An evaluation of Guided Regularized Random Forest for classification and regression tasks in remote sensing

New Earth observation missions and technologies are delivering large amounts of data. Processing this data requires developing and evaluating novel dimensionality reduction approaches to identify the most informative features for classification and regression tasks. Here we present an exhaustive evaluation of Guided Regularized Random Forest (GRRF), a feature selection method based on Random Forest. GRRF does not require fixing a priori the number of features to be selected or setting a threshold of the feature importance. Moreover, the use of regularization ensures that features selected by GRRF are non-redundant and representative. Our experiments based on various kinds of remote sensing images, show that GRRF selected features provides similar results to those obtained when using all the available features. However, the comparison between GRRF and standard random forest features shows substantial differences: in classification, the mean overall accuracy increases by almost 6% and, in regression, the decrease in RMSE almost reaches 2%. These results demonstrate the potential of GRRF for remote sensing image classification and regression. Especially in the context of increasingly large geodatabases that challenge the application of traditional methods.     

Size-Flux Relation in Solar Active Regions

We present a study of the relationship between integral area and corresponding total magnetic flux for solar active regions. It is shown that some of these relationships are satisfied to simple power laws. Fractal examination showed that some of these power laws can not be justified inside the simple models of stationary magnetic flux tube aggregation. All magnetic fluxes and corresponding areas were calculated using the data measured with the Solar Magnetic Field Telescope of the Huairou Solar Observing Station in Beijing.     

The depolarization properties of powerful extragalactic radio sources as a function of cosmic epoch

We use the observed polarization properties of a sample of 26 powerful radio galaxies and radio-loud quasars to constrain the conditions in the Faraday screens local to the sources. We adopt the cosmological redshift, low-frequency radio luminosity and physical size of the large-scale radio structures as our 'fundamental' parameters. We find no correlation of the radio spectral index with any of the fundamental parameters. The observed rotation measure is also independent of these parameters, suggesting that most of the Faraday rotation occurs in the Galactic foreground. The difference between the rotation measures of the two lobes of an individual source, as well as the dispersion of the rotation measure, shows significant correlations with the source redshift, but not with the radio luminosity or source size. This is evidence that the small-scale structure observed in the rotation measure is caused by a Faraday screen local to the sources. The observed asymmetries between the lobes of our sources show no significant trends with each other or other source properties. Finally, we show that the commonly used model for the depolarization of synchrotron radio emission by foreground Faraday screens is inconsistent with our observations. We apply alternative models to our data and show that they require a strong increase of the dispersion of the rotation measure inside the Faraday screens with cosmological redshift. Correcting our observations with these models for redshift effects, we find a strong correlation of the depolarization measure with redshift and a significantly weaker correlation with radio luminosity. We do not find any (anti-)correlation of depolarization measure with source size. All our results are consistent with a decrease in the order of the magnetic field structure of the Faraday screen local to the sources for increasing cosmological redshift.     

Can the inner gap sparking take place in millisecond pulsars？

The inner vacuum gap model has become the foundation stone of most theories on pulsar radio emission. The fundamental picture of this model is the sparking, which was conjectured to be induced by magnetic absorption of background gamma photons. However, a question is, can the sparking be triggered in the millisecond pulsars (MSPs) with magnetic fields (B) only about 108 G? We investigate this problem by including the pair production above the inner gap. Under the assumption that the magnetic field is dipolar, our results show the background gamma-ray emission can not be the key factor that triggers the sparking, at least not in MSPs with B - 108 G, if the temperature in the polar cap region is only so high as is observed (< 4 × 106 K). Some other mechanisms are required.     

Minimal Energy and Gravitation

Quantum theory in Robertson – Walker spacetime suggests the existence of a minimal energy ε of the order of 10⁻⁴⁵ erg. Reasonable forms for ε give the expansion factor R=R(t)(t= the cosmic time) with no need of gravitational field equations.Einstein's theory should be modified in gravitational fields of strength less than ε c/ħ ∼ 10⁻⁸ cm/s² where c is the speed of light and ħ is the reduced Planck constant. The cosmological term λ is expected to decrease as the universe expands.In the Appendix, ε is derived from a big bang – big crunch Newtonian cosmology.     

Three Super Active Regions in the Descending Phase of Solar Cycle 23

We analyze the magnetic configurations of three super active regions, NOAA 10484, 10486 and 10488, observed by the Huairou Multi-Channel Solar Telescope (MCST) from 2003 October 18 to November 4. Many energetic phenomena, such as flares (including a X-28 flare) and coronal mass ejections (CMEs), occurred during this period. We think that strong shear and fast emergence of magnetic flux are the main causes of these events. The question is also of great interest why these dramatic eruptions occurred so close together in the descending phase of the solar cycle.     

On the Line Formation in Stellar Magnetized Atmospheres

1 INTRODUCTIONIn the previous paper (on et al. 1999, hereafter Paper I), we investigated the wavelength-dependence of four colltribution functions (CFs) derived from dmerent formal solutions andreferring to different emergellt quantities in the unpolarized case. Because one cannot generallyassign a single formation region to the whole line band in a real stellar atmosphere, e.g., the solaratmosphere, instead, the line formation region can be defined as the layers deviating farthestfrom t…     

Ellerman Bombs,Type Ⅱ White—light Flares and Magnetic Reconnection in the Solar Lower Atmosphere

Ellerman bombs and type Ⅱ white-light flares share many common features despite the large energy gap between them.Both are considered to result from local heation in the solar lower atmosphere.This paper presents numerical simulations of magnetic reconnection occurring in such a deep atmosphere,with the aim to account for the common features of the two phenomena.Our numerical results manifest the following two typical characteristics of the assumed reconnection process:(1)magnetic reconnection saturates in-600-900s,which is just the lifetime of the two phenomens;(2)ionization in the upper chromosphere consumes quite a large part of the energy released through reconnection,making the heation effect most significant in the lower chromosphere.The applicationof the reconnection model to the two phenomena is discussed in detail.