Soil salinity prediction in the Lower Cheliff plain(Algeria) based on remote sensing and topographic feature analysis

Soil salinity and ground surface morphology in the Lower Cheliff plain（Algeria） can directly or indirectly impact the stability of environments. Soil salinization in this area is a major pedological problem related to several natural factors, and the topography appears to be important in understanding the spatial distribution of soil salinity. In this study, we analyzed the relationship between topographic parameters and soil salinity, giving their role in understanding and estimating the spatial distribution of soil salinity in the Lower Cheliff plain. Two satellite images of Landsat 7 in winter and summer 2013 with reflectance values and the digital elevation model（DEM） were used. We derived the elevation and slope gradient values from the DEM corresponding to the sampling points in the field. We also calculated the vegetation and soil indices（i.e. NDVI（normalized difference vegetation index）, RVI（ratio vegetation index）, BI（brightness index） and CI（color index）） and soil salinity indices, and analyzed the correlations of soil salinity with topography parameters and the vegetation and soil indices. The results showed that soil salinity had no correlation with slope gradient, while it was significantly correlated with elevation when the EC（electrical conductivity） values were less than 8 d S/m. Also, a good relationship between the spectral bands and measured soil EC was found, leading us to define a new salinity index, i.e. soil adjusted salinity index（SASI）. SASI showed a significant correlation with elevation and measured soil EC values. Finally, we developed a multiple linear regression for soil salinity prediction based on elevation and SASI. With the prediction power of 45%, this model is the first one developed for the study area for soil salinity prediction by the combination of remote sensing and topographic feature analysis.

Prediction of the diet nutrients digestibility of dairy cows using Gaussian process regression

In order to effectively evaluate the diet nutritional value of dairy cows,it is essential to accurately predict the diet nutrients digestibility(DND).Conventional predicting DND methods are usually based on the least squares linear regression analysis(LS-LRA),which often relies on a large amount of training samples to accomplish reliable predictions.However,in real-world applications,it is often extremely difficult,costly and time-consuming to obtain a large number of measured samples,especially for the DND prediction of dairy cows.This paper applies a Gaussian process regression(GPR)technique to predict the DND indicators of dairy cows in small samples.To evaluate prediction accuracy effectively,we compared the GPR technique with the LS-LRA,radial basis function artificial neural network(RBF-ANN),support vector regression(SVR)and least squares support vector regression(LS-SVR)methods,using the required sample data obtained from actual digestion experiments.The prediction results indicate that the GPR technique is superior to other conventional methods(especially the LS-LRA method)in predicting the main DND indicators of dairy cows such as dry matter digestibility(DMD),organic matter digestibility(OMD),neutral detergent fiber(NDFD),acid detergent fiber(ADFD),and crude protein digestibility(CPD).It is worth mentioning that the developed GPR-based prediction technique is more suitable for the prediction problems with small samples,which is often the case in the prediction of DND indicators of dairy cows,and then more coincide with actual needs.

Static Frame Model Validation with Small Samples Solution Using Improved Kernel Density Estimation and Confidence Level Method

An improved method using kernel density estimation （KDE） and confidence level is presented for model validation with small samples. Decision making is a challenging problem because of input uncertainty and only small samples can be used due to the high costs of experimental measurements. However, model validation provides more confidence for decision makers when improving prediction accuracy at the same time. The confidence level method is introduced and the optimum sample variance is determined using a new method in kernel density estimation to increase the credibility of model validation. As a numerical example, the static frame model validation challenge problem presented by Sandia National Laboratories has been chosen. The optimum bandwidth is selected in kernel density estimation in order to build the probability model based on the calibration data. The model assessment is achieved using validation and accreditation experimental data respectively based on the probability model. Finally, the target structure prediction is performed using validated model, which are consistent with the results obtained by other researchers. The results demonstrate that the method using the improved confidence level and kernel density estimation is an effective approach to solve the model validation problem with small samples.