1
|
Li Z, Zhou X, Cheng Q, Zhai W, Mao B, Li Y, Chen Z. An integrated feature selection approach to high water stress yield prediction. FRONTIERS IN PLANT SCIENCE 2023; 14:1289692. [PMID: 38111876 PMCID: PMC10726204 DOI: 10.3389/fpls.2023.1289692] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Subscribe] [Scholar Register] [Received: 09/06/2023] [Accepted: 11/17/2023] [Indexed: 12/20/2023]
Abstract
The timely and precise prediction of winter wheat yield plays a critical role in understanding food supply dynamics and ensuring global food security. In recent years, the application of unmanned aerial remote sensing has significantly advanced agricultural yield prediction research. This has led to the emergence of numerous vegetation indices that are sensitive to yield variations. However, not all of these vegetation indices are universally suitable for predicting yields across different environments and crop types. Consequently, the process of feature selection for vegetation index sets becomes essential to enhance the performance of yield prediction models. This study aims to develop an integrated feature selection method known as PCRF-RFE, with a focus on vegetation index feature selection. Initially, building upon prior research, we acquired multispectral images during the flowering and grain filling stages and identified 35 yield-sensitive multispectral indices. We then applied the Pearson correlation coefficient (PC) and random forest importance (RF) methods to select relevant features for the vegetation index set. Feature filtering thresholds were set at 0.53 and 1.9 for the respective methods. The union set of features selected by both methods was used for recursive feature elimination (RFE), ultimately yielding the optimal subset of features for constructing Cubist and Recurrent Neural Network (RNN) yield prediction models. The results of this study demonstrate that the Cubist model, constructed using the optimal subset of features obtained through the integrated feature selection method (PCRF-RFE), consistently outperformed the RNN model. It exhibited the highest accuracy during both the flowering and grain filling stages, surpassing models constructed using all features or subsets derived from a single feature selection method. This confirms the efficacy of the PCRF-RFE method and offers valuable insights and references for future research in the realms of feature selection and yield prediction studies.
Collapse
Affiliation(s)
| | | | | | | | | | | | - Zhen Chen
- Institute of Farmland Irrigation, Chinese Academy of Agricultural Sciences, Xinxiang, China
| |
Collapse
|
2
|
Kuhl E, Zang C, Esper J, Riechelmann DFC, Büntgen U, Briesch M, Reinig F, Römer P, Konter O, Schmidhalter M, Hartl C. Using machine learning on tree‐ring data to determine the geographical provenance of historical construction timbers. Ecosphere 2023. [DOI: 10.1002/ecs2.4453] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 03/11/2023] Open
Affiliation(s)
- Eileen Kuhl
- Department of Geography Johannes Gutenberg University Mainz Germany
| | - Christian Zang
- Department of Forestry University of Applied Science Weihenstephan‐Triesdorf Freising Germany
| | - Jan Esper
- Department of Geography Johannes Gutenberg University Mainz Germany
- Global Change Research Centre (CzechGlobe) Brno Czech Republic
| | | | - Ulf Büntgen
- Global Change Research Centre (CzechGlobe) Brno Czech Republic
- Department of Geography University of Cambridge Cambridge UK
- Swiss Federal Research Institute (WSL) Birmensdorf Switzerland
- Department of Geography Masaryk University Brno Czech Republic
| | - Martin Briesch
- Department of Information Systems and Business Administration Johannes Gutenberg University Mainz Germany
| | - Frederick Reinig
- Department of Geography Johannes Gutenberg University Mainz Germany
| | - Philipp Römer
- Department of Geography Johannes Gutenberg University Mainz Germany
| | - Oliver Konter
- Department of Geography Johannes Gutenberg University Mainz Germany
| | | | - Claudia Hartl
- Nature Rings ‐ Environmental Research and Education Mainz Germany
| |
Collapse
|
3
|
Tang C, Ye Z, Long J, Liu Z, Zhang T, Xu X, Lin H. Mapping forest and site quality of planted Chinese fir forest using sentinel images. FRONTIERS IN PLANT SCIENCE 2022; 13:949598. [PMID: 36267948 PMCID: PMC9577201 DOI: 10.3389/fpls.2022.949598] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Subscribe] [Scholar Register] [Received: 05/21/2022] [Accepted: 09/02/2022] [Indexed: 06/16/2023]
Abstract
Normally, forest quality (FQ) and site quality (SQ) play an important role in evaluating actual and potential forest productivity. Traditionally, these assessment indices (FQ and SQ) are mainly based on forest parameters extracted from ground measurement (forest height, age, density, forest stem volume (FSV), and DBH), which is labor-intensive and difficult to access in certain remote forest areas. Recently, remote sensing images combined with a small number of samples were gradually applied to map forest parameters because of the various advantages of remote sensing technology, such as low cost, spatial coverage, and high efficiency. However, FQ and SQ related to forest parameters are rarely estimated using remote sensing images and machine learning models. In this study, the Sentinel images and ground samples of planted Chinese fir forest located in the ecological "green-core" area of Changzhutan urban cluster, were initially employed to explore the feasibility of mapping the FQ and SQ. And then, four types of alternative variables (backscattering coefficients (VV and VH), multi-spectral bands, vegetation indices, and texture characteristics) were extracted from Sentinel-1A and Sentinel-2A images, respectively. After selecting variables using a stepwise regression model, three machine learning models (SVR, RF, and KNN) were employed to estimate various forest parameters. Finally, the FQ of the study region was directly mapped by the weights sum of related factors extracted by the factor analysis method, and the SQ was also extracted using mapped forest height and age. The results illustrated that the accuracy of estimated forest parameters (DBH, H, and Age) was significantly higher than FSV, FCC, and Age and the largest and smallest rRMSEs were observed from FSV (0.38~0.40) and forest height (0.20~0.21), respectively. Using mapped forest parameters, it also resulted that the rRMSEs of estimated FQ and SQ were 0.19 and 0.15, respectively. Furthermore, after normalization and grading, the grades of forest quality were mainly concentrated in grades I, II, and III in the study region. Though the accuracy of mapping FQ and SQ is limited by the saturation phenomenon, it is significantly proved that using machine learning models and Sentinel images has great potential to indirectly map FQ and SQ.
Collapse
Affiliation(s)
- Chongjian Tang
- Research Center of Forestry Remote Sensing & Information Engineering, Central South University of Forestry and Technology, Changsha, China
- Key Laboratory of Forestry Remote Sensing Based Big Data & Ecological Security for Hunan Province, Changsha, China
- Key Laboratory of State Forestry Administration on Forest Resources Management and Monitoring in Southern Area, Changsha, China
| | - Zilin Ye
- Research Center of Forestry Remote Sensing & Information Engineering, Central South University of Forestry and Technology, Changsha, China
- Key Laboratory of Forestry Remote Sensing Based Big Data & Ecological Security for Hunan Province, Changsha, China
| | - Jiangping Long
- Research Center of Forestry Remote Sensing & Information Engineering, Central South University of Forestry and Technology, Changsha, China
- Key Laboratory of Forestry Remote Sensing Based Big Data & Ecological Security for Hunan Province, Changsha, China
- Key Laboratory of State Forestry Administration on Forest Resources Management and Monitoring in Southern Area, Changsha, China
| | - Zhaohua Liu
- Research Center of Forestry Remote Sensing & Information Engineering, Central South University of Forestry and Technology, Changsha, China
- Key Laboratory of Forestry Remote Sensing Based Big Data & Ecological Security for Hunan Province, Changsha, China
- Key Laboratory of State Forestry Administration on Forest Resources Management and Monitoring in Southern Area, Changsha, China
| | - Tingchen Zhang
- Research Center of Forestry Remote Sensing & Information Engineering, Central South University of Forestry and Technology, Changsha, China
- Key Laboratory of Forestry Remote Sensing Based Big Data & Ecological Security for Hunan Province, Changsha, China
| | - Xiaodong Xu
- Research Center of Forestry Remote Sensing & Information Engineering, Central South University of Forestry and Technology, Changsha, China
- Key Laboratory of Forestry Remote Sensing Based Big Data & Ecological Security for Hunan Province, Changsha, China
- Key Laboratory of State Forestry Administration on Forest Resources Management and Monitoring in Southern Area, Changsha, China
| | - Hui Lin
- Research Center of Forestry Remote Sensing & Information Engineering, Central South University of Forestry and Technology, Changsha, China
- Key Laboratory of Forestry Remote Sensing Based Big Data & Ecological Security for Hunan Province, Changsha, China
- Key Laboratory of State Forestry Administration on Forest Resources Management and Monitoring in Southern Area, Changsha, China
| |
Collapse
|
4
|
Individual Tree Basal Area Increment Models for Brazilian Pine (Araucaria angustifolia) Using Artificial Neural Networks. FORESTS 2022. [DOI: 10.3390/f13071108] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 02/06/2023]
Abstract
This research aimed to develop statistical models to predict basal area increment (BAI) for Araucaria angustifolia using Artificial Neural Networks (ANNs). Tree species were measured for their biometric variables and identified at the species level. The data were subdivided into three groups: (1) intraspecific competition with A. angustifolia; (2) the first group of species that causes interspecific competition with A. angustifolia; and (3) the second group of species that causes interspecific competition with A. angustifolia. We calculated both the dependent and independent distance and the described competition indices, considering the impact of group stratification. Multi-layer Perceptron (MLP) ANN was structured for modeling. The main results were that: (i) the input variables size and competition were the most significant, allowing us to explain up to 77% of the A. angustifolia BAI variations; (ii) the spatialization of the competing trees contributed significantly to the representation of the competitive status; (iii) the separate variables for each competition group improved the performance of the models; and (iv) besides the intraspecific competition, the interspecific competition also proved to be important to consider. The ANN developed showed precision and generalization, suggesting it could describe the increment of a species common in native forests in Southern Brazil and with potential for upcoming forest management initiatives.
