双亮度差方法测量白天气象能见度时的误差分析

对数字摄像能见度观测系统采用双亮度差方法测量白天气象能见度时非标准观测条件引起的误差进行理论分析, 并给出了相应的误差控制措施。结果表明:对于正南、正北方地平线附近天空亮度的分布, 除了天空布满均匀云层和晴天的中午较为均匀以外, 其他时刻垂直相对梯度都较大, 通常在4%/°～10%/°范围内, 而水平相对梯度大多较小, 低于2%/°;不同组的目标-背景视线方向不一致是不容忽视的一个误差源, 为了限制和最大程度地减小其不利影响, 最好选择具有相同视线仰角的目标物和天空背景, 且视线方位角的差异越小越好;可对目标物两侧的天空背景的亮度进行测量, 内插获得与目标物视线方向一致的天空背景的亮度;视线方向上气柱照明不均匀会导致测量误差且人工无法调控, 应尽量选择下垫面特性较为均匀的观测场地, 通过多次采样平均在一定程度上能够减小其影响, 也可实时监测视野范围内场地和天空的亮度分布, 识别较差的照明状况以免获得误差较大的观测记录;利用树木、草丛、山体、墙壁和窗户等暗目标物难以满足双亮度差方法对观测条件的要求, 最好采用反射率足够低的人工实用黑体目标物。

Error Analyses of Daytime Meteorological Visibility Measurement Using Dual Differential Luminance Algorithm

The theoretical analyses of measurement error caused by non-standard viewing conditions, and the corresponding error-control methods through using a Dual Differential Luminance (DDL)algorithm in daytime meteorological visibility measurements are presented based on digital photography.It show s that :(i)the vertical relative gradients of sky luminance near due south and north horizons are mostly in the range of 4%/°～10%/°bigger than horizontal ones, which are mostly less than 2%/°, except at noon in sunny day or w hen the homogeneous cloud layer spreads over the w hole sky.The deviation angles between sight paths of different target-backg rounds can bring much bigger measurement errors and their effects should no t be neglected.The best solution is to set targets with same viewing elevation angle and to restrict the viewing azimuth angle deviations as small as possible during the operations of Digital Photography Visiometer System (DPVS).With the important advantage of matrix imaging technique, there is the possibility to further reduce their effects if the sky luminance along the same sight path of a given target is interpolated from the luminance distribution of the sky at both sides of the target.(ii)The nonuniform illumination along the sight path is also a measurement error source, which can' t be cont rolled manually.It is preferred to choose a site field with more uniform surface.And the error can also be reduced to a certain ex tent through averaging multi-samples.It is feasible to avoid result s with muchbigger error by monitoring the luminance distributions of site and sky and recognizing worse illumination conditions in real time.(iii)Observing some dark target s, such as trees, grasses, mountains, w alls and window s etc., can' t satisfy the demand of DDL measurements.I t is an optimal solution to adopt artificial black target s which have enough low reflectivity.