Search

The air temperature is a critical factor in many societal challenges to protect human health and the environment. Moreover, a vital climatic parameter, the temperature has a direct impact on evaporation, frost, and snow melting. Temperature predictions are based mainly on numerical and statistical models. Sometimes it is a challenge to improve the weather forecast accuracy. The article aims to implement a weather prediction technique based on machine learning methods and approaches to improve the hourly air temperature prediction for up to 24 hours in the Ararat valley (Armenia). Due to intense heat and low relative humidity, the high temperatures and hot winds occur between 120 and 160 days per year in Ararat valley, as one of the aridest regions of Armenia. The approach utilizes the earth observation data received from several meteorological stations and the large satellite analysis-ready datasets at different frequencies and resolutions. The experiments have been conducted with multiple neural networks to forecast air temperatures for 24 hours that happened over the Ararat valley. The suggested model has 87.31% and 75.57% accuracies to predict the temperature for the next 3 and 24 hours, which is sufficient to be used alongside the current state-of-the-art techniques. [ABSTRACT FROM AUTHOR]

The main goal of this research was to estimate heavy metals (HMs) (molybdenum (Mo), copper (Cu), nickel (Ni), cadmium (Cd)) contents in crop leaves through multispectral satellite imagery. During the acquisition of a SPOT 7 satellite image (28 July 2017) in situ sampling (38 samples) was done from the leaves of potatoes and beans growing close to the mining town of Kajaran (Armenia). To estimate HMs contents, multivariate regression (multiple linear regression (MLR), partial least squares regression (PLSR)), and artificial neural network (ANN) were used. As input data for the models raw, atmospherically corrected (Dark Object Subtraction (DOS)) and hyperspherical direction cosines (HSDC) normalized values of SPOT 7 spectral data in combination with one or combined log10, multiplicative scatter correction (MSC), standard normal variate transform (SNV) preprocessing methods were utilized. The best results were obtained for Cu using MLR (R2cal. = 0.79, R2CV = 0.70, RMSEcal. = 11.27, RMSECV = 13.47) and ANN (R2Train ≈ 0.80, R2Test ≈ 0.72, RMSETrain ≈ 11, RMSETest ≈ 13) models in case of bean leaves. The results are quite optimistic, however, further research with the use of high spatial/spectral resolution satellite images is needed to improve the accuracy of models. [ABSTRACT FROM AUTHOR]