Collapse
|
5
|
Disentangling the Factors That Contribute to the Growth of Betula spp. and Cunninghami lanceolata in China Based on Machine Learning Algorithms. SUSTAINABILITY 2022. [DOI: 10.3390/su14148346] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 01/25/2023]
Abstract
Forests are indispensable materials and spiritual foundations for promoting ecosystem circulation and human survival. Exploring the environmental impact mechanism on individual-tree growth is of great significance. In this study, the effects of biogeoclimate, competition, and topography on the growth of Betula spp. and Cunninghamia lanceolata (Lamb.) Hook., two tree species with high importance value in China, were explored by gradient boosting regression tree (GBRT), k-nearest neighbor (KNN), and random forest (RF) machine learning (ML) algorithms. The results showed that the accuracy of RF was better than KNN, which was better than GBRT. All ML algorithms performed well for future diameter at breast height (DBH) predictions; the Willmott’s indexes of agreement (WIA) of each ML algorithm in predicting the future DBH were all higher than 0.97, and the R2 was higher than 0.98 and 0.90, respectively. The individual tree annual growth rate is mainly affected by the single-tree size, and the external environment can promote or inhibit tree growth. Climate and stand structure variables were relatively more important for tree growth than the topographic factors. Lower temperature and precipitation, higher stand density, and canopy closure were more unfavorable for their growth. In afforestation, the following factors should be considered in order: geographic location, meteorological climate, stand structure, and topography.
Collapse
|
6
|
He J, Fan C, Geng Y, Zhang C, Zhao X, von Gadow K. Assessing scale-dependent effects on Forest biomass productivity based on machine learning. Ecol Evol 2022; 12:e9110. [PMID: 35845366 PMCID: PMC9277413 DOI: 10.1002/ece3.9110] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/23/2022] [Revised: 06/21/2022] [Accepted: 06/24/2022] [Indexed: 12/02/2022] Open
Abstract
Estimating forest above‐ground biomass (AGB) productivity constitutes one of the most fundamental topics in forest ecological research. Based on a 30‐ha permanent field plot in Northeastern China, we modeled AGB productivity as output, and topography, species diversity, stand structure, and a stand density variable as input across a series of area scales using the Random Forest (RF) algorithm. As the grain size increased from 10 to 200 m, we found that the relative importance of explanatory variables that drove the variation of biomass productivity varied a lot, and the model accuracy was gradually improved. The minimum sampling area for biomass productivity modeling in this region was 140 × 140 m. Our study shows that the relationship of topography, species diversity, stand structure, and stand density variables with biomass productivity modeled using the RF algorithm changes when moving from scales typical of forest surveys (10 m) to larger scales (200 m) within a controlled methodology. These results should be of considerable interest to scientists concerned with forest assessment.
