A new technique for measuring global rainfall

Precipitation is a vital element for life on Earth. Agriculture and the food supply depend upon the global distribution of precipitation, so a knowledge of when, where and how much rain falls is of paramount importance to society, but excessive amounts lead to flooding, loss of life and damage to property. We need to improve weather forecast models so they can better predict when and where heavy rain is likely to cause flash floods and any mitigating actions can be focused on areas at risk. We also need better confidence in the ability of climate models to predict changes in global rainfall patterns so that long term policy decisions are better informed.

Global climate and weather forecasting models have a resolution of several km and each model 'grid-box' (size 1km or greater) can have just two or three variables expressing the properties of the clouds in the grid box. The individual collisions between cloud particles to produce precipitation cannot be modelled, but instead the rate of conversion of cloud water into precipitation is approximated or 'parameterised' in terms of the large-scale variables such as the mass of cloud water per cubic metre within the grid box. We know that these parameterisation schemes are imperfect, and need global observations of rainfall to check how well these models capture the statistical properties of the rainfall in the present climate so we can identify when and where the schemes are failing and how they could be improved.

It is surprisingly difficult to measure global rainfall. Rain gauges have been in use for hundreds of years, but they are only a point measurement and are restricted to land. CloudSat, launched in 2006, provides the best estimates of global rainfall and these data have been used for model validation. The technique relies on the fall of the radar signal from the ocean surface caused by the attenuating rain, the so-called PIA ('path integrated attenuation') method, but direct validation is difficult.

We propose implementation of a new 'Gradient' technique that derives the rain rate directly from the gradient of the radar reflectivity profile that results from the attenuating rain and has two unique advantages: a) the error in the rain rate can be estimated from the goