一种注意力卷积与差值特征结合的遥感变化检测模型

为了有效地提取变化目标,保证变化目标提取的准确性与完整性,文章提出了一种结合了注意力卷积与差异强化结构的三分支网络ACSCN。该网络具有一个变化学习主分支与两个双时相特征学习附属分支,使用包含通道注意力模块和空间注意力模块的注意力卷积模块来代替普通卷积与池化,并利用强化差异模块来强化两个分支的变化差异特征。为了进一步证明ACSCN的性能,我们用不同方法做了对比。实验证明,相比其它方法,ACSCN能更好地检测目标变化,在边界完整性,提取准确性上具有明显的优势,注意力卷积模块与强化差异模块的加入对方法的性能做了有效地提升。

利用遥感影像自动提取变化信息的方法研究

目前土地利用遥感动态监测已进入实用化阶段,但其自动化程度不高,影响作业效率。本文在使用增强法对两时相的遥感影像自动发现变化基础上,结合两时相影像的分类结果,对变化图斑进行类别标记,然后对标记图像进行矢量化、变化信息提取和统计,从而提高变化监测工作中的自动化程度、减轻工作量。

Modeling an emissions peak in China around 2030: Synergies or trade-offs between economy, energy and climate security

China has achieved a political consensus around the need to transform the path of economic growth toward one that lowers carbon intensity and ultimately leads to reductions in carbon emissions, but there remain different views on pathways that could achieve such a transformation. The essential question is whether radical or incremental reforms are required in the coming decades. This study explores relevant pathways in China beyond 2020, particularly modeling the major target choices of carbon emission peaking in China around 2030 as China-US Joint Announcement by an integrated assessment model for climate change IAMC based on carbon factor theory. Here scenarios DGS-2020, LGS2025, LBS-2030 and DBS-2040 derived from the historical pathways of developed countries are developed to access the comprehensive impacts on the economy, energy and climate security for the greener development in China. The findings suggest that the period of 2025--2030 is the window of opportunity to achieve a peak in carbon emissions at a level below 12 Gt CO2 and 8.5 t per capita by reasonable trade-offs from economy growth, annually -0.2% in average and cumulatively -3% deviation to BAU in 2030. The oil and natural gas import dependence will exceed 70% and 45% respectively while the non-fossil energy and electricity share will rise to above 20% and 45%. Meantime, the electrification level in end use sectors will increase substantially and the electricity energy ratio approaching 50%, the labor and capital productivity should be double in improvements and the carbon intensity drop by 65% by 2030 compared to the 2005 level, and the cumulative emission reductions are estimated to be more than 20 Gt CO2 in 2015-2030.

Enhancement of signal-noise-ratio in a distributed polarization mode coupling detection system

A distributed polarization-mode coupling measurement system was designed and implemented using white light interferometry. It can be used for the measurement of polarization mode coupling in a high-birefringence fiber of up to i km. This system can be used in both fiber-optic sensors and optical fiber communications. Wavelet Transform was adopted in data processing to improve the signal-noise-ratio. The signal-noise-ratio of this system was improved more than 15 dB after denoising. The influence of denoising threshold on signal-noise-ratio and measurement accuracy was also discussed. Hilbert Transform and non-linear regression can be used in conjunction with Wavelet Transform to enhance the signal-noise-ratio and spatial resolution of this system.

Study on the Anomaly of Small Earthquake Activity Before Moderate-strong Earthquake in the Northeast Margin of the Qinghai-Xizang (Tibet) Block