Collapse
Affiliation(s)
- Jingyuan He
- Research Center of Forest Management Engineering of State Forestry Administration Beijing Forestry University Beijing China
| | - Chunyu Fan
- Research Center of Forest Management Engineering of State Forestry Administration Beijing Forestry University Beijing China
| | - Yan Geng
- Research Center of Forest Management Engineering of State Forestry Administration Beijing Forestry University Beijing China
| | - Chunyu Zhang
- Research Center of Forest Management Engineering of State Forestry Administration Beijing Forestry University Beijing China
| | - Xiuhai Zhao
- Research Center of Forest Management Engineering of State Forestry Administration Beijing Forestry University Beijing China
| | - Klaus von Gadow
- Research Center of Forest Management Engineering of State Forestry Administration Beijing Forestry University Beijing China.,Faculty of Forestry and Forest Ecology Georg-August-University Göttingen Germany.,Department of Forest and Wood Science University of Stellenbosch Matieland South Africa
| |
Collapse
|
7
|
Quantifying the Effects of Stand and Climate Variables on Biomass of Larch Plantations Using Random Forests and National Forest Inventory Data in North and Northeast China. SUSTAINABILITY 2022. [DOI: 10.3390/su14095580] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 02/05/2023]
Abstract
The accurate estimation of forest biomass is crucial for supporting climate change mitigation efforts such as sustainable forest management. Although traditional regression models have been widely used to link stand biomass with biotic and abiotic predictors, this approach has several disadvantages, including the difficulty in dealing with data autocorrelation, model selection, and convergence. While machine learning can overcome these challenges, the application remains limited, particularly at a large scale with consideration of climate variables. This study used the random forests (RF) algorithm to estimate stand aboveground biomass (AGB) and total biomass (TB) of larch (Larix spp.) plantations in north and northeast China and quantified the contributions of different predictors. The data for modelling biomass were collected from 445 sample plots of the National Forest Inventory (NFI). A total of 22 independent variables (6 stand and 16 climate variables) were used to develop and train climate-sensitive stand biomass models. Optimization of hyper parameters was implemented using grid search and 10-fold cross-validation. The coefficient of determination (R2) and root mean square error (RMSE) of the RF models were 0.9845 and 3.8008 t ha−1 for AGB, and 0.9836 and 5.1963 t ha−1 for TB. The cumulative contributions of stand and climate factors to stand biomass were >98% and <2%, respectively. The most crucial stand and climate variables were stand volume and annual heat-moisture index (AHM), with relative importance values of >60% and ~0.25%, respectively. The partial dependence plots illustrated the complicated relationships between climate factors and stand biomass. This study illustrated the power of RF for estimating stand biomass and understanding the effects of stand and climate factors on forest biomass. The application of RF can be useful for mapping of large-scale carbon stock.
Collapse
|
8
|
The Carbon Neutral Potential of Forests in the Yangtze River Economic Belt of China. FORESTS 2022. [DOI: 10.3390/f13050721] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/16/2022]
Abstract
Prediction of forest carbon sink in the future is important for understanding mechanisms concerning the increase in carbon sinks and emission reduction, and for realizing the climate goals of the Paris Agreement and global carbon neutrality. Based on stand volume data of permanent monitoring plots of the successive national forest inventories from 2004 to 2018, and combined with multiple variables, such as climatic factors, soil properties, stand attributes, and topographic features, the random forest algorithm was used to predict the stand volume growth-loss and then calculated the forest biomass and its carbon sink potential between 2015 to 2060 in the Yangtze River Economic Belt of China. From 2015 to 2060, the predicted forest biomass carbon storage and density increased from 3053.27 to 6721.61 Tg C and from 33.75 to 66.12 Mg C hm−2, respectively. The predicted forest biomass carbon sink decreased from 90.58 to 73.98 Tg C yr−1, and the average forest biomass carbon storage and sink were ranked in descending order: Yunnan, Sichuan, Jiangxi, Hunan, Guizhou, Hubei, Zhejiang, Chongqing, Anhui, Jiangsu, and Shanghai. The forest biomass carbon storage in the Yangtze River Economic Belt will increase by 3.67 Pg C from 2015 to 2060. The proportion of forest C sinks on the regional scale to C emissions on the national scale will increase from 2.9% in 2021–2030 to 4.3% in 2041–2050. These results indicate higher forest carbon sequestration efficiency in the Yangtze River Economic Belt in the future. Our results also suggest that improved forest management in the upper and middle reaches of the Yangtze River will help to enhance forest carbon sink in the future.
