Comparison of Four Machine Learning Methods for Generating the GLASS Fractional Vegetation Cover Product from MODIS Data

Yang, Linqing; Jia, Kun; Liang, Shunlin; Liu, Jingcan; Wang, Xiaoxia

doi:10.3390/rs8080682

Cited by 63 publications

(51 citation statements)

References 36 publications

(59 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…RF has broader scope than SVM and NN, because it has no limitation on the distribution pattern of the training data. The complexity of the parameter setting may limit the performance of SVM and NN for model training [118]. Although, SVM has the advantage on solving non-linear problems, a major downside of SVM is that it can be painfully inefficient to train [119,120].…”

Section: Integration Of Various Remote Sensing Metricsmentioning

confidence: 99%

Developing an Integrated Remote Sensing Based Biodiversity Index for Predicting Animal Species Richness

Liang

2018

Remote Sensing

Self Cite

View full text Add to dashboard Cite

Many remote sensing metrics have been applied in large-scale animal species monitoring and conservation. However, the capabilities of these metrics have not been well compared and assessed. In this study, we investigated the correlation of 21 remote sensing metrics in three categories with the global species richness of three different animal classes using several statistical methods. As a result, we developed a new index by integrating several highly correlated metrics. Of the 21 remote sensing metrics analyzed, evapotranspiration (ET) had the greatest impact on species richness on a global scale (explained variance: 52%). The metrics with a high explained variance on the global scale were mainly in the energy/productivity category. The metrics in the texture category exhibited higher correlation with species richness at regional scales. We found that radiance and temperature had a larger impact on the distribution of bird richness, compared to their impacts on the distributions of both amphibians and mammals. Three machine learning models (i.e., support vector machine, random forests, and neural networks) were evaluated for metric integration, and the random forest model showed the best performance. Our newly developed index exhibited a 0.7 explained variance for the three animal classes' species richness on a global scale, with an explained variance that was 20% higher than any of the univariate metrics.

show abstract

Section: Integration Of Various Remote Sensing Metricsmentioning

confidence: 99%

Developing an Integrated Remote Sensing Based Biodiversity Index for Predicting Animal Species Richness

Liang

2018

Remote Sensing

Self Cite

View full text Add to dashboard Cite

show abstract

“…The comparison between the satellite FVC and UAS FVC gave interesting results. The satellite FVC was based on the neural network approach, which generally tends to show a high accuracy [21,45]. However, from our observation, the lower boundary of the FVC was overestimated and the higher FVC was underestimated.…”

Section: Discussionmentioning

confidence: 64%

“…The remote sensing techniques for FVC development utilize the multispectral information observed from space and validates its product with ground truth information (e.g., field surveys). The estimation methods can vary depending on the model type used, including simple empirical models [18], linear spectral models [19], decision tree method [20], machine learning techniques [21], and so on. Although the input information is rather simple, remote sensing can use various spectral information or the computed vegetation indices (Vis) for its estimation, although correctly delineating the FVC for various regions of the world is still a challenge.…”

Section: Introductionmentioning

confidence: 99%

Estimating and Examining the Sensitivity of Different Vegetation Indices to Fractions of Vegetation Cover at Different Scaling Grids for Early Stage Acacia Plantation Forests Using a Fixed-Wing UAS

