundamental to applications in forestry and ecology, but estimates vary substantially depending on type of field measurement, imagery, or active sensing used. Our objective was to improve estimates of stand-level canopy cover from standard tree inventory measurements, using representative data collected across diverse forest plant association groups across Oregon, USA. Canopy cover was measured with line intercept sampling on 1706 inventory plots and compared to calculations from individually tallied trees. We investigated adjustments of tree crown area equations, adjustments of crown overlap factors, and modeling from climatic variables and standard forest measurements to estimate line intercept cover. Estimates based on simple crown width equations adjusted for tree social position and caps on maximum cover, had the lowest error (RMSE = 14% cover) of crown width approaches across all vegetation types. Random crown overlap applied to unadjusted crown area only performed well in drier forest types and was unable to match high line intercept cover levels (>90%) often found in productive forest types. Although statistical models had somewhat greater precision than the simpler crown-width summation approaches (RMSE of 12%), accuracy was comparable. The greater flexibility of crown width summation approaches could make them more useful in forestry applications and beyond our study area.

Tree diversity and composition in Mexican traditional smallholder cocoa agroforestry systems

The cultivation of cocoa is a commodity of great importance worldwide. In Mexico, cocoa is grown in Tabasco and Chiapas states in agroforestry systems since pre-Hispanic times, where cocoa is grown under the canopy of shade trees. Crops such as sugarcane and extensive livestock production have gradually reduced the area devoted to cocoa cultivation in the region. Yet, farmers keep small portions of the cocoa agroforestry systems to maintain the local floristic diversity. This practice seems to be a successful contribution to plant conservation. To determine the management practices used by farmers in the agroforestry cocoa systems and the diversity of products they harvest, a non-probabilistic sampling was carried out in 38 shade cocoa plots of 20 × 50 m each (19 in Tabasco and 19 in Chiapas). We counted and identified all trees with DBH ≥ 5 cm, recording their height. We estimated canopy cover at 20 points within each plot. Our results show no tree species richness differences between states, but there were differences at the municipality level; Pichucalco (Chiapas) had the highest tree species richness. Considering all tree species, there was a higher tree density in Tabasco than in Chiapas. Regarding only cocoa trees, there was also a higher abundance in Tabasco than in Chiapas. In both cases, farmers obtain a wide variety of products for self-consumption and local market sale from their cocoa agroforestry systems (e.g., timber, fruit, and grains). The wide floristic diversity in these agroecosystems aids native plant species conservation and could favor the increase of agroforestry plantations associated with cocoa cultivation.

Estimating canopy cover in artificial forests using high spatial resolution GF-1 and ZY-3 images: across-sensor and across-site comparison

ABSTRACT Canopy cover is an important parameter in forest ecosystems and has diverse applications in a wide variety of fields. However, estimation of forest canopy cover from high spatial resolution images is an arduous task. To evaluate the utility of high spatial resolution images for estimation of canopy cover of artificial forests, the performance of canopy cover estimation models across sensors and across sites was evaluated. Choosing Wangyedian Forest Farm and Gaofeng Forest Farm as experimental areas, based on Chinese high spatial resolution Gaofen-1 (GF-1) and Ziyuan-3 (ZY-3) satellite images, three models, namely multiple linear regression (MLR), generalized additive model (GAM), and Random Forest (RF) were established. The capabilities of the two data sources and three model types to estimate canopy cover of the forests were compared and the portability of the three models across sensors and sites was analysed. The ranges of the root mean square error (RMSE) and relative root mean square error (rRMSE) of the MLR, GAM, and RF models for each study area established from GF-1 and ZY-3 were 0.0632–0.1205 and 9.98–19.93%, respectively. GF-1 showed better performance than ZY-3 when using the same model type. Whether the objective was across sensors or across sites transplantation, the accuracy of the model decreased with transplantation of the model. The MLR model was non-significant when transplanted across sites and model migration failed, whereas across sensors and across sites transplantation of RF and GAM models was successful. The three evaluation indices of transplanted GAM models were significantly superior to those of transplanted RF and MLR models. Thus, the GAM model showed strong portability and the highest model stability.

A FULLY AUTOMATED AND FAST APPROACH FOR CANOPY COVER ESTIMATION USING SUPER HIGH-RESOLUTION REMOTE SENSING IMAGERY

Abstract. Canopy cover is a key agronomic variable for understanding plant growth and crop development status. Estimation of canopy cover rapidly and accurately through a fully automated manner is significant with respect to high throughput plant phenotyping. In this work, we propose a simple, robust and fully automated approach, namely a rule-based method, that leverages the unique spectral pattern of green vegetation at visible (VIS) and near-infrared red (NIR) spectra regions to distinguish the green vegetation from background (i.e., soil, plant residue, non-photosynthetic vegetation leaves etc.), and then derive canopy cover. The proposed method was applied to high-resolution hyperspectral and multispectral imagery collected from gantry-based scanner and Unmanned Aerial Vehicle (UAV) platforms to estimate canopy cover. Additionally, machine learning methods, i.e., Support Vector Machine (SVM) and Random Forest (RF) were also employed as bench mark methods. The results show that: the rule-based method demonstrated promising classification accuracies that are comparable to SVM and RF for both hyperspectral and multispectral datasets. Although the rule-based method is more sensitive to mixed pixels and shaded canopy region, which potentially resulted in classification errors and underestimation of canopy cover in some cases; it showed better performance to detect smaller leaves than SVM and RF. Most importantly, the rule-based method substantially outperformed machine learning methods with respect to processing speed, indicating its greater potential for high-throughput plant phenotyping applications.