Collapse
|
9
|
Geng A, Tu Q, Chen J, Wang W, Yang H. Improving litterfall production prediction in China under variable environmental conditions using machine learning algorithms. JOURNAL OF ENVIRONMENTAL MANAGEMENT 2022; 306:114515. [PMID: 35063829 DOI: 10.1016/j.jenvman.2022.114515] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Subscribe] [Scholar Register] [Received: 07/17/2021] [Revised: 12/20/2021] [Accepted: 01/13/2022] [Indexed: 06/14/2023]
Abstract
Litterfall production is a major process within forest ecosystems that plays a crucial role in the global carbon cycle. Accordingly, studies have explored the abiotic and biotic features that influence litterfall production. In addition to traditional statistical models, the rapid development of nonparametric and nonlinear machine learning models, such as random forest (RF), light gradient boosting machine (LightGBM), and categorical boosting (CatBoost), have provided new methods of predicting the production of forest litterfall. Here, we evaluated the ability of the abovementioned models and mixed effect random forest (MERF) models to predict total annual litterfall production-based on several abiotic and biotic features-using 968 records from 314 forest sites covering the full geographical range of Chinese forests. In general, machine learning models were found to outperform linear mixed models. In particular, the MERF models ranked the highest in terms of performance (R2 = 0.7), which may be attributed to their ability to characterize nonlinear relationships between features and litterfall production. The key drivers were climate-related features and forest age, with the mean annual temperature and age positively correlated with litterfall production. Furthermore, the correlation between forest type and litterfall production was more significant for needleleaf forests than for other forest types. For needleleaf and broadleaf forests in several regions in China, the future litterfall production was predicted to be the highest under IPCC representative concentration pathway (RCP) 8.5, followed by RCP 4.5, RCP 2.6, and the original scenarios (sample data). Improved models to better understand and estimate litterfall production in forests at present and in the future are required for forest management planning to minimize the negative impacts of climate change on forest ecosystems.
Collapse
Affiliation(s)
- Aixin Geng
- College of Economics and Management, Nanjing Forestry University, Nanjing, 210037, China; Research Center for Economics and Trade in Forest Products of the State Forestry Administration, Nanjing, 210037, China
| | - Qingshi Tu
- Department of Wood Science, University of British Columbia, Vancouver, V6T 1Z4, Canada
| | - Jiaxin Chen
- Ontario Forest Research Institute, Ministry of Natural Resources and Forestry, 1235 Queen Street East, Sault Ste. Marie, Ontario, P6A 2E5, Canada.
| | - Weifeng Wang
- College of Biology and the Environment, Nanjing Forestry University, Nanjing, 210037, China
| | - Hongqiang Yang
- College of Economics and Management, Nanjing Forestry University, Nanjing, 210037, China; Research Center for Economics and Trade in Forest Products of the State Forestry Administration, Nanjing, 210037, China; Yangtze River Delta Economics and Social Development Research Center, Nanjing University, Nanjing, 210093, China.
| |
Collapse
|
10
|
Machine Learning for the Estimation of Diameter Increment in Mixed and Uneven-Aged Forests. SUSTAINABILITY 2022. [DOI: 10.3390/su14063386] [Citation(s) in RCA: 5] [Impact Index Per Article: 2.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 02/01/2023]
Abstract
Estimating the diameter increment of forests is one of the most important relationships in forest management and planning. The aim of this study was to provide insight into the application of two machine learning methods, i.e., the multilayer perceptron artificial neural network (MLP) and adaptive neuro-fuzzy inference system (ANFIS), for developing diameter increment models for the Hyrcanian forests. For this purpose, the diameters at breast height (DBH) of seven tree species were recorded during two inventory periods. The trees were divided into four broad species groups, including beech (Fagus orientalis), chestnut-leaved oak (Quercus castaneifolia), hornbeam (Carpinus betulus), and other species. For each group, a separate model was developed. The k-fold strategy was used to evaluate these models. The Pearson correlation coefficient (r), coefficient of determination (R2), root mean square error (RMSE), Akaike information criterion (AIC), and Bayesian information criterion (BIC) were utilized to evaluate the models. RMSE and R2 of the MLP and ANFIS models were estimated for the four groups of beech ((1.61 and 0.23) and (1.57 and 0.26)), hornbeam ((1.42 and 0.13) and (1.49 and 0.10)), chestnut-leaved oak ((1.55 and 0.28) and (1.47 and 0.39)), and other species ((1.44 and 0.32) and (1.5 and 0.24)), respectively. Despite the low coefficient of determination, the correlation test in both techniques was significant at a 0.01 level for all four groups. In this study, we also determined optimal network parameters such as number of nodes of one or multiple hidden layers and the type of membership functions for modeling the diameter increment in the Hyrcanian forests. Comparison of the results of the two techniques showed that for the groups of beech and chestnut-leaved oak, the ANFIS technique performed better and that the modeling techniques have a deep relationship with the nature of the tree species.