Iizuka

Kato²,

Silsigia³

et al. 2019

Remote Sensing

View full text Add to dashboard Cite

Understanding the information on land conditions and especially green vegetation cover is important for monitoring ecosystem dynamics. The fraction of vegetation cover (FVC) is a key variable that can be used to observe vegetation cover trends. Conventionally, satellite data are utilized to compute these variables, although computations in regions such as the tropics can limit the amount of available observation information due to frequent cloud coverage. Unmanned aerial systems (UASs) have become increasingly prominent in recent research and can remotely sense using the same methods as satellites but at a lower altitude. UASs are not limited by clouds and have a much higher resolution. This study utilizes a UAS to determine the emerging trends for FVC estimates at an industrial plantation site in Indonesia, which utilizes fast-growing Acacia trees that can rapidly change the land conditions. First, the UAS was utilized to collect high-resolution RGB imagery and multispectral images for the study area. The data were used to develop general land use/land cover (LULC) information for the site. Multispectral data were converted to various vegetation indices, and within the determined resolution grid (5, 10, 30 and 60 m), the fraction of each LULC type was analyzed for its correlation between the different vegetation indices (Vis). Finally, a simple empirical model was developed to estimate the FVC from the UAS data. The results show the correlation between the FVC (acacias) and different Vis ranging from R 2 = 0.66-0.74, 0.76-0.8, 0.84-0.89 and 0.93-0.94 for 5, 10, 30 and 60 m grid resolutions, respectively. This study indicates that UAS-based FVC estimations can be used for observing fast-growing acacia trees at a fine scale resolution, which may assist current restoration programs in Indonesia.

show abstract

“…MARS is a nonparametric and multivariate regression analysis model and has been demonstrated to obtain satisfactory FVC results from MODIS reflectance data [29,56]. Without strong assumptions, MARS is capable of modeling complex nonlinear relationships among variables by fitting piecewise linear regressions.…”

Section: The Mars Modelmentioning

confidence: 99%

“…However, in terms of FVC products, the current FVC products were obtained mainly from low-or medium-resolution remote sensing data such as SPOT-VGT, SEAWIFS, MERIS, MODIS and AVHRR data [1,22,[28][29][30], which limits the FVC applications to the regional and local scales [31]. The development of FVC products from decametric spatial resolution sensors will be better for addressing these applications closely related to agriculture, ecosystem and environmental management.…”

Section: Introductionmentioning

confidence: 99%

A Robust Algorithm for Estimating Surface Fractional Vegetation Cover from Landsat Data

et al. 2017

Self Cite

View full text Add to dashboard Cite

Abstract:Fractional vegetation cover (FVC) is an essential land surface parameter for Earth surface process simulations and global change studies. The currently existing FVC products are mostly obtained from low or medium resolution remotely sensed data, while many applications require the fine spatial resolution FVC product. The availability of well-calibrated coverage of Landsat imagery over large areas offers an opportunity for the production of FVC at fine spatial resolution. Therefore, the objective of this study is to develop a general and reliable land surface FVC estimation algorithm for Landsat surface reflectance data under various land surface conditions. Two machine learning methods multivariate adaptive regression splines (MARS) model and back-propagation neural networks (BPNNs) were trained using samples from PROSPECT leaf optical properties model and the scattering by arbitrarily inclined leaves (SAIL) model simulations, which included Landsat reflectance and corresponding FVC values, and evaluated to choose the method which had better performance. Thereafter, the MARS model, which had better performance in the independent validation, was evaluated using ground FVC measurements from two case study areas. The direct validation of the FVC estimated using the proposed algorithm (Heihe: R 2 = 0.8825, RMSE = 0.097; Chengde using Landsat 7 ETM+: R 2 = 0.8571, RMSE = 0.078, Chengde using Landsat 8 OLI: R 2 = 0.8598, RMSE = 0.078) showed the proposed method had good performance. Spatial-temporal assessment of the estimated FVC from Landsat 7 ETM+ and Landsat 8 OLI data confirmed the robustness and consistency of the proposed method. All these results indicated that the proposed algorithm could obtain satisfactory accuracy and had the potential for the production of high-quality FVC estimates from Landsat surface reflectance data.

show abstract

Comparison of Four Machine Learning Methods for Generating the GLASS Fractional Vegetation Cover Product from MODIS Data

Cited by 63 publications

References 36 publications

Developing an Integrated Remote Sensing Based Biodiversity Index for Predicting Animal Species Richness

Developing an Integrated Remote Sensing Based Biodiversity Index for Predicting Animal Species Richness

Estimating and Examining the Sensitivity of Different Vegetation Indices to Fractions of Vegetation Cover at Different Scaling Grids for Early Stage Acacia Plantation Forests Using a Fixed-Wing UAS

A Robust Algorithm for Estimating Surface Fractional Vegetation Cover from Landsat Data

Contact Info

Product

Resources

About