Land-based observations are collected from instruments sited at locations on every continent. They include temperature, dew point, relative humidity, precipitation, wind speed and direction, visibility, atmospheric pressure, and types of weather occurrences such as hail, fog, and thunder. The bulk of these observations are carried out by National Meteorological Services as part of the WMO World Weather Watch, which networks the observing stations to national, regional and global weather and climate prediction centres 24 hours a day in real-time.

The World Weather Watch collects meteorological, climatological, hydrological and oceanographic data from over 15 satellites, 100 moored buoys, 600 drifting buoys, 3 000 aircraft, 7 300 ships and some 10 000 land-based observation stations. This data has to be comparable and up to standards in order to be usable by the prediction centres in the numerical weather prediction models that produce daily weather forecasts and early warnings for natural hazards such as hurricanes. Thus, the World Weather Watch also produces the standards for measurement of the data collected.

Further reading: Reanalyses and Observations: What’s the Difference?

Re-analysis data is a gridded estimate of the state of the atmosphere at a certain time. To do this, data from many sources are assimilated and includes both observations from a variety of sources ground based stations, ships, airplanes, radar and satellites—and forecasts from numerical weather prediction (NWP) models. It is used to produce the analysis of current conditions that serves as the starting point for the next NWP forecast cycle; from gappy observations, the assimilation system delivers a complete gridded state estimate that provides values (initial conditions) for all NWP model variables at all grid points. Since the 1990s, data assimilation also has been used to construct long-term datasets for use in climate and other research, in a process known as retrospective analysis, or reanalysis (Trenberth and Olson 1988; Bengtsson and Shukla 1988). Reanalysis involves performing data assimilation for past periods, using a current NWP model and data assimilation method and some or all of the data that are now available for those past periods. It produces a long sequence of comprehensive snapshots (analyses) of atmospheric conditions—a reanalysis dataset.

Further reading: Reanalyses and Observations: What’s the Difference?