声速剖面反演预测方法

利用2006—2017年我国南海部分区域(112°～114°E,10°～12°N)的Argo观测数据,对该海区声速剖面进行了仿真分析和研究。在此基础上,利用遗传算法(GA)优化的径向基函数(RBF)神经网络建立反演预测模型(GA-RBF),结合海区表面实测温度和历史数据,研究了该区域2016—2017年的声速剖面实时预测情况,并获得该海区6月和12月的声速剖面平均均方根误差值为0.845 m/s和0.815 m/s;而采用平均声速剖面方法获得该海区6月和12月的声速均方根误差分别是2.393 m/s和2.176 m/s。仿真结果表明:基于GA-RBF网络模型并利用海区表面实测温度的反演预测结果更趋近实测声速剖面,该模型可用于海区垂直声速剖面的实时预测。     

不同天气条件下微波辐射计和风廓线雷达探测数据误差特征分析

为探讨微波辐射计和风廓线雷达探测数据的准确性和可用性,利用天津全运会期间获取的GPS探空资料,分析不同天气条件下微波辐射计探测温湿度、风廓线雷达测风的误差特征.结果表明:晴天、云天和雨天条件下,微波辐射计反演低空温度廓线效果均较好,反演高空温度廓线误差较大,云天条件下,反演的整层温度廓线与探空实测值相关性最优;3种天气...     

区域四维变分资料同化的数值试验

针对中尺度数值预报模式预报误差的主要来源,尝试利用四维变分资料同化的方法来改善预报效果。在已建立的中尺度模式（ＭＭ４）四维变分资料同化系统基础上,进行了若干数值试验,通过比较同化前后的预报来检验同化的效果。这些试验中初始场、模式误差和侧边界条件被分别或同时作为控制变量来进行调整,主要探讨了模式误差和侧边界条件对同化及预报的影响,以及同时结合两者或三者的途径和方法。对两组个例分别进行的试验结果表明,区域中尺度模式预报误差除了来源于初始误差外,模式误差、侧边界条件也有不可忽视的作用。同化时应同时考虑初始场、模式误差和侧边界条件这三方面的共同作用,仅修正其中某一个或某两个会把由于其它方面造成的预报误差转嫁到它们之上,从而出现尽管目标函数下降很快而预报结果并没有相应改善的现象。     

Error Calculation in Maritime Delimitation between States with Opposite or Adjacent Coasts

The sea boundary between two states with opposite or adjacent coasts is defined by the median line between baselines realized on shore lines. As defined in UNCLOS (United Nations Convention on the Law of the Sea), every point of the median line must be equidistant from the nearest points on the baselines from which the breadth of the territorialseas of each ofthe two states is measured. The coordinates of breakpoints of the median line are computed based on known coordinates of baseline points on the shores. What is the precision of computed coordinates? Here we study the error propagation on two reference surfaces: (1) plane and (2) sphere. Moreover, four approaches to breakpoint determination are investigated. Algorithms for computation of the covariance matrix of median line breakpoints both on sphere and plane were developed.     

数字水准仪的原理及误差分析

本文论述了数字水准仪（Ｌｅｉｃａ ＮＡ３００３和ＺＥＩＳＳ ＤＩＮＩ１２）的基本测量原理和误差来源,并与光学水准仪的误差进行了比较,通过误差分析,对数字水准仪的使用提出了一些建议。     

全站仪在数字化测图中的误差分析

论述了全站仪在数字化测图中的综合误差以及对测量成果的影响大小,并对其作了详细的探讨和分析。     

Analysis of inversion errors of ionospheric radio occultation

The retrieved electron density profile of ionospheric radio occultation (RO) simulation data can be compared with the background model value during the simulation and the inversion error can be obtained exactly. This paper studies the inversion error of ionospheric RO through simulation. The sources of the inversion errors are analyzed. The impacts of measurement errors, such as the errors in phase measurements and satellite orbits, are very small and can be neglected. The approximation of straight-line propagation introduces errors at the height of the F1 layer under solar maximum condition. The spherical symmetry approximation of the electron density distribution is found to be the main source of the inversion error. The statistical results reveal some characteristics of the inversion errors. (1) The relative error increases with enhanced solar activity. (2) It is larger in winter than in equinox season, and it is smallest in summer. (3) For all seasons, it is smaller at middle latitude than at other latitudes. (4) For all seasons and geomagnetic latitudes, it is smaller at daytime than at other times. The NmF2 of the ROs from COSMIC are compared with the measurements of ionosondes, and the relative differences show the same dependencies on season, geomagnetic latitude and local time, as the relative errors of the simulated ionospheric ROs.

An uncertainty model for deep ocean single beam and multibeam echo sounder data

Comparing single beam and multibeam echo sounder data where surveys overlap we find that: 95% of multibeam measurements are repeatable to within 0.47% of depth; older single beam data can be at least as accurate as multibeam; single beam and multibeam profiles show excellent agreement at full-wavelengths longer than 4 km; archival sounding errors are not Gaussian; 95% of archival soundings in the northwest Atlantic are accurate to within 1.6% of depth; the 95th percentile error is about five times greater in pre-1969 data than in post-1968 data; many of the largest errors are located over large seafloor slopes, where small navigation errors can lead to large depth errors. Our uncertainty model has the form σ ² = a ² + (bz)² + (cs)², where 2σ is approximately the 95th percentile error, z is the depth, s is the slope, and a, b, c are constants we determine separately for pre-1969 and post-1968 data.     

泊松分布下的非线性拟合

对一数据涨落服从泊松概率分布的实验数据组进行拟合,如果拟合函数是非线性的,常常只能采用某些近似方法。近似性之一,表现在将拟合函数线性化,或者采用某种参数寻优的方法。近似性之二,则是将泊松问题近似地看作高斯分布问题。本文从最大或然法出发,对两类不同概率分布情况下的曲线拟合进行了比较,给出了对泊松分布问题更严格的求解途径。对于特殊的非线性拟合函数y(x)=ae~(bx~N),直接给出了迭代解。本文对曲线拟合的“面积亏损”问题,也进行了讨论。     

抗差估计在粗差探测及平差计算中的应用

常规的粗差深测技术，不仅辅助计算工作量大，而且将粗差探测与平差计算分两步进行。本文根据抗差估计原理，通过对观测值附加粗差的办法，制订不同的方案，分别进行抗差估计和ＬＳ估计。