Dataset Information

Estimation of paddy rice leaf area index using machine learning methods based on hyperspectral data from multi-year experiments.

ABSTRACT: The performance of three machine learning methods (support vector regression, random forests and artificial neural network) for estimating the LAI of paddy rice was evaluated in this study. Traditional univariate regression models involving narrowband NDVI with optimized band combinations as well as linear multivariate calibration partial least squares regression models were also evaluated for comparison. A four year field-collected dataset was used to test the robustness of LAI estimation models against temporal variation. The partial least squares regression and three machine learning methods were built on the raw hyperspectral reflectance and the first derivative separately. Two different rules were used to determine the models' key parameters. The results showed that the combination of the red edge and NIR bands (766 nm and 830 nm) as well as the combination of SWIR bands (1114 nm and 1190 nm) were optimal for producing the narrowband NDVI. The models built on the first derivative spectra yielded more accurate results than the corresponding models built on the raw spectra. Properly selected model parameters resulted in comparable accuracy and robustness with the empirical optimal parameter and significantly reduced the model complexity. The machine learning methods were more accurate and robust than the VI methods and partial least squares regression. When validating the calibrated models against the standalone validation dataset, the VI method yielded a validation RMSE value of 1.17 for NDVI(766,830) and 1.01 for NDVI(1114,1190), while the best models for the partial least squares, support vector machine and artificial neural network methods yielded validation RMSE values of 0.84, 0.82, 0.67 and 0.84, respectively. The RF models built on the first derivative spectra with mtry = 10 showed the highest potential for estimating the LAI of paddy rice.

SUBMITTER: Wang L

PROVIDER: S-EPMC6281281 | biostudies-literature | 2018

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Estimation of paddy rice leaf area index using machine learning methods based on hyperspectral data from multi-year experiments.

Wang Li L Chang Qingrui Q Yang Jing J Zhang Xiaohua X Li Fenling F

PloS one 20181205 12

The performance of three machine learning methods (support vector regression, random forests and artificial neural network) for estimating the LAI of paddy rice was evaluated in this study. Traditional univariate regression models involving narrowband NDVI with optimized band combinations as well as linear multivariate calibration partial least squares regression models were also evaluated for comparison. A four year field-collected dataset was used to test the robustness of LAI estimation model ...[more]

PMID: 30517144

Similar Datasets

Project description:Accurate estimations of the vertical leaf nitrogen (N) distribution within a rice canopy is helpful for understanding the nutrient supply and demand of various functional leaf layers of rice and for improving the predictions of rice productivity. A two-year field experiment using different rice varieties, N rates, and planting densities was performed to investigate the vertical distribution of the leaf nitrogen concentration (LNC, %) within the rice canopy, the relationship between the LNC in different leaf layers (LNCLi, i = 1, 2, 3, 4), and the relationship between the LNCLi and the LNC at the canopy level (LNCCanopy). A vertical distribution model of the LNC was constructed based on the relative canopy height. Furthermore, the relationship between different vegetation indices (VIs) and the LNCCanopy, the LNCLi, and the LNC vertical distribution model parameters were studied. We also compared the following three methods for estimating the LNC in different leaf layers in rice canopy: (1) estimating the LNCCanopy by VIs and then estimating the LNCLi based on the relationship between the LNCLi and LNCCanopy; (2) estimating the LNC in any leaf layer of the rice canopy by VIs, inputting the result into the LNC vertical distribution model to obtain the parameters of the model, and then estimating the LNCLi using the LNC vertical distribution model; (3) estimating the model parameters by using VIs directly and then estimating the LNCLi by the LNC vertical distribution model. The results showed that the LNC in the bottom of rice canopy was more susceptible to different N rates, and changes in the LNC with the relative canopy height could be simulated by an exponential model. Vegetation indices could estimate the LNC at the top of rice canopy. R705/(R717+R491) (R2 = 0.763) and the renormalized difference vegetation index (RDVI) (1340, 730) (R2 = 0.747) were able to estimate the parameter "a" of the LNC vertical distribution model in indica rice and japonica rice, respectively. In addition, method (2) was the best choice for estimating the LNCLi (R2 = 0.768, 0.700, 0.623, and 0.549 for LNCL1, LNCL2, LNCL3, and LNCL4, respectively). These results provide technical support for the rapid, accurate, and non-destructive identification of the vertical distribution of nitrogen in rice canopies.

