Begin with a moving measurement window
The Global Precipitation Measurement mission combines a joint Japanese–American Core Observatory with partner satellites carrying microwave radiometers. Each spacecraft observes along its orbit, so a complete-looking global map cannot be understood as one instantaneous photograph. JAXA describes the central design as combining the Core Observatory's detailed measurements with the more frequent coverage supplied by the wider constellation. The first question for any rainfall pixel is therefore how an observation, a calibration relationship and a gap-filling method contributed to its estimated value.
Sources: gpm-jaxa
The radar supplies a vertical reference
Japan's Dual-frequency Precipitation Radar uses Ku- and Ka-band signals to examine precipitation's three-dimensional structure. The two frequencies bring different sensitivities to the measurement, improving information about weak precipitation as well as stronger systems. Its role extends beyond drawing vertical slices through storms: observations help anchor precipitation estimates made with the microwave imager carried on the same spacecraft. That connection allows the detailed radar information to influence a much larger observing network, although it does not turn every constellation pixel into a direct radar sounding.
Sources: gpm-jaxa
The microwave imager makes the bridge
NASA's GPM Microwave Imager has 13 channels spanning approximately 10 to 183 gigahertz. They measure microwave energy naturally emitted by, or affected by, precipitation and associated atmospheric conditions. Its approximately 885-kilometer swath is wider than the radar's, with an overlapping central region where the two instruments can be compared. Different channels respond to different aspects of rain, frozen precipitation and water vapor. These overlapping observations help improve radiometer-based retrievals and make the imager a reference for other constellation instruments.
Sources: gpm-gmi
A gridded product reconstructs the gaps
NASA's IMERG Version 07 documentation describes a further processing chain. Estimates from passive-microwave sensors are placed on a grid, intercalibrated against a combined radar–radiometer product, and merged with infrared-derived information using time interpolation. The resulting fields have half-hour time steps and 0.1-degree grid spacing. Those are product coordinates, not a claim that a microwave sensor directly visited each cell every half hour. This distinction is central to reading the map: a regular output schedule can contain irregularly timed observations and calculated values between them.
Sources: gpm-imerg-technical
A half-hour rain rate is still a rate
IMERG documentation describes values based on a microwave snapshot or an interpolated estimate when microwave data are unavailable. They are expressed in millimeters per hour, even though the analysis interval is half an hour. Under the documented whole-interval assumption, a value of 2 millimeters per hour corresponds to 1 millimeter over that half hour. This simple example illustrates a frequent interpretation trap: the number's units and its time interval are separate pieces of information. A map's timestamp alone does not establish an accumulated depth.
Sources: gpm-imerg
Later analysis can change the answer
IMERG's Final processing incorporates monthly gauge information where suitable observations are available. That adjustment helps constrain the longer-term amount, but gauge coverage is uneven and most ocean regions lack comparable gauge input. A revised historical pixel is consequently not evidence that rain happened twice or that an earlier map was a direct reading later erased. It reflects another stage of estimation. Comparisons should retain the product version, processing run and averaging interval, because those choices determine which observations and adjustments are represented.
Sources: gpm-imerg