Estimating Surface Precipitation from Space (Invited Presentation)

Precipitation is the ultimate source of fresh water around the globe, so it is labeled as an Essential Climate Variable (ECV) by the Global Climate Observing System (GCOS).  As such, precipitation is a key observation across a range of scientific and societal benefit areas.  This starts with fundamental understanding of the microphysical precipitation process; how it scales up to the global water and energy cycles, including the effects of climate change; and informing the representation of precipitation in numerical models for weather forecasting and climate.  The practical uses of precipitation data includes analyses of localized flash flooding, landslides, and transportation disruptions; input for forecasts of agricultural yields, disease outbreaks, water resources, and drought; input to decision-making in disaster relief agencies; use in microinsurance programs in developing countries; application to developing/reviewing building codes and civil engineering standards; and use as a control variable in other fields, including political economics, ecology, and public health.

Observations by surface gauges and surface-based weather radars are the conventional way to obtain precipitation data, but satisfactory coverage is only available over some land areas.  Satellite observations are critical for providing fine-scale observations over the vast expanses of the oceans and the remaining land areas.  Space-based radars on Global Precipitation Measurement (GPM) mission Core Observatory, Tropical Rainfall Measuring Mission (TRMM), and CloudSat provide key (but limited) observations.  The workhorse is the passive microwave (PMW) radiometer, providing precipitation-relevant information in channels between 6 and 183 GHz.  Ideally, a precipitation satellite has both a radar and radiometer, such as in TRMM and GPM.  Finally, infrared and visible (IR/VIS) sensor data can be used for precipitation estimates, but with lower quality results.

The constellation of precipitation-relevant sensors has a relatively long heritage and heavier investment by the national and international agencies due to the importance, and because the sensors are typically applicable to a range of Earth science variables.  The current configuration of a precipitation radar in a precessing low Earth orbit (LEO), about ten “modern” PMW satellites in LEO, and five geostationary IR/VIS satellites mostly satisfies the minimum sampling requirements for many global precipitation-related applications.  However, one unmet need is additional relevant PMW channels in the range 100-200 GHz to better define light and solid precipitation.

Progress in precipitation algorithms has allowed us to demand more from retrievals and their accuracy.  PMW retrievals oriented toward rainfall, especially over tropical oceans, have a three-decade history and are reasonably mature, while those specifically addressing problematic precipitation regimes and surface conditions have a much shorter history and still require further algorithm development.  In particular, retrieving light precipitation and falling snow is still challenging, which are important in specific regions for rainfall, and in nearly all cases of snowfall, making continued algorithm development across the range of climate regimes a second key unmet need.  Once the retrievals are computed for the sensors, it is “obvious” that their individual overpasses need to be combined onto a uniform space-time grid.  Reaching back almost three decades for Global Precipitation Climatology Project, and increasingly over the last decade, multi-satellite algorithms have been developed that merge the individual sensors’ estimates.  Key results include the need for intercalibration across the estimates, and the importance of using surface observations to control bias in the satellite products.  Thus, a third unmet need is improved access to more precipitation gauge data, with shorter latencies than are now available to fully realize the critical time-dependent surface calibration of the satellite estimates.  A final unmet need is the routine ability to assimilate precipitation estimates into numerical models.

In the near term, there is concern in the precipitation community that the launch schedule for new precipitation-relevant PMW sensors to replace the aging fleet is not likely to keep up with sensor failures.  This raises the possibility that the quality of products available for research and societal use might suffer.  In the longer term, new technologies are being explored that provide PMW sensors and satellite concepts that are lower power and lighter weight.  As these are demonstrated and scaled up to meet the channel and resolution requirements for quality precipitation retrieval, the national and international agencies should be able to coordinate a stable, capable constellation.  Another possible technology is PMW in geosynchronous orbit, which could provide excellent sampling, but currently presents technical challenges.  Multi-sensor mission concepts are also being considered, such as the Clouds and Precipitation Processes Mission, to more directly address the precipitation processes that vary from one weather regime to another.

Looking to the future for algorithms, continued development, particularly for snowfall, a focus in GPM, will continue to yield improved estimates from current-generation instruments.  Furthermore, the modern paradigm of systemically processing the entire legacy record of precipitation-relevant satellite data with new algorithms allows users to improve their understanding of the entire time series of global precipitation.  Finally, it is planned to augment the observation-only multi-satellite datasets with datasets that combine observations and numerical model estimates.  Satellite estimates at low latitudes characteristically show much more skill than model estimates, while the opposite is currently true at high latitudes.  The balance between the two is regime-, algorithm-, and model-dependent, and requires research to develop the appropriate combination approach.  The goal is a hierarchy of precipitation datasets that allow different users to employ specific sensors, the entire multi-satellite constellation, or the “best” satellite-model estimate according to their application.

There has been tremendous progress over the last three decades in precipitation sensors and algorithms, but there is still significant room for enhanced capabilities in equipment and data sets.