Weather Observations and their use in NWP

Like medicine, a weather prognosis needs a diagnosis based on observation. In the case of weather this involves a complex system of measurements of the atmosphere from terstrial and space based systems.

For millennia, the only data were what could be seen from your location. Is the morning sky red? Is high cloud spreading from the northwest? As my page on Single Observer Forecasting points out, such techniques are limited.

After the late 1870s forecasters used synoptic charts with plots of observed values. Initially, these were only from places on land. Then, W/T allowed data to be obtained from ships on passage. Then radio techniques made it possible to send up balloons and get measurements throughout the atmosphere.

Since the early 1950s, technology has advanced greatly in computing, instrument design both in situ, and remotely sensed, data. This has led to the rapid evolution of meteorological observing techniques necessary to fee NWP models..

Over land weather observations are made for various purposes such as aviation, shipping, agriculture, offshore exploration etc. Sometimes the observations are made entirely by a human being reading instruments and describing what he or she sees but, increasingly, observing is being automated.

Fairly obviously, there are no great problems in automating instruments such as barometers, thermometers and anemometers. There are now instruments that can measure visibility, detect lightning, estimate cloud base, cloud amount and, even, "weather".

Over the sea, and dating back to the mid-19th century, there are routine reports from ships. Again, these are being automated although this may reduce their value. In particular, it is nigh impossible to site an anemometer on a ship because of the superstructure. Experience officers-of-the-watch have, over the past produced usefully consistent estimates of wind.

Ships tend to follow well worn routes and leave many areas with few reports. In busy shipping lanes, pilotage must take precedence over weather observing. Around the continental shelf This has been tackled by the use of robust automatic weather stations. These are also used on remote islands and some oil or gas rigs make such data available for general use.

Data gaps over the open ocean are partially filled by drifting buoys measuring, usually, surface pressure and sea temperature only. These are often launched in data sparse areas and produce much useful , albeit limited, data.

Around the continental shelf and on remote islands, use has been made of robust automatic weather stations. These are also used on some oil or gas rigs and the data made available for general use.

The ECMWF produces a list of data used in its predictions. The first diagram shows locations of weather reports from land ships at sea. "SYNOPS" refers to reports from lands station; "METARS" are abbreviated reports from airfields and "SHIPS" are reports from ships on passage.

The geographically uneven distribution of data will be obvious. Gaps in areas where there are few ships, and especially in the Southern Ocean, are filled by drifting buoys. Around the continental shelf, tethered buoys provide comprehensive reports.

Ever since the late 1930s/early 40s radios-sonde balloons have been used to measure temperatures, humid ties and winds from ground level up to about 12 km generally and sometimes up to 30 km. With the exception of a handful of ships, these, as the diagram show, are heavily weighted towards land masses and some islands. There are large gaps over the oceans and southern hemisphere generally .

Some of these gaps are partly filled by aircraft reports. These are, for the most part at flight level, typically 10 to 12 km above the surface but some are also from the ascent/descent phases of a flight. The diagram shows that the southern hemisphere is still not well covered. There are three types of report shown but the differences are mainly in the communication route by which they are received.

These are of two broad forms; first, used for many years now, are visual and infra-red instruments. The most simple are, essentially cameras that produce pictures of whatever the satellite can "see". The infra-red sensors, measure the temperature of the top of any cloud or, if there is no cloud in the beam, the surface of the earth.

The atmosphere is transparent to some of these wavelengths and absorbs others. The absorption depends on the pressure, temperature and water content. By looking at the sea surface at several frequencies, information can be gained on the variation of temperature and water vapour in cloud free air.

Micro-wave instruments sense radiation at frequencies between about 20 and 200 GHz and signals are received from troposphere and stratosphere. The information relates to temperature, liquid water and water vapour.

Active sensors

These transmit a beam and the return signal is measured. A principal function is the estimation of surface winds by the scattering of a radar beam fired at the sea - a scatterometer. The figures here show the tracks of a satellite carrying this instrument and some of the estimated winds.

• Advantages:
  • Wide space coverage (60 to 65 degrees from the sub-satellite point)
  • Very high temporal coverage (every few minutes)
  • Suitable for meteorological feature tracking (Atmospheric Motion winds)
• Drawbacks:
  • Spatial coverage limited to the disk (need for constellation)
  • Cannot observe the Polar Regions

Shows the geographical coverage from the current constellation.

Particularly useful is their use to "see" weather move and measure its movement. In the early space era this was done by human beings identifying small cloud features and, literally, using a ruler on a picture. Nowadays it is done by pattern matching on the assumption that, for short period, there will be little change. Using infra-red imagery, this can be done at any time. The technique can be used with infra-red pictures of cloud as well as areas of water vapour that the satellite can detect.

Wind data can also be derived from images in the visible spectrum but, of course, only during daylight hours. In all cases, that data can only refer to the top of a cloud or water vapour layer. The advantage of using visible light as well as infra-red is that the former can be detected through the latter.

Low Earth Orbiting satellites

• Advantages:
  • Cover the whole earth after several cycles (near polar orbiting satellites)
  • More suitable to sound the atmosphere in the microwave spectrum.
• Drawbacks:
  • Moderate temporal sampling (several hours to go back to the same point)
  • Requires constellation to ensure a reasonable temporal sampling

Data coverage depends upon how many satellites are carrying a particular instrument. An instrument known as AIRS (Advance Infra Red Sounder) is only active on one satellite. Its coverage is

.

AMSU-A (Advanced Micro-Wave Sounding Unit is active on several satellites. Its data coverage is

One of the major problems to those using the data in NWP is that sounding from space measure the effect of the weather on the atmosphere rather than the atmosphere itself. The meteorologist would like data in the usual form of temperatures, humidifies and winds. What he gets are values of radiation that depend upon temperature and humidity. Surface wind data are estimates based on scattering of a radar beam.

The nearest to “real” data are the cloud winds but even these cannot be precise because clouds are continually changing and the

 pattern matching contains consequent errors.

The UK Met Office (and others) approach to the assimilation of satellite data is to treat radiance data as just another variable that can be predicted by NWP models and analysed like other, conventional, weather data. However, despite their ever increasing sophistication, satellites can only provide estimates over areas rather than at specific locations.

The areas are usually fairly small - tens of km – substantially larger than the detail that we sailors observe at sea. Nevertheless, studies have shown that satellite data make significant and worthwhile improvements to NWP.

This raises questions of the predictability of small weather detail. If it cannot be measured, how can it be predicted? It seems fairly clear that those forecasts based on the finest detail available are more likely to succeed than those that do not use detailed data input to their models. National Weather services eg the UK, Météo France, the HIRLAM countries, the US all use the most detailed data available. As far as I can determine, the same cannot be said of private sect6or companies running detailed, meso-scale models.

The whole process of data collection, quality evaluation and assimilation into NWP models is a massive task involving some “heavy” mathematics. From a nominal data time of 00 UTC, it takes about three hours before the computer can begain its prediction calculations. It takes just about as long again, or even les, to compute a 5 day global forecast.