Collapse
|
11
|
On the Need to Further Refine Stock Quality Specifications to Improve Reforestation under Climatic Extremes. FORESTS 2022. [DOI: 10.3390/f13020168] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 12/10/2022]
Abstract
The achievement of goals in forest landscape restoration strongly relies on successful plantation establishment, which is challenging in drylands, especially under climate change. Improvement of field performance through stock quality has been used for decades. Here, we use machine learning (ML) techniques to identify key stock traits involved in successful survival and to refine previous specifications that were developed under more conventional stock quality assessments carried out at the lifting–shipping phases in the nursery. Two differentiated stocklots in each species were used, both fitting in the regional quality standard. ML was used to infer a set of attributes for planted seedlings that were subsequently related to survival at the short-term (two years) and mid-term (ten years) in six different species planted in a harsh site with shallow soil that suffered the driest year on record during this study. Whilst stocklot quality, as measured in the lifting–shipping stage, had very poor importance to the survival response, individual plant traits presented a moderate to high diagnostic ability for seedling survival (area under the receiver operating characteristic (ROC) curve between 0.59 and 0.99). Early growth traits catch most of the importance in these models (≈40%), followed by individual morphology traits (≈28%) and site variation (≈2%), with overall means varying across species. Aleppo pine and Phoenician juniper stocklots presented survival rates of 66–78% after ten years, and these rates were below 27% for the remaining species that suffered during the historical drought. In Aleppo pine, the plant attributes related to early field performance (growth in the first growing season) were more important in the drought-mediated mid-term performance than stock quality at the nursery stage. Within the technical framework of this study, our results allow for both testing and refining the regional quality standard specifications for harsh conditions such as those found in our study.
Collapse
|
12
|
The Effect of Stand Density, Biodiversity, and Spatial Structure on Stand Basal Area Increment in Natural Spruce-Fir-Broadleaf Mixed Forests. FORESTS 2022. [DOI: 10.3390/f13020162] [Citation(s) in RCA: 3] [Impact Index Per Article: 1.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 02/04/2023]
Abstract
Forest trees exhibit a large variation in the basal area increment (BAI), and the variation is attributed to the stand density, biodiversity, and stand spatial structure. Studying and quantifying the effect of these above variables on tree growth is vital for future forest management. However, the stand spatial structure based on neighboring trees has rarely been considered, especially in the mixed forests. This study adopted the random-forest (RF) algorithm to model and interpret BAI based on stand density, biodiversity, and spatial structure. Fourteen independent variables, including two stand density predictors, four biodiversity predictors, and eight spatial structure predictors, were evaluated. The RF model was trained for the whole stand, three tree species groups (gap, neutral, and shade_tolerant), and two tree species (spruce and fir). A 10-fold blocked cross-validation was then used to optimize the hyper-parameters and evaluate the models. The squared correlation coefficients (R2) for the six groups were 0.233 for the whole stand, 0.575 for fir, 0.609 for shade_tolerant, 0.622 for neutral, 0.722 for gap, and 0.730 for spruce. The Stand Density Index (SDI) was the most-important predictor, suggesting that BAI is primarily restricted by competition. BAI and species biodiversity were positively correlated for the whole stand. The stands were expected to be randomly distributed based on the relationship between the uniform angle index (W) and growth. The relationship between dominance (U) and BAI indicated that small trees should be planted around the light-demanding tree species and vice versa. Of note, these findings emphasize the need to consider the three types of variables in mixed forests, especially the spatial structure factors. This study may help make significant advances in species composition, spatial arrangement, and the sustainable development of mixed forests.
Collapse
|
13
|
A Crown Contour Envelope Model of Chinese Fir Based on Random Forest and Mathematical Modeling. FORESTS 2020. [DOI: 10.3390/f12010048] [Citation(s) in RCA: 3] [Impact Index Per Article: 0.8] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/16/2022]
Abstract
The tree crown is an important part of a tree and is closely related to forest growth status, forest canopy density, and other forest growth indicators. Chinese fir (Cunninghamia lanceolata (Lamb.) Hook) is an important tree species in southern China. A three-dimensional (3D) visualization assistant decision-making system of plantations could be improved through the construction of crown contour envelope models (CCEMs), which could aid plantation production. The goal of this study was to establish CCEMs, based on random forest and mathematical modeling, and to compare them. First, the regression equation of a tree crown was calculated using the least squares method. Then, forest characteristic factors were screened using methods based on mutual information, recursive feature elimination, least absolute shrink and selection operator, and random forest, and the random forest model was established based on the different screening results. The accuracy of the random forest model was higher than that of the mathematical modeling. The best performing model based on mathematical modeling was the quartic polynomial with the largest crown radius as the variable (R-squared (R2) = 0.8614 and root mean square error (RMSE) = 0.2657). Among the random forest regression models, the regression model constructed using mutual information as the feature screening method was the most accurate (R2 = 0.886, RMSE = 0.2406), which was two percentage points higher than mathematical modeling. Compared with mathematical modeling, the random forest model can reflect the differences among trees and aid 3D visualization of a Chinese fir plantation.
