| description abstract | Vertical gradients are usually calculated by a finite-difference approximation, where the level of interest is midway between the levels at which measurements are made. When this approach is used, considerable error may occur if the gradient varies with height, i.e., if the profile is not linear. Near the ground, non-linearity is the rule rather than the exception. The finite-difference approach may still be used, but measurements must be made at heights determined by the form of the profile, and not at heights equally distant from the level of interest. A set of charts is presented showing the heights at which the sensors must be mounted to give the gradient at the level of interest, and the height to which the gradient as usually measured actually applies. Correction factors are derived so that the gradient at any level may be determined from measurements at any two heights, provided the form of the profile is known. Use of this technique eliminates one source of error in comparison of gradient measurements from different locations, and in determination of parameters in which one or more gradient measurements at a predetermined level are required. | |