Adjudicating Perspectives on Forest Structure: How Do Airborne, Terrestrial, and Mobile Lidar-Derived Estimates Compare?

Applications of lidar in ecosystem conservation and management continue to expand as technology has rapidly evolved. An accounting of relative accuracy and errors among lidar platforms within a range of forest types and structural configurations was needed. Within a ponderosa pine forest in northern Arizona, we compare vegetation attributes at the tree-, plot-, and stand-scales derived from three lidar platforms: fixed-wing airborne (ALS), fixed-location terrestrial (TLS), and hand-held mobile laser scanning (MLS). We present a methodology to segment individual trees from TLS and MLS datasets, incorporating eigen-value and density metrics to locate trees, then assigning point returns to trees using a graph-theory shortest-path approach. Overall, we found MLS consistently provided more accurate structural metrics at the tree- (e.g., mean absolute error for DBH in cm was 4.8, 5.0, and 9.1 for MLS, TLS and ALS, respectively) and plot-scale (e.g., R2 for field observed and lidar-derived basal area, m2 ha−1, was 0.986, 0.974, and 0.851 for MLS, TLS, and ALS, respectively) as compared to ALS and TLS. While TLS data produced estimates similar to MLS, attributes derived from TLS often underpredicted structural values due to occlusion. Additionally, ALS data provided accurate estimates of tree height for larger trees, yet consistently missed and underpredicted small trees (≤35 cm). MLS produced accurate estimates of canopy cover and landscape metrics up to 50 m from plot center. TLS tended to underpredict both canopy cover and patch metrics with constant bias due to occlusion. Taking full advantage of minimal occlusion effects, MLS data consistently provided the best individual tree and plot-based metrics, with ALS providing the best estimates for volume, biomass, and canopy cover. Overall, we found MLS data logistically simple, quickly acquirable, and accurate for small area inventories, assessments, and monitoring activities. We suggest further work exploring the active use of MLS for forest monitoring and inventory.

Comparing airborne and spaceborne photon-counting LiDAR canopy structural estimates across different boreal forest types

The monitoring of forested ecosystems relies on an accurate description of forest structure. The Ice, Cloud and land Elevation Satellite-2 (ICESat-2), launched in September 2018, carries the Advanced Topographic Laser Altimeter System (ATLAS), a Light Detection and Ranging (LiDAR) instrument capable of detecting individual photons reflected back from vegetation canopy. ICESat-2 data is delivering global estimates of forest structure; however, analysis of the performance of ICESat-2 on-orbit data across a range of forest conditions remains limited. This study derives structural estimates of (i) canopy height, (ii) canopy cover and (iii) canopy height variability from ICESat-2 data acquired in snow-free and low atmospheric scattering conditions over different boreal forest structural types in Ontario, Canada. ICESat-2 structural estimates were derived from the Global Geolocated Photon Data (ATL03) and Land and Vegetation Height (ATL08) data products and compared against single-photon detection airborne LiDAR (Leica SPL100). An extensive network of ground plots were used to stratify the study area into three distinct forest structural groups, each resulting from different stand development stages. ICESat-2 and SPL100 estimates of canopy height were compared at the ATL03 photon level, whereas estimates of height variability and canopy cover were compared for spatial analysis units (AU; mean size = 1287 m2). ICESat-2 photons returned from the top of the canopy underestimated canopy height relative to SPL100 by an average of 2.3 m overall and corresponded most strongly to the 90th percentile (P90) of coincident airborne SPL100 returns (root mean square difference (RMSD) = 2.9 m and correlation coefficient (r) = 0.84). The lowest average underestimation of SPL P90 was observed in homogeneous stands that were relatively simple, and single-layered with a single dominant species (RMSD = 2.5 m, r = 0.84). We observed the least agreement of ICESat-2 and SPL forest structural metrics in over-mature stands with complex structure and greater variability in canopy heights (RMSD = 3.5 m, r = 0.64). For the AUs, the strength of the relationship between SPL100 and ICESat-2 canopy height percentiles increased with increasing height percentiles (e.g. P25 RMSD% = 77.8%; P95 RMSD% = 23.7%). ICESat-2 generally underestimated canopy height variability relative to the SPL100 data, with both data having similar absolute variability (standard deviation of canopy heights RMSD% = 26.8%, r = 0.75), but lower agreement in relative variability (coefficient of variation of canopy heights RMSD% = 33.9%, r = 0.45). Herein we propose the use of the vegetation fill index as a method to estimate canopy cover with ICESat-2. Comparison of SPL100 and ICESat-2 vegetation fill indices at the AU level resulted in strong agreement overall (RMSD% = 19.7%; r = 0.57). These observations and results contribute to the overall objective of building a comprehensive understanding of the performance of ICESat-2 for characterizing vegetation structure in boreal forest environments.