Learning in neural spatial interaction models: A statistical perspective

 In this paper we view learning as an unconstrained non-linear minimization problem in which the objective function is defined by the negative log-likelihood function and the search space by the parameter space of an origin constrained product unit neural spatial interaction model. We consider Alopex based global search, as opposed to local search based upon backpropagation of gradient descents, each in combination with the bootstrapping pairs approach to solve the maximum likelihood learning problem. Interregional telecommunication traffic flow data from Austria are used as test bed for comparing the performance of the two learning procedures. The study illustrates the superiority of Alopex based global search, measured in terms of Kullback and Leibler's information criterion. Received: December 2001 / Accepted: August 2002 The author gratefully thanks Martin Reismann (Department of Economic Geography & Geoinformatics) for his valuable research assistance.     

SAR干涉图的静态小波域MAP法滤波

SAR干涉图中的斑点噪声是对相位解缠的极大障碍，为了得到更精确的地形模型，必须在相位解缠前对干涉图进行斑点噪声滤波。提出了一种干涉图的小波域多尺度滤波算法。对干涉图的实部和虚部分别进行处理，用Pearson分布系确定了信号和噪声项小波系数的概率密度函数，然后用最大后验概率（Maximum A Posterior,MAP)准则估计无噪数据的小波系数。将本算法和另外3种滤波算法进行比较。结果表明，本滤波算法对SAR干涉图有很强的噪声抑制能力，并能保持干涉图的相位细节信息和相位分布。     

功率谱分析中几个重要问题的讨论

本文从理论上推导了谱分析中最大后延M的取值,并且从理论和实际计算中得出了样本长度和其隐含周期之间的关系。只要抽样是平稳的,样本长度并非一定要求太长,仅略为超过其隐含周期的长度即可。但谱分析绝非是万能的。如果样本中仅含有两个波动,且当这两个波动的周期相差一个整数倍或更大时,这两个波动才可以通过谱分析分辨出,反之就会歪曲了其隐含周期。由此可见,在具体应用时,应作一些滤波和去掉趋势是必要的。因而这将对广大气象、地理科学工作者以及其它相邻学科的科研工作者具有重要的参考价值。     

MM5 Simulations of the China Regional Climate During the LGM.I: In uence of CO2 and Earth Orbit Change

Using a regional climate model MM5 nested to an atmospheric global climate model CCM3, a series of simulations and sensitivity experiments have been performed to investigate the relative Last Glacial Maximum (LGM) climate response to different mechanisms over China. Model simulations of the present day (PD) climate and the LGM climate change are in good agreement with the observation data and geological records, especially in the simulation of precipitation change. Under the PD and LGM climate,changes of earth orbital parameters have a small influence on the annual mean temperature over China.However, the magnitude of the effect shows a seasonal pattern, with a significant response in winter. Thus,this influence cannot be neglected. During the LGM, CO2 concentration reached its lowest point to 200 ppmv. This results in a temperature decrease over China. The influences of CO2 concentration on climate show seasonal and regional patterns as well, with a signi cant influence in winter. On the contrary, CO2 concentration has less impact in summer season. In some cases, temperature even increases with decreasing in CO2 concentration. This temperature increase is the outcome of decrease in cloud amount; hence increase the solar radiation that reached the earth's surface. This result suggests that cloud amount plays a very important role in climate change and could direct the response patterns of some climate variables such as temperature during certain periods and over certain regions. In the Tibetan Plateau, the temperature responses to changes of the above two factors are generally weaker than those in other regions because the cloud amount in this area is generally more than in the other areas. Relative to the current climate, changes in orbital parameters have less impact on the LGM climate than changes in CO2 concentration. However,both factors have rather less contributions to the climate change in the LGM. About 3%-10% changes in the annual mean temperature are contributed by CO2.     

黏滑实验的震级评估和应力降分析

本文通过三种结构模型的黏滑地震模拟实验,利用高频速度连续观测系统获得了地震失稳过程的速度特征,讨论了最大位移量的选取方法,估算了实验室黏滑型地震的矩震级,探讨了黏滑类型、应力降大小与震级的关系.结果表明,黏滑型地震的应力降过程可能包含一次到多次高频振荡,对应若干次黏滑子事件.高频振荡的摆动幅度很大,包含有多种频率成分,峰值速度0.003~0.008 m·s^-1.初步估计黏滑型地震的震级范围为-4.4~-3级,断层构造面的差异对各种黏滑模型的地震震级分布有明显影响.总体来看应力降与地震震级没有明显相关性,决定地震震级的主要因素应当是震源尺度.     

