Mónica Bocco

Comparison of Regression and Neural Networks Models to Estimate Solar Radiation

La radiacion solar incidente en el suelo es una variable importante usada en aplicaciones agronomicas, ademas es relevante en hidrologia, meteorologia y fisica del suelo, entre otros. Para estimarla se han desarrollado modelos empiricos que utilizan distintos parametros meteorologicos y, recientemente, modelos de pronostico y prediccion basados en tecnicas de inteligencia artificial tales como redes neuronales. El objetivo de este trabajo fue desarrollar modelos lineales y de redes neuronales, del tipo perceptron multicapa, para estimar la radiacion solar global diaria y comparar la eficiencia de los mismos en su aplicacion para una region de la Provincia de Salta, Argentina. Se utilizaron datos de heliofania relativa, temperaturas maxima y minima, precipitacion, precipitacion binaria y radiacion solar astronomica provistos por la Estacion Experimental Salta, Instituto Nacional de Tecnologia Agropecuaria (INTA), Argentina, correspondientes al periodo 1996-2002. Tanto para los modelos de redes neuronales como para las regresiones lineales se consideraron tres alternativas de combinaciones de los parametros meteorologicos, obteniendose buenos resultados con ambas metodologias de prediccion, con valores de la raiz del error cuadratico medio variando desde 1.99 a 1.66 MJ m-2 d-1 y coeficientes de correlacion de 0.88 a 0.92. Se concluye que ambos, los modelos de redes neuronales y las regresiones lineales, pueden ser usados para predecir en forma adecuada la radiacion solar global diaria; si bien las redes neuronales produjeron mejores resultados.

Development and evaluation of neural network models to estimate daily solar radiation at Córdoba, Argentina

The objective of this work was to develop neural network models of backpropagation type to estimate solar radiation based on extraterrestrial radiation data, daily temperature range, precipitation, cloudiness and relative sunshine duration. Data from Cordoba, Argentina, were used for development and validation. The behaviour and adjustment between values observed and estimates obtained by neural networks for different combinations of input were assessed. These estimations showed root mean square error between 3.15 and 3.88 MJ m -2 d -1 . The latter corresponds to the model that calculates radiation using only precipitation and daily temperature range. In all models, results show good adjustment to seasonal solar radiation. These results allow inferring the adequate performance and pertinence of this methodology to estimate complex phenomena, such as solar radiation.

Neural Network Model for Land Cover Classification from Satellite Images

Los datos de cobertura de suelo representan informacion ambiental clave para aplicaciones cientificas y politicas, por esto su clasificacion a partir de imagenes satelitales es importante. Las redes neuronales (NN) constituyen una herramienta valiosa para clasificar imagenes satelitales pues no requieren hipotesis sobre la distribucion de los datos. Los objetivos de este trabajo fueron desarrollar modelos de NN para clasificar datos de cobertura de suelo a partir de informacion proveniente de imagenes satelitales y evaluarlos cuando se utilizan diferentes variables de entrada. Se utilizaron imagenes satelitales MODIS-MYD13Q1 y datos de 85 parcelas en Cordoba (Argentina). Se disenaron cinco NN del tipo perceptron multicapa feed-forward. Cuatro de ellas recibieron como patrones de entrada valores de NDVI (Normalized Difference Vegetation Index), EVI (Enhanced Vegetation Index), de reflectancias en la bandas roja (RED) y en infrarroja cercana (NIR), respectivamente. La quinta NN tuvo como valores de entrada las reflectancias RED y NIR. La validacion, permitio concluir que todos los modelos presentan un buen desempeno. El modelo que muestra mejor comportamiento es aquel que considera conjuntamente valores de reflectancias RED y NIR, cuya precision en la clasificacion es del 93% con un estadistico Kappa excelente. Las redes construidas individualmente a partir de valores de NDVI y EVI tienen un comportamiento similar (86 y 83% de exactitud, respectivamente), con estadisticos Kappa muy bueno y bueno, respectivamente. Las NN que incluyen solo valores de RED o NIR presentaron los menores porcentajes de exactitud (76 y 81%, respectivamente) con indices Kappa regular y bueno, respectivamente.

Estimating soybean ground cover from satellite images using neural-networks models

The ground cover is a necessary parameter for agronomic and environmental applications. In Argentina, soybean (Glycine max (L.) Merill) is the most important crop; therefore it is necessary to determine its amount and configuration. In this work, neural-network (NN) models were developed to calculate soybean percentage ground cover (fractional vegetation cover, fCover) and to compare the accuracy of the estimate from Moderate-Resolution Imaging Spectroradiometer (MODIS) and Landsat satellites data. The NN design included spectral values of the red and near-infrared (NIR) bands as input variables and one neuron output, which expressed the estimated coverage. Data of fCover were acquired throughout the growing season in the central plains of Córdoba (Argentina); they were used for training and validating the networks. The results show that the NNs are an appropriate methodology for estimating the temporal evolution of soybean coverage fraction from MODIS and Landsat images, with coefficients of determination (R 2) equal to 0.90 and 0.91, respectively.

