A shortage of observational climate data—or limitations in its quality—in different regions can reduce our understanding of climate processes and weaken management capacity in meteorology, hydrology, and agriculture. Today, advances in forecasting centres, climate-data modelling, modern processors, and supercomputers make it possible to predict weather by combining numerical models with data assimilation: the incorporation of direct observations and remotely sensed measurements into model calculations.
What is atmospheric reanalysis?
 |
With sufficient computing resources, advanced numerical models can predict air temperature and other atmospheric variables over short- and medium-range future periods. Increasingly, they can also incorporate near-real-time data—observations assimilated into numerical forecasts soon after they are collected.
The same basic modelling framework can be used in the opposite temporal direction. By repeatedly assimilating observations recorded in the past—such as measurements from weather stations, radiosondes, and meteorological satellites—researchers can reconstruct atmospheric conditions during historical periods, including places and times for which direct observations are incomplete.
This process is called atmospheric reanalysis. In simplified terms, reanalysis combines historical climate and weather observations with numerical modelling to create a physically consistent reconstruction of past atmospheric conditions.
Iran's temperature anomalies
The time series shown below comes from reanalysis produced by the European Centre for Medium-Range Weather Forecasts (ECMWF). It shows Iran's average air-temperature anomaly over approximately seventy years relative to the 1981-2010 baseline period.
 |
The reanalysis output shows a clear rise in Iran's average temperature. In particular, during roughly the final 25 years represented by the series, the average departure from the baseline temperature becomes increasingly positive.
This example demonstrates the usefulness of reanalysis products for studying long-term atmospheric change where observational records alone may be incomplete.