Effect of overconsolidation ratio on dynamic properties of binary mixtures of silica particles

Overconsolidated soils are ubiquitous in nature due to multiple mechanisms; however, the stress-history-based studies of small strain stiffness on binary mixtures, such as silty sand, are limited even though natural sand deposits are commonly mixtures of sand particles with varying amounts of fines. Consequently, this study quantified the stress-history-based dynamic properties of binary mixtures, such as sand-sand mixtures with different size small particles, and silty sand mixtures with small amounts of non-plastic fines, up to the critical fines content. By performing bender element tests on those mixtures according to fines content, size ratio, and overconsolidation ratio, the stress-history-based G_max of binary mixtures was evaluated. For the relevant data analysis, the OCR (overconsolidation ratio) exponent in the G_max formulation was expressed in terms of stress exponents during loading and unloading. It was found that the effect of OCR on the estimation of G_max increased with a decrease in size ratio (or increase in size difference), since the stress exponents during loading increased more significantly with a decrease in size ratio due to the pronounced change in interparticle coordination between large grains. However, the variation of stress exponents during unloading of different mixtures was relatively small due to the prevalent elastic deformation. It was demonstrated that the maximum stress history effect of tested mixed soils was observed at a fines content of approximately 5%, which was smaller than critical fines content of silty sand. This behavior was attributable to the delay in critical fines content observed during unloading, when compared to that observed during loading.     

ɽ��CORS����ʱ������������������

?????CORS??2?????????д?????????????????????????У???????CATS???????????????????CORS??????????????е??????????????з??????????????1??CORS??????????????в????а???????????????????2??CORS?????????????????N??E??U??????????????????????????￡????У?N??E????????????????WN+FN??U????????????????WN+FN+RWN;3???????CORS?????????????????????????????????????????????????????????????????????????????????     

中国大陆应变应力场研究

根据1999～2009年网络工程GPS观测资料计算得到的应变率参数,研究了中国大陆地壳的应变应力场及其地壳现今的水平活动特征。结果表明,中国大陆地壳西部青藏亚板块的压应力主方向围绕藏南和阿萨姆构造结向北、东、南依次辐射撒开。新疆亚板块自西向东由近SN向变为NE向。中国大陆东部地壳的压应力主方向自北向南由NEE变为近EW向,再变为SEE向。中国大陆主压应力作用强度西部显著大于东部。中国大陆地壳西部强于东部,南部强于北部,现今西部地壳以挤压、走滑为主,东部地壳既有挤压、走滑,也有拉张。     

Evaluation and legal assessment of certified seafood

This study presents a legal review of international treaties to derive sound definitions of overfishing. It examines seafood stocks that were certified by the Marine Stewardship Council (MSC) or Friend of the Sea (FOS). Stock size and fishing pressure were compared with the internationally agreed reference points which both organizations have accepted. No suitable status information was found for 11% (MSC) to 53% (FOS) of the certified stocks. For the stocks with available status information, 19% (FOS) to 31% (MSC) had overfished stock sizes and were subject to ongoing overfishing. An analysis of legal implications of certification of overfished stocks suggests that a certifying body cannot be held liable for a violation of internationally agreed standards unless the domestic law of its home country so regulates. States may ban the import of fish products from overfished stocks, but only in very specific cases. Possible causes for the certification of overfished stocks are discussed and recommendations are given on how the certifiers could improve their performance. The study concludes that it is still reasonable to buy certified seafood, because the percentage of moderately exploited, healthy stocks is 3–4 times higher in certified than in non-certified seafood.     

Estimation of Maximum Earthquakes in Northeast India

We attempt to estimate possible maximum earthquakes in the northeast Indian region for four seismic source zones, namely EHZ, MBZ, EBZ, and SHZ, which encapsulates the various seismogenic structures of the region and also for combined source zones taken as a single seismic source regime. The latter case exhibits a high maximum earthquake estimate of M_W 9.4 (±0.85) through Bayesian interpretation of frequency magnitude distribution with Gamma function implicating a moderate deviation from the standard Gutenberg Richter model at the higher magnitudes. However, tapering Gutenberg Richter models with corner magnitudes at M_W 8.01, 8.7 and 9.1, respectively indicated maximum values corresponding to M_W 8.4, 9.0, and 9.3. The former approach was applied to each of the source zones wherein the data are presented in parts according to the data completeness, thereof. EHZ, MBZ, EBZ and SHZ are seen with maximum earthquakes of M_W 8.35 (±0.59), 8.79 (±0.31), 8.20 (±0.50), and 8.73 (±0.70), respectively. The maximum possible earthquakes estimated for each individual zone are seen to be lower than that estimated for the single regime. However, the pertaining return periods estimated for the combined zone are far less than those estimated for the demarcated ones.