Multitime scale variations of sea surface temperature in the China seas based on the HadISST dataset

The variability of the sea surface temperature（SST） in the China seas has been studied in seasonal,interannual and interdecadal scales based on the monthly data of HadISST spanning from 1870 to 2007. The main results obtained are SST in the China offshore changes most actively at the seasonal scale with the intensity diminishing from north to south,as the temperature differences between summer and winter reaching 17 and 4 C in the northern and southern areas,respectively. Moreover,seasonal variation near the coastal regions seems relatively stronger than that far from the coastline;significant interannual variations are detected,with the largest positive anomaly occurring in 1998 in the overall area. But as far as different domains are concerned,there exists great diversity,and the difference is also found between winter and summer. Differed from the seasonal variations,where the strongest interannual variability takes place,resides to the south of that of the seasonal ones in the northern section,nevertheless in the South China Sea,the most significant interannual variability is found in the deep basin;interdecadal changes of summer,winter and annual mean SST in different domains likewise present various features. In addition,a common dominant warming in recent 20 a are found in the overall China offshore with the strongest center located in the vicinity of the Changjiang Estuary in the East China Sea,which intensifies as high as 1.3 C during the past 130 a.

Scale correlation between vegetation and soil in larch forest, NE China

After the analysis on the linages of species, the fractal dimensions of vegetation and soil in a managed larch(Larix gmelini) forest in Daxingan Mountains, NE China were estimated separately, and their scale correlation was discussed. (1) The dominant species of the larch forest, larch and the important accompany species, birch (Betula platyphylla) were homogeneously distributed along the transect. The other species were heterogeneous, and can be divided into three groups located on the right, left and central parts of the transect respectively. The transect can be separated into two parts at the position for the 29th quadrat(580m). (2) The fractal dimension in the large scale range from 200 to 400m was lower than that in the small scale range from 0 to 200m in the forest, indicating the different variation regularities of spatial heterogeneity in different scale ranges. The inflection point at 200m was one of the key scales of spatial hierarchy of the larch forest. (3) The scale variation of the forest was correlated with that of soil pH on large scales, which reflected the indirect control of the dominant species to the spatial pattern and species distribution of understorey by changing soil acidity in the forest and the indirect effects of forest management.

Discrete Fractional Birkhoff Equations in Terms of Time Scales

In order to study discrete fractional Birkhoff equations for Birkhoffian systems,the method of isochronous variational principle is used in this paper. Discrete fractional Pfaff-Birkhoff principle in terms of time scales is presented. Discrete fractional Birkhoff equations with left and right discrete operators of Riemann-Liouville type are established and some special cases including classical discrete Birkhoff equations,discrete fractional Hamilton equations and discrete fractional Lagrange equations are discussed. Finally,an example is devoted to illustrate the results.

Object Counting Using a Refinement Network

To address the scale variance and uneven distribution of objects in scenarios of object-counting tasks,an algorithm called Refinement Network(RefNet) is exploited.The proposed top-down scheme sequentially aggregates multiscale features,which are laterally connected with low-level information.Trained by a multiresolution density regression loss,a set of intermediate-density maps are estimated on each scale in a multiscale feature pyramid,and the detailed information of the density map is gradually added through coarse-to-fine granular refinement progress to predict the final density map.We evaluate our RefNet on three crowd-counting benchmark datasets,namely,ShanghaiTech,UCFC0,and UCSD,and our method achieves competitive performances on the mean absolute error and root mean squared error compared to the state-of-the-art approaches.We further extend our RefNet to cell counting,illustrating its effectiveness on relative counting tasks.