Project description:The growth status of winter wheat in irrigated field and rainfed field are obviously different and the field types may have an effect on the predictive accuracy of hyperspectral model. The objectives of the present study were to understand the difference of spectral sensitive wavelengths for leaf area index (LAI) in two field types and realize its hyperspectral prediction. In study, a total of 31 and 28 sample sites in irrigated fields and rainfed fields respectively were selected from Wenxi County, and the LAI and canopy spectra were also collected at the main grow stage of winter wheat. The method of successive projections algorithm (SPA) was applied by selecting the important wavelengths, and the multiple linear regression (MLR) and partial least squares regression (PLSR) were used to construct the predictive model based on the important wavelengths and full wavelengths, respectively. Moreover, the parameters of variable importance project (VIP) and B-coefficient derived from PLSR analysis were implemented to validate the evaluated wavelengths using the SPA method. The sensitive wavelengths of LAI for irrigated field and rainfed field were 404, 407, 413, 417, 450, 677, 715, 735, 816, 1127 and 404, 406, 432, 501, 540, 679, 727, 779, 1120, 1290 nm, respectively, and these wavelengths proved to be highly correlated with LAI. Compared with the model performance based on the SPA-MLR and PLSR methods, the method of SPA-MLR was proved to be better (rainfed field: R2 = 0.736, RMSE = 1.169, RPD = 1.6245; irrigated field: R2 = 0.716, RMSE = 1.059, RPD = 1.538). Moreover, the predictive model of LAI in rainfed fields had a better accuracy than the model in irrigated fields. The results from this study indicated that it was necessary to classify the field type while monitoring the winter wheat using the remote sensing technology. This study also demonstrated that the multivariate method of SPA-MLR could accurately evaluate the sensitive wavelengths and construct the predictive model of LAI.

Project description:Timely monitoring nitrogen status of rice crops with remote sensing can help us optimize nitrogen fertilizer management and reduce environmental pollution. Recently, the use of near-surface imaging spectroscopy is emerging as a promising technology that can collect hyperspectral images with spatial resolutions ranging from millimeters to decimeters. The spatial resolution is crucial for the efficiency in the image sampling across rice plants and the separation of leaf signals from the background. However, the optimal spatial resolution of such images for monitoring the leaf nitrogen concentration (LNC) in rice crops remains unclear. To assess the impact of spatial resolution on the estimation of rice LNC, we collected ground-based hyperspectral images throughout the entire growing season over 2 consecutive years and generated ten sets of images with spatial resolutions ranging from 1.3 to 450 mm. These images were used to determine the sensitivity of LNC prediction to spatial resolution with three groups of vegetation indices (VIs) and two multivariate methods Gaussian Process regression (GPR) and Partial least squares regression (PLSR). The reflectance spectra of sunlit-, shaded-, and all-leaf leaf pixels separated from background pixels at each spatial resolution were used to predict LNC with VIs, GPR and PLSR, respectively. The results demonstrated all-leaf pixels generally exhibited more stable performance than sunlit- and shaded-leaf pixels regardless of estimation approaches. The predictions of LNC required stage-specific LNC~VI models for each vegetative stage but could be performed with a single model for all the reproductive stages. Specifically, most VIs achieved stable performances from all the resolutions finer than 14 mm for the early tillering stage but from all the resolutions finer than 56 mm for the other stages. In contrast, the global models for the prediction of LNC across the entire growing season were successfully established with the approaches of GPR or PLSR. In particular, GPR generally exhibited the best prediction of LNC with the optimal spatial resolution being found at 28 mm. These findings represent significant advances in the application of ground-based imaging spectroscopy as a promising approach to crop monitoring and understanding the effects of spatial resolution on the estimation of rice LNC.

Dataset Information

Estimation of paddy rice leaf area index using machine learning methods based on hyperspectral data from multi-year experiments.

Publications

Estimation of paddy rice leaf area index using machine learning methods based on hyperspectral data from multi-year experiments.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets