The application of artificial neural networks to the analysis of remotely sensed data

Mas, Jean-François; Flores, Juan J.

doi:10.1080/01431160701352154

Cited by 492 publications

(285 citation statements)

References 247 publications

Supporting

Mentioning

279

Contrasting

Unclassified

Order By: Relevance

“…This capacity to generalize means that ANNs can be effective in situ ations where data may be missing or imprecise. ANNs are also able to incorporate prior knowledge and physical constraints into the analysis, while making no assumptions about the statistical nature of the input data [35,36]. This allows for the incorporation of disparate data from many remote sensing and ancillary sources, and can include variables such as terrain height, slope, aspect, soil texture and land cover.…”

Section: Introductionmentioning

confidence: 99%

Artificial Neural Network Modeling of High Arctic Phytomass Using Synthetic Aperture Radar and Multispectral Data

et al. 2014

View full text Add to dashboard Cite

Vegetation in the Arctic is often sparse, spatially heterogeneous, and difficult to model. Synthetic Aperture Radar (SAR) has shown some promise in above-ground phytomass estimation at sub-arctic latitudes, but the utility of this type of data is not known in the context of the unique environments of the Canadian High Arctic. In this paper, Artificial Neural Networks (ANNs) were created to model the relationship between variables derived from high resolution multi-incidence angle RADARSAT-2 SAR data and optically-derived (GeoEye-1) Soil Adjusted Vegetation Index (SAVI) values. The modeled SAVI values (i.e., from SAR variables) were then used to create maps of above-ground phytomass across the study area. SAVI model results for individual ecological classes of polar semi-desert, mesic heath, wet sedge, and felsenmeer were reasonable, with r 2 values of 0.43, 0.43, 0.30, and 0.59, respectively. When the outputs of these models were combined to analyze the relationship between the model output and SAVI as a group, the r 2 value was 0.60, with an 8% normalized root mean square error (% of the total range of phytomass values), a positive indicator of a relationship. The above-ground phytomass model also resulted in a very strong relationship (r 2 = 0.87) between SAR-modeled and field-measured phytomass. A positive relationship was also found between optically derived SAVI values and field measured phytomass (r 2 = 0.79). These relationships demonstrate the utility of SAR data, compared to using optical data alone, for modeling OPEN ACCESSRemote Sens. 2014, 6 2135 above-ground phytomass in a high arctic environment possessing relatively low levels of vegetation.

show abstract

Section: Introductionmentioning

confidence: 99%

Artificial Neural Network Modeling of High Arctic Phytomass Using Synthetic Aperture Radar and Multispectral Data

et al. 2014

View full text Add to dashboard Cite

show abstract

“…De acuerdo a Mas y Flores (2008), la definición presentada por Haykin (1999) asemeja a las ANNs con el cerebro en dos aspectos, en primer lugar resalta el hecho de que el conocimiento es adquirido por la red desde su medio, a través de un proceso de aprendizaje y por otra parte las fuerzas de las conexiones inter-neuronas, conocidas como pesos sinápticos, son usadas para almacenar el conocimiento adquirido. El primer aspecto está relacionado a las características de los datos de entrada y salida, así como con los mecanismos de entrenamiento de la red, mientras que el segundo aspecto está relacionado con la estructura de la ANN.…”

Section: Generalidadesunclassified

“…La forma en que las neuronas son dispuestas en una red determinan la arquitectura o topología de la red, la cual está estrechamente relacionada con la selección del algoritmo de entrenamiento (Mas y Flores, 2008). …”

Section: Generalidadesunclassified

Análisis Comparativo de Ajuste en Entrenamiento de Redes Neuronales artificiales a partir de las Librerías Open NN y ALGLIB

Muñoz

Seijas

2015

lgr

View full text Add to dashboard Cite

Manuscrito recibido el 19 de agosto de 2014. Aceptado, tras revisión, el 1 de junio de 2015. ResumenEn las últimas décadas son muchos los avances que han tenido lugar en el desarrollo de aplicaciones y alcances de las redes neuronales artificiales, y de igual modo el desarrollo tecnológico en el área de la computación. Este tipo de avances han incidido directamente en el número de publicaciones de aplicaciones, en diversas áreas del conocimiento, basadas en este método de inteligencia artificial. Ahora bien, hasta el presente sigue siendo tema de discusión la idoneidad y aplicabilidad de herramientas de software libre para facilitar la implementación y la calidad de resultados. En este contexto, este trabajo representa un análisis comparativo de aplicaciones usando las librerías ALGLIB y Open NN (Open Source Neural Networks C++ Library), orientadas al entrenamiento y reproducción de redes neuronales artificiales. De igual modo, se establece una evaluación de los resultados obtenidos a partir de los niveles de correlación entre la salida de valores para redes entrenadas y un conjunto de datos de entrenamiento simulados. Palabras claves:Redes Neuronales Artificiales, ALGLIB, Open NN, Algoritmo Quasi-Newton, C++. AbstractIn the last decades, there have been a considerable amount of innovations in the development of applications and the scope of artificial neural networks, and likewise the technological development in computer science. These improvements have had a direct effect in the number of publications on applications, in diverse areas of knowledge, based on this artificial intelligence method. Until now, the adequacy and applicability of free software tools to facilitate the implementation and the quality of results is still under discussion. In this context, this work presents a comparative analysis of such applications using libraries ALGLIB and Open NN, oriented to training and reproduction of artificial neural networks. Also, we propose an evaluation of the results obtained from the levels of correlation between the output values for trained networks and a set of data for simulated training.

show abstract

“…As for dust aerosols, by further assuming that IR AOT can be expressed by IR BT and associated parameters (e.g., surface emissivity and surface temperature)-as demonstrated in the recent success measuring IR AOTs from AIRS and IASI IR hyperspectral (Pierangelo et al 2004;Peyridieu et al 2010;Yao et al 2011)-the implicit relationship between VIS AOT and IR BT can be expected. The expected relationship may be highly nonlinear, and thus the ANN approach is used to obtain the relationship since the ANN is well known as a mathematical tool to solve forward and inverse remote sensing problems (Krasnopolsky and Schiller 2003;Mas and Flores 2008). In this study, we use the so-called Multi-Layer Perceptron (MLP) ANN model, which allows a feed-forward network to link AOT at 550 nm with BTs in the IR bands of MODIS.…”

Section: Artificial Neural Network (Ann) Modelmentioning

confidence: 99%

Nighttime AOT Retrieval for Asian Dusts from MODIS IR Measurements: An Artificial Neural Network Approach

Lee

Sohn²

2012

Journal of the Meteorological Society of Japan

View full text Add to dashboard Cite

In order to estimate MODIS-equivalent aerosol optical thicknesses (AOTs) for dust particles during the nighttime over East Asia, an Artificial Neural Network (ANN) model approach was used to combine MODIS-measured infrared (IR) brightness temperatures (BTs) and visible (VIS) AOTs. For training the ANN model, IR BTs were used together with surface type and geometrical information as inputs to predict MODIS-derived AOTs as target data during the daytime when VIS-based AOTs are available. The training was done exclusively over dust-laid pixels during the spring (MarchMay) of 2006 over the East Asian domain (20It should be noted that the obtained daytime AOTs from the ANN model are in good agreement with MODIS-derived AOTs, with a correlation coefficient of 0.77 over the analysis domain. Although a weaker correlation is found during the nighttime when derived AOTs are compared against AOTs from CALIPSO, the case study indicates that the developed ANN method appears to effectively depict both the evolutionary features and intensity of the Asian dust plume during the nighttime. Results further indicate that nighttime VIS-like AOTs can be readily used for monitoring dust movement and its intensity during the nighttime, filling the gap between consecutive daytime AOT distributions.

show abstract

The application of artificial neural networks to the analysis of remotely sensed data

Cited by 492 publications

References 247 publications

Artificial Neural Network Modeling of High Arctic Phytomass Using Synthetic Aperture Radar and Multispectral Data

Artificial Neural Network Modeling of High Arctic Phytomass Using Synthetic Aperture Radar and Multispectral Data

Análisis Comparativo de Ajuste en Entrenamiento de Redes Neuronales artificiales a partir de las Librerías Open NN y ALGLIB

Nighttime AOT Retrieval for Asian Dusts from MODIS IR Measurements: An Artificial Neural Network Approach

Contact Info

Product

Resources

About