IDENTIFICATION OF MOTION PARAMETERS FROM SPATIO-TEMPORAL ATMOSPHERIC TEMPERATURE PATTERNS L. Ju. FRADKIN and F. B. KNOX Physics and Engineering Laboratory, Department of Scientific and Industrial Research. Lower Hutt. New Zealand (Received in final form 21 June, 1984) Abstract. In near-calm conditions it is difficult to make direct measurements of atmospheric advection reliably and cheaply, particularly at many points over a large area. An alternative indirect method is examined using time-series measurements of atmospheric temperature (or any other convenient conserved field variable) at points on a spatial grid. System identification methodology is applied to analyze atmospheric temperature data obtained in such near-calm conditions during an experiment with a low-flying helicopter. A three-parameter numerical model of atmospheric advection has been identified in the temperature data, the advection parameters being an eddy diffusion coefficient and horizontal components of the wind velocity. It has been demonstrated by analysis of all the assumptions, and using simulated data, that in this case the standard least-square procedure can be applied to recover sensible parameter estimates, even though the model is of the error-in-variables type and parameters appear to change abruptly at certain moments of time (but by a reasonably small amount). The parameter estimates have been validated against independent data. 1. Introduction Much of geophysics involves modeling large continuous and incompletely understood systems on the basis of a relatively few discrete measurements in spaceand time. In this paper we will demonstrate the use of a branch of control theory known as system identification (e.g., AstrUm and Eykhoff, 1971; Young, 1978; Bennett, 1979) on such data to assess the suitability of a model and determine the value of its parameters, or to understand why this cannot be done. The data used here to illustrate the methodology are a spatio-temporal series of air temperatures measured near the ground during an experimental helicopter flight to prevent a radiation frost. The initial aim of the experiment was to get an idea of the magnitude, extent and duration of air temperature rise in the wake of the helicopter. No great precautions were taken against radiation heat loss by the thermometers to ensure high accuracy. However, later inspection of time-varying spatial contours of measured temperature (seesample in Figure 2) suggested the possibility of extracting information on overall air velocity and eddy difIiusion. It was decided to attempt to realize the above possibility. Under radiation frost conditions, air movement,though important, is often so slow asto be ditIicult to measure reliably and cheaply, cost being of particular concern when measurements are to be taken at many different points simultaneously. Thus a method of deducing a velocity Boundary-Layer Meteorology 31 (1985) 65-88. 0006-8314/85/031 l-0065$03.60. 0 1985 by D. Reidel Publishing Company.