Gaussian Process Regression Model for Crop Biophysical Parameter Retrieval from Multi-Polarized C-Band SAR Data

Ghosh, Swarnendu Sekhar; Dey, Subhadip; Bhogapurapu, Narayanarao; Homayouni, Saeid; Bhattacharya, Avik; McNairn, Heather

doi:10.3390/rs14040934

Cited by 16 publications

(7 citation statements)

References 58 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In comparison to other non-linear prediction algorithms, such as SVM or kernel ridge regression, GP offer the possibility to automatically tune their hyper-parameters θ. GP also provide the variance of the point-wise estimation. These properties made GP for regression widely used by the remote sensing community in the last decade [61]- [64].…”

Section: A Univariate Gaussian Processesmentioning

confidence: 99%

Land Cover Classification With Gaussian Processes Using Spatio-Spectro-Temporal Features

Bellet

Fauvel

Inglada

2023

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

In this article, we propose an approach based on Gaussian Processes (GP) for large scale land cover pixel-based classification with Sentinel-2 satellite image time-series (SITS). We used a sparse approximation of the posterior combined with variational inference to learn the GP's parameters. We applied stochastic gradient descent and GPU computing to optimize our GP models on massive data sets. The proposed GP model can be trained with hundreds of thousands of samples, compared to few thousands for traditional GP methods. Moreover, we included the spatial information by adding the geographic coordinates into the GP's covariance function to efficiently exploit the spatiospectro-temporal structure of the SITS. We ran experiments with Sentinel-2 SITS of the full year 2018 over an area of 200 000 km 2 (about 2 billion pixels) in the south of France, which is representative of an operational setting. Adding the spatial information significantly improved the results in terms of classification accuracy. With spatial information, GP models have an overall accuracy of 79.8. They are more than three points above Random Forest (the method used for current operational systems) and more than one point above a multi-layer perceptron. Compared to a Transformer-based model (which provides state of the art results in the literature, but are not applied in operational systems), GP models are only one point below.

show abstract

Section: A Univariate Gaussian Processesmentioning

confidence: 99%

Land Cover Classification With Gaussian Processes Using Spatio-Spectro-Temporal Features

Bellet

Fauvel

Inglada

2023

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Furthermore, the result showed that the GPR model has a lower prediction accuracy with soil and UAV imagery derived from data fusion compared with the UAV vegetation indices scenario. In contrast, previous research showed that hyperspectral UAV and soil data fusion improve GPR modelling precision while providing more accurate results with vegetation indices for estimating wheat above-ground biomass [42,115]. The improved performance of GPR can be linked to its use of kernel functions when dealing with input [46,47,82].…”

Section: Discussionmentioning

confidence: 94%

Field-Scale Winter Wheat Growth Prediction Applying Machine Learning Methods with Unmanned Aerial Vehicle Imagery and Soil Properties

Nduku,

Munghemezulu,

Mashaba-Munghemezulu

et al. 2024

Land

View full text Add to dashboard Cite

Monitoring crop growth conditions during the growing season provides information on available soil nutrients and crop health status, which are important for agricultural management practices. Crop growth frequently varies due to site-specific climate and farm management practices. These variations might arise from sub-field-scale heterogeneities in soil composition, moisture levels, sunlight, and diseases. Therefore, soil properties and crop biophysical data are useful to predict field-scale crop development. This study investigates soil data and spectral indices derived from multispectral Unmanned Aerial Vehicle (UAV) imagery to predict crop height at two winter wheat farms. The datasets were investigated using Gaussian Process Regression (GPR), Ensemble Regression (ER), Decision tree (DT), and Support Vector Machine (SVM) machine learning regression algorithms. The findings showed that GPR (R2 = 0.69 to 0.74, RMSE = 15.95 to 17.91 cm) has superior accuracy in all models when using vegetation indices (VIs) to predict crop growth for both wheat farms. Furthermore, the variable importance generated using the GRP model showed that the RedEdge Normalized Difference Vegetation Index (RENDVI) had the most influence in predicting wheat crop height compared to the other predictor variables. The clay, calcium (Ca), magnesium (Mg), and potassium (K) soil properties have a moderate positive correlation with crop height. The findings from this study showed that the integration of vegetation indices and soil properties predicts crop height accurately. However, using the vegetation indices independently was more accurate at predicting crop height. The outcomes from this study are beneficial for improving agronomic management within the season based on crop height trends. Hence, farmers can focus on using cost-effective VIs for monitoring particular areas experiencing crop stress.

show abstract

“…The covariance between any two points in the input space determines the similarity between those points and is used to make predictions about the output variable. The hyper parameters of the Gaussian process, such as the length scale and amplitude, control the smoothness and variability of the functions in the prior distribution [38]. The posterior distribution over functions is obtained by conditioning on the observed data and is also a Gaussian process with updated mean and covariance functions.…”

Section: Gaussian Process Regression (Gpr) Modelmentioning

confidence: 99%

Prediction of Greenhouse Indoor Air Temperature Using Artificial Intelligence (AI) Combined with Sensitivity Analysis

Hosseini Monjezi,

Taki,

Abdanan Mehdizadeh

et al. 2023

Horticulturae

View full text Add to dashboard Cite

Greenhouses are essential for agricultural production in unfavorable climates. Accurate temperature predictions are critical for controlling Heating, Ventilation, Air-Conditioning, and Dehumidification (HVACD) and lighting systems to optimize plant growth and reduce financial losses. In this study, several machine models were employed to predict indoor air temperature in an even-span Mediterranean greenhouse. Radial Basis Function (RBF), Support Vector Machine (SVM), and Gaussian Process Regression (GPR) were applied using external parameters such as outside air, relative humidity, wind speed, and solar radiation. The results showed that an RBF model with the LM learning algorithm outperformed the SVM and GPR models. The RBF model had high accuracy and reliability with an RMSE of 0.82 °C, MAPE of 1.21%, TSSE of 474.07 °C, and EF of 1.00. Accurate temperature prediction can help farmers manage their crops and resources efficiently and reduce energy inefficiencies and lower yields. The integration of the RBF model into greenhouse control systems can lead to significant energy savings and cost reductions.

show abstract

Gaussian Process Regression Model for Crop Biophysical Parameter Retrieval from Multi-Polarized C-Band SAR Data

Cited by 16 publications

References 58 publications

Land Cover Classification With Gaussian Processes Using Spatio-Spectro-Temporal Features

Land Cover Classification With Gaussian Processes Using Spatio-Spectro-Temporal Features

Field-Scale Winter Wheat Growth Prediction Applying Machine Learning Methods with Unmanned Aerial Vehicle Imagery and Soil Properties

Prediction of Greenhouse Indoor Air Temperature Using Artificial Intelligence (AI) Combined with Sensitivity Analysis

Contact Info

Product

Resources

About