26 earthquakes with MS ≥5. 0 have been recorded in the northeast margin of the Qinghai- Xizang (Tibet) block since 1980,22 of which were relatively independent of other moderate- strong earthquakes. Research on the increase of small earthquake activity before the 22 moderate-strong earthquakes has indicated that small earthquake activity was enhanced before 17 of the moderate-strong earthquakes. Though the increased seismicity is a common phenomenon in the northeast margin of the Qinghai-Xizang ( Tibet ) block,we have difficulty in predicting the moderate-strong earthquakes by this phenomenon. In order to predict the moderate-strong earthquakes through the increased seismicity of small earthquakes,this paper attempts to propose a new method, which calculates small earthquake frequency through the change of distribution pattern of small earthquakes, based on the characteristics of small earthquake activity in the northeastern Qinghai-Xizang (Tibet) block,and then make primary applications. The result shows that we are able to obtain obvious anomalies in the frequency of small earthquakes before moderate strong earthquakes through the new method,with little spatial range effect on the amplitude of this small earthquake frequency anomaly. We can obtain mid to short-term anomaly indices for moderate-strong earthquakes in the northeast margin of the Qinghai-Xizang (Tibet) block.

Study on Stability of Ion-exchange Resin Catalysts:Ⅱ. Desulfonation during Isobutylene Dimerization

The performance of ion-exchange resin catalysts during isobutene (IB) dimerization was investigated under different IB contents,temperatures and liquid-volume hourly space velocity (LHSV) using a plug flow reactor in the absence of any selectivity enhancing component.High IB content and temperature resulted in a high conversion and C12 selectivity bu low C8 selectivity.The influence of LHSV was related with the IB content:LHSV had great effect at high IB content,while the performance of ion-exchange resin changed little with LHSV if IB content was low.The effect of water on the stability of resins was also studied.Desulfonation was observed during the C4 dimerization reaction when water was added to the feed.Chlorinated resin was more stable than conventional polystyrene-based resins during the test.

An Experimental Investigation of Glass Transition Temperature of Composite Materials Using Bending Test

In this study, the bending test is used to investigate the glass transition temperature for epoxy reinforced with three types ot fibers, fiberglass, Kevlar and synthetic wool, these materials have a wide used in many application which they are used composite materials. The glass transition temperature can be measured at the point of inflection for ＂the curve of variation of the deflection and temperature. The results show that, the glass transition temperature is affected by the type of the reinforcement of the composites. On the other hand, the glass transition temperature of the wool composite is higher than the other.

Temporal Variations in Growth and Aboveground Biomass of Phragmites australis and EVI Analysis in Jiaozhou Bay Coastal Salt Marshes, China

To better ascertain leaf, stem and flower traits, and analyze aboveground allocation during the vegetation period, three sampling vegetation transects were settled and reed samples were collected at intervals to determine morphological parameters and dry and wet weights in Jiaozhou Bay wetland. Remote sensing data were also combined to evaluate regional biomass through EVI regression models. Results show that growth dynamics of leaves and stems changed during the vegetation period. Stem length increased rapidly and peaked in September（194.40 ± 23.89 cm）, whereas leaf width peaked in July. There was a significantly negative correlation between stem length and stem diameter with a value of-0.785. Stem biomass was higher than that of leaves, and the maximum value of aboveground biomass was 27.17 ± 3.56 g. F/C exhibited a tendency to increase and values ranged from 0.37–0.76. The aboveground biomass of sample plots reached a peak of 2356 ± 457 g/m^2 in September. EVI was 0.05–0.5; EVI and biomass had a better fitting effect using the power-exponent model compared with other models and its function was y = 4219.30 x^0.88（R^2 = 0.7810）. R^2 of the other three models ranked as linear function〉 polynomial function 〉exponent function, with the values being 0.7769, 0.7623 and 0.6963, respectively. EVI can be used to estimate vegetation biomass, and effectively solved the problems of the destructive effect to sample plots resulting from traditional harvest methods.

Method of infrared image enhancement based on histogram

Aiming at the problem in infrared image enhancement,a new method is given based on histogram.Using the gray characteristics of target,the upper-bound threshold is selected adaptively and the histogram is processed by the threshold.After choosing the gray transform function based on the gray level distribution of image,the gray transformation is done during histogram equalization.Finally,the enhanced image is obtained.Compared with histogram equalization(HE),histogram double equalization(HDE) and plateau histogram equalization(PE),the simulation results demonstrate that the image enhancement effect of this method has obvious superiority.At the same time,its operation speed is fast and real-time ability is excellent.