Collapse
|
14
|
Comparison of Spatially and Nonspatially Explicit Nonlinear Mixed Effects Models for Norway Spruce Individual Tree Growth under Single-Tree Selection. FORESTS 2020. [DOI: 10.3390/f11121338] [Citation(s) in RCA: 3] [Impact Index Per Article: 0.8] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/17/2022]
Abstract
Background and Objectives: Continuous cover forestry is of increasing importance, but operational forest growth models are still lacking. The debate is especially open if more complex spatial approaches would provide a worthwhile increase in accuracy. Our objective was to compare a nonspatial versus a spatial approach for individual Norway spruce tree growth models under single-tree selection cutting. Materials and Methods: We calibrated nonlinear mixed models using data from a long-term experiment in Finland (20 stands with 3538 individual trees for 10,238 growth measurements). We compared the use of nonspatial versus spatial predictors to describe the competitive pressure and its release after cutting. The models were compared in terms of Akaike Information Criteria (AIC), root mean square error (RMSE), and mean absolute bias (MAB), both with the training data and after cross-validation with a leave-one-out method at stand level. Results: Even though the spatial model had a lower AIC than the nonspatial model, RMSE and MAB of the two models were similar. Both models tended to underpredict growth for the highest observed values when the tree-level random effects were not used. After cross-validation, the aggregated predictions at stand level well represented the observations in both models. For most of the predictors, the use of values based on trees’ height rather than trees’ diameter improved the fit. After single-tree selection cutting, trees had a growth boost both in the first and second five-year period after cutting, however, with different predicted intensity in the two models. Conclusions: Under the research framework here considered, the spatial modeling approach was not more accurate than the nonspatial one. Regarding the single-tree selection cutting, an intervention regime spaced no more than 15 years apart seems necessary to sustain the individual tree growth. However, the model’s fixed effect parts were not able to capture the high growth of the few fastest-growing trees, and a proper estimation of site potential is needed for uneven-aged stands.
Collapse
|
15
|
Leaf Area Index Estimation Algorithm for GF-5 Hyperspectral Data Based on Different Feature Selection and Machine Learning Methods. REMOTE SENSING 2020. [DOI: 10.3390/rs12132110] [Citation(s) in RCA: 17] [Impact Index Per Article: 4.3] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/16/2022]
Abstract
Leaf area index (LAI) is an essential vegetation parameter that represents the light energy utilization and vegetation canopy structure. As the only in-operation hyperspectral satellite launched by China, GF-5 is potentially useful for accurate LAI estimation. However, there is no research focus on evaluating GF-5 data for LAI estimation. Hyperspectral remote sensing data contains abundant information about the reflective characteristics of vegetation canopies, but these abound data also easily result in a dimensionality curse. Therefore, feature selection (FS) is necessary to reduce data redundancy to achieve more reliable estimations. Currently, machine learning (ML) algorithms have been widely used for FS. Moreover, the same ML algorithm is usually conducted for both FS and regression in LAI estimation. However, no evidence suggests that this is the optimal solution. Therefore, this study focuses on evaluating the capacity of GF-5 spectral reflectance for estimating LAI and the performances of different combination of FS and ML algorithms. Firstly, the PROSAIL model, which coupled leaf optical properties model PROSPECT and the scattering by arbitrarily inclined leaves (SAIL) model, was used to generate simulated GF-5 reflectance data under different vegetation and soil conditions, and then three FS methods, including random forest (RF), K-means clustering (K-means) and mean impact value (MIV), and three ML algorithms, including random forest regression (RFR), back propagation neural network (BPNN) and K-nearest neighbor (KNN) were used to develop nine LAI estimation models. The FS process was conducted twice using different strategies: Firstly, three FS methods were conducted to search the lowest dimension number, which maintained the estimation accuracy of all bands. Then, the sequential backward selection (SBS) method was used to eliminate the bands having minimal impact on LAI estimation accuracy. Finally, three best estimation models were selected and evaluated using reference LAI. The results showed that although the RF_RFR model (RF used for feature selection and RFR used for regression) achieved reliable LAI estimates (coefficient of determination (R2) = 0.828, root mean square error (RMSE) = 0.839), the poor performance (R2 = 0.763, RMSE = 0.987) of the MIV_BPNN model (MIV used for feature selection and BPNN used for regression) suggested using feature selection and regression conducted by the same ML algorithm could not always ensure an optimal estimation. Moreover, RF selection preserved the most informative bands for LAI estimation so that each ML regression method could achieve satisfactory estimation results. Finally, the results indicated that the RF_KNN model (RF used as feature selection and KNN used for regression) with seven GF-5 spectral band reflectance achieved the better estimation results than others when validated by simulated data (R2 = 0.834, RMSE = 0.824) and actual reference LAI (R2 = 0.659, RMSE = 0.697).
