Artificial neural networks for mapping regional‐scale upland vegetation from high spatial resolution imagery

Mills, Henny; Cutler, Mark E. J.; Fairbairn, David

doi:10.1080/01431160500396501

Cited by 23 publications

(10 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…), although acquisitions around September are considered optimum as most upland vegetation types are fully developed (Mills et al. ). The spectral similarity between different vegetation types during the summer often limits the ability of acquisitions during these months to reliably distinguish between vegetation types.…”

Section: Discussionmentioning

confidence: 99%

“…In this regard, the time of year of image acquisition will have a strong bearing on the classification accuracy and the ability to distinguish different types of vegetation. Nonetheless, it is difficult to identify an optimal temporal window for operational monitoring of all upland vegetation types (Cole et al 2014), although acquisitions around September are considered optimum as most upland vegetation types are fully developed (Mills et al 2006). The spectral similarity between different vegetation types during the summer often limits the ability of acquisitions during these months to reliably distinguish between vegetation types.…”

Section: Eo Data Acquisition Timing and Spatial Resolutionmentioning

confidence: 99%

See 1 more Smart Citation

Upland vegetation mapping using Random Forests with optical and radar satellite data

Barrett

Raab

Cawkwell

et al. 2016

Remote Sens Ecol Conserv

View full text Add to dashboard Cite

Uplands represent unique landscapes that provide a range of vital benefits to society, but are under increasing pressure from the management needs of a diverse number of stakeholders (e.g. farmers, conservationists, foresters, government agencies and recreational users). Mapping the spatial distribution of upland vegetation could benefit management and conservation programmes and allow for the impacts of environmental change (natural and anthropogenic) in these areas to be reliably estimated. The aim of this study was to evaluate the use of medium spatial resolution optical and radar satellite data, together with ancillary soil and topographic data, for identifying and mapping upland vegetation using the Random Forests (RF) algorithm. Intensive field survey data collected at three study sites in Ireland as part of the National Parks and Wildlife Service (NPWS) funded survey of upland habitats was used in the calibration and validation of different RF models. Eight different datasets were analysed for each site to compare the change in classification accuracy depending on the input variables. The overall accuracy values varied from 59.8% to 94.3% across the three study locations and the inclusion of ancillary datasets containing information on the soil and elevation further improved the classification accuracies (between 5 and 27%, depending on the input classification dataset). The classification results were consistent across the three different study areas, confirming the applicability of the approach under different environmental contexts.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Eo Data Acquisition Timing and Spatial Resolutionmentioning

confidence: 99%

Upland vegetation mapping using Random Forests with optical and radar satellite data

Barrett

Raab

Cawkwell

et al. 2016

Remote Sens Ecol Conserv

View full text Add to dashboard Cite

show abstract

“…(2007) used backpropagation to train NNs with AVHRR data to accurately estimate cloud parameters. Mills et al. (2006) used a multi‐layer perceptron NN and Ikonos satellite data to map regional vegetation patterns.…”

Section: Artificial Neural Network Studies In Remote Sensingmentioning

confidence: 99%

Artificial Neural Networks and Remote Sensing

Jensen

Hardin

2009

Geography Compass

View full text Add to dashboard Cite

Accurate land cover classifications and biophysical estimations derived from remotely sensed data are important to generate map products and provide information about the Earth's changing surface. Statistical classifiers are often used to generate many of these data, but these classifiers rely on assumptions that may limit their utility for many datasets. Conversely, Artificial Neural Networks (ANNs) provide an accurate way for researchers to classify land cover and estimate biophysical properties of earthly phenomena without having to rely on statistical procedures or assumptions. This article describes the role of ANNs in remote sensing from their inception and contrasts ANNs with statistical remote sensing classifiers. We conclude that ANNs will continue to play a crucial role in remote sensing because of their many benefits.

show abstract

“…The basic methodology adopted was to use artificial neural networks as the predictors on a pixel by pixel basis, with the satellite image data as input variables. Neural networks were chosen because they have proven particularly effective in integrating multiple types of spatial data in the past 19 , including in agricultural applications [20][21][22][23] , and because they can easily be adapted to handle a large number of input variables, as required in a multi-temporal context and with large numbers of channels 24 (for example with hyperspectral imagery 25 ) or with combinations of multispectral imagery and multi-polarization imagery as required in this particular application.…”

Section: Yield Prediction Experimentsmentioning

confidence: 99%

Crop yield prediction using multipolarization radar and multitemporal visible/infrared imagery

Davis

Wilkinson

2006

SPIE Proceedings

View full text Add to dashboard Cite

This paper describes research undertaken on the improvement of within-field late season yield forecasting for crops such as wheat using multi-temporal visible/infrared satellite imagery and multi-polarization radar satellite imagery. Experiments have been carried out using ASAR imagery from Envisat combined with nine bands of ASTER imagery from the NASA Terra satellite. An experimental test site in an agricultural area in the county of Lincolnshire, UK, has been used. The satellite imagery has been integrated using artificial neural networks which have been trained as predictors of the spatial distributions of yield per unit area in a variety of fields. Ground truth data in the form of yield maps from GPS-enabled combine harvesters have been used to train the neural networks and to evaluate accuracy. The results show that the combinations of ASTER and ASAR imagery can provide enhanced yield predictions with overall correlations of up to 0.77 between predicted and actual yield patterns. The results also show that the use of dual polarization radar data alone is not sufficient to give reasonable yield predictions even in a multi-temporal mode. It has also been shown that varying the architectures of the neural networks with ensembles can improve the overall results. KeywordsAgriculture, remote sensing, yield prediction, radar imagery, multi-temporal data, multi-polarization. Disciplines Computer Sciences | Databases and Information Systems | Numerical Analysis and Scientific ComputingComments ©Copyright 2009 SPIE Society of Photo-Optical Instrumentation Engineers. One print or electronic copy may be made for personal use only. Systematic electronic or print reproduction and distribution, duplication of any material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited. This paper was given at the SPIE ABSTRACTThis paper describes research undertaken on the improvement of within-field late season yield forecasting for crops such as wheat using multi-temporal visible/infrared satellite imagery and multi-polarization radar satellite imagery. Experiments have been carried out using ASAR imagery from Envisat combined with nine bands of ASTER imagery from the NASA Terra satellite. An experimental test site in an agricultural area in the county of Lincolnshire, UK, has been used. The satellite imagery has been integrated using artificial neural networks which have been trained as predictors of the spatial distributions of yield per unit area in a variety of fields. Ground truth data in the form of yield maps from GPS-enabled combine harvesters have been used to train the neural networks and to evaluate accuracy. The results show that the combinations of ASTER and ASAR imagery can provide enhanced yield predictions with overall correlations of up to 0.77 between predicted and actual yield patterns. The results also show that the use of dual polarization radar data alone is not sufficient to give reasonable yield predictions even in a multi-temporal mode. It...

show abstract

Artificial neural networks for mapping regional‐scale upland vegetation from high spatial resolution imagery

Cited by 23 publications

References 31 publications

Upland vegetation mapping using Random Forests with optical and radar satellite data

Upland vegetation mapping using Random Forests with optical and radar satellite data

Artificial Neural Networks and Remote Sensing

Crop yield prediction using multipolarization radar and multitemporal visible/infrared imagery

Contact Info

Product

Resources

About