In recent years meteorological services everywhere have expanded considerably, and meteorology as a science may be said to have come of age. Following the inauguration of regular air services in New Zealand in 1935, the number of reporting stations and the frequency of reports were greatly increased, while measurements of upper winds were introduced at selected stations, using small hydrogen-filled balloons tracked by theodolite. This method was effective only in the absence of low cloud, but the more recent use of radar for tracking enables the winds to be measured up to 50,000 ft and higher in all weather conditions. It also became possible to measure the physical properties of the atmosphere through the development of reasonably priced, lightweight radiosondes which automatically transmit pressure, temperature, and humidity while carried aloft by balloon. Daily radiosonde ascents commenced at Auckland in 1942; the present network consists of seven radiosonde and nine radiowind stations. As large rain-drops are able to reflect ultra-short radio waves, radar equipment is also used for locating and tracking areas of heavy rain, such as fronts and thunderstorms.

Regular upper-air sounding have enabled meteorologists for the first time to study, in their true three-dimensional setting, the physical and dynamical processes of the atmosphere which produce what we recognise as weather. Although the surface weather map has lost none of its former importance, it is now supplemented with several upper-level charts showing the height contours of the 700, 500, and 300 millibar surfaces, similar to isobars at about 10,000, 18,000, and 30,000 feet, and bearing a similar relationship to the wind flow. A thickness chart is also drawn for the layer 1,000–500 millibars.

The air flow at high levels is usually quite different from that at sea level; at times, it may even be almost opposite in direction. The upper-level charts reveal that many of the pressure systems that appear at the earth's surface are relatively shallow. On the other hand, the circulation of air around a depression may sometimes be stronger at high levels than at low levels. Narrow belts of extremely strong winds are often found in the 30,000–40,000 ft layer. Within these “jet-streams”, as they are called, winds often exceed 100 knots and may reach 200 knots or more. These naturally create a special problem in forecasting winds for high-flying aircraft.

The differences between the flow at low and high levels in the atmosphere often help to explain features of the weather that are not adequately made clear by the sea-level analysis. Furthermore, a number of semi-empirical rules have been found, relating the contour patterns on the upper-level and thickness charts to future changes in the weather systems. For example, a shallow system usually moves in the general direction of the winds blowing over it at high levels. Such rules are applied by the forecaster in the preparation of a “prognostic chart” representing the weather map as he expects it to appear, usually 24 hours ahead. In parts of the Northern Hemisphere, where upper-air stations are much more numerous, prognostic charts are now prepared automatically with the aid of large, very fast, electronic computers. The computer takes the latest observations and rapidly performs many thousands of intricate calculations to produce the expected values for the required period ahead. Even a computer can, at present, solve only an approximate form of the dynamical and physical equations, and the resulting prognostic map cannot be perfectly accurate. Moreover, the human forecaster is still required to bridge a very large gap in translating the prognostic map into terms of expected weather. Numerical prediction methods have been tried experimentally in New Zealand with promising results, but lack of upper-air data from the surrounding oceans, and restricted computer facilities, at present limit application on a routine basis.

Further developments in numerical prediction methods can confidently be expected in the future, but it seems unrealistic to expect that one day it may become possible to compute an accurate weather almanac for the following year, similar to the Nautical Almanac. The inherent instability of atmospheric motions seems to rule out this possibility.