Collapse
|
16
|
Estimation of Tree Heights in an Uneven-Aged, Mixed Forest in Northern Iran Using Artificial Intelligence and Empirical Models. FORESTS 2020. [DOI: 10.3390/f11030324] [Citation(s) in RCA: 19] [Impact Index Per Article: 4.8] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/17/2022]
Abstract
The diameters and heights of trees are two of the most important components in a forest inventory. In some circumstances, the heights of trees need to be estimated due to the time and cost involved in measuring them in the field. Artificial intelligence models have many advantages in modeling nonlinear height–diameter relationships of trees, which sometimes make them more useful than empirical models in estimating the heights of trees. In the present study, the heights of trees in uneven-aged and mixed stands in the high elevation forests of northern Iran were estimated using an artificial neural network (ANN) model, an adaptive neuro-fuzzy inference system (ANFIS) model, and empirical models. A systematic sampling method with a 150 × 200 m network (0.1 ha area) was employed. The diameters and heights of 516 trees were measured to support the modeling effort. Using 10 nonlinear empirical models, the ANN model, and the ANFIS model, the relationship between height as a dependent variable and diameter as an independent variable was analyzed. The results show, according to R2, relative root mean square error (RMSE), and other model evaluation criteria, that there is a greater consistency between predicted height and observed height when using artificial intelligence models (R2 = 0.78; RMSE (%) = 18.49) than when using regression analysis (R2 = 0.68; RMSE (%) = 17.69). Thus, it can be said that these models may be better than empirical models for predicting the heights of common, commercially-important trees in the study area.
Collapse
|
17
|
|
18
|
Estimating Forest Stock Volume in Hunan Province, China, by Integrating In Situ Plot Data, Sentinel-2 Images, and Linear and Machine Learning Regression Models. REMOTE SENSING 2020. [DOI: 10.3390/rs12010186] [Citation(s) in RCA: 28] [Impact Index Per Article: 7.0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/16/2022]
Abstract
The forest stock volume (FSV) is one of the key indicators in forestry resource assessments on local, regional, and national scales. To date, scaling up in situ plot-scale measurements across landscapes is still a great challenge in the estimation of FSVs. In this study, Sentinel-2 imagery, the Google Earth Engine (GEE) cloud computing platform, three base station joint differential positioning technology (TBSJDPT), and three algorithms were used to build an FSV model for forests located in Hunan Province, southern China. The GEE cloud computing platform was used to extract the imagery variables from the Sentinel-2 imagery pixels. The TBSJDPT was put forward and used to provide high-precision positions of the sample plot data. The random forests (RF), support vector regression (SVR), and multiple linear regression (MLR) algorithms were used to estimate the FSV. For each pixel, 24 variables were extracted from the Sentinel-2 images taken in 2017 and 2018. The RF model performed the best in both the training phase (i.e., R2 = 0.91, RMSE = 35.13 m3 ha−1, n = 321) and in the test phase (i.e., R2 = 0.58, RMSE = 65.03 m3 ha−1, and n = 138). This model was followed by the SVR model (R2 = 0.54, RMSE = 65.60 m3 ha−1, n = 321 in training; R2 = 0.54, RMSE = 66.00 m3 ha−1, n = 138 in testing), which was slightly better than the MLR model (R2 = 0.38, RMSE = 75.74 m3 ha−1, and n = 321 in training; R2 = 0.49, RMSE = 70.22 m3 ha−1, and n = 138 in testing) in both the training phase and test phase. The best predictive band was Red-Edge 1 (B5), which performed well both in the machine learning methods and in the MLR method. The Blue band (B2), Green band (B3), Red band (B4), SWIR2 band (B12), and vegetation indices (TCW, NDVI_B5, and TCB) were used in the machine learning models, and only one vegetation index (MSI) was used in the MLR model. We mapped the FSV distribution in Hunan Province (3.50 × 108 m3) based on the RF model; it reached a total accuracy of 63.87% compared with the official forest report in 2017 (5.48 × 108 m3). The results from this study will help develop and improve satellite-based methods to estimate FSVs on local, regional and national scales.
Collapse
|