Neural network and crop residue index multiband models for estimating crop residue cover from Landsat TM and ETM+ images

Crop residues on the soil surface provide not only a barrier against water and wind erosion, but they also contribute to improving soil organic matter content, infiltration, evaporation, temperature, and soil structure, among others. In Argentina, soybean (Glycine max (L.) Merill) and corn (Zea mays L.) are the most important crops. The objective of this work was to develop and evaluate two different types of model for estimating soybean and corn residue cover: neural networks (NN) and crop residue index multiband (CRIM) index, from Landsat images. Data of crop residue were acquired throughout the summer growing season in the central plains of Córdoba (Argentina) and used for training and validating the models. The CRIM, a linear mixing model of composite soil and residue, and the NN design, included reflectance and digital numbers from a combination of different TM bands to estimate the fractional residue cover. The results show that both methodologies are appropriate for estimating the residue cover from Landsat data. The best developed NN model yielded R2 = 0.95 when estimating soybean and corn residue cover fraction, whereas the best fit using CRIM yielded R2 = 0.87; in addition, this index is dependent on the soil and residue lines considered.

REDES NEURONALES PARA MODELAR PREDICCIÓN DE HELADAS

A B S T R A C T In this work models based on neural networks of the backpropagation type were developed in order to predict the occurrence of frosts from meteorological data such as temperature, relative humidity, cloudiness and wind direction and speed. The training and the validation of the networks were made on the basis of 24 years of meteorological data corresponding to the Rio Cuarto station, Cordoba, Argentina. These data were grouped as follows: 10 years for the training data set and 14 years for the validation data set. Different models were built to evaluate the performance of the networks when different numbers of input variables and/or neurons in the hidden layer are used, and the probabilities of success in the prediction results on considering different input variables. In the models used, the percentage of days with prediction error was 2%, approximately, for the 14 years of application; when effective frosts days are considered the percentage varies between 10 and 23%, for the same period. The simulation results demonstrated the good performance and the relevance of this methodology for the estimation of the behavior of non-linear phenomena like frosts.

MODELOS MULTICRITERIO: UNA APLICACIÓN A LA SELECCIÓN DE ALTERNATIVAS PRODUCTIVAS

En este trabajo se desarrollan modelos basados en los metodos de programacion multicriterio: se utiliza la programacion multiobjetivo, complementada con la programacion compromiso y por metas, con el fin de evaluar la optimizacion de mas de un objetivo economico. Estos modelos tienen como proposito predecir ex-ante los resultados economicos, maximizacion del margen bruto y minimizacion del riesgo empresarial, que se observaran en los sistemas horticolas al adoptar nuevas alternativas de produccion, contemplando distintas restricciones agroeconomicas. Se realizo su aplicacion a las pequenas explotaciones horticolas del Cinturon Verde de Cordoba, Argentina, para explicar el cambio de la situacion economica de la empresa al seleccionar y adoptar nuevos planes de produccion, representados por la incorporacion de productos horticolas alternativos para su produccion. La comparacion de cada modelo con la situacion actual de las empresas, en cuanto a las variables involucradas, permite concluir que la adopcion de cualquiera de las propuestas modeladas significara una disminucion importante de la relacion riesgo/margen bruto en valores de hasta un 50% para la estacion estival, y un 66% para la estacion fria. La eleccion de las soluciones que, sin ser optimas seran eficientes, corresponden a una mayor diversificacion de los cultivos a realizar.

Assessment of surface solar irradiance retrieved by CERES

ABSTRACT This study compared and evaluated the monthly global solar radiation generated by Clouds and the Earth’s Radiant Energy System (CERES) with the surface radiation registered in 232 meteorological stations located in whole Spain, for the period of July 2006–July 2015. Results showed strong correlations between CERES and registered data with R2 values greater than 0.96, for all stations considered. When the temporal evolution of recorded and provided by CERES solar radiation was analysed, a systematic overestimation by CERES was detected from July 2011, although the shapes of both curves were respected in the whole period. This finding led us to propose a linear adjustment model since July 2011. After applying the developed model to rescale CERES data for the whole period, an improvement in solar radiation fit was observed. Our finding offers an insight into error patterns of CERES solar radiation, since July 2011, and proposes a model for improving accuracy allowing therefore a reliable use of this product. Moreover, our study on radiation data of Spain provides a case example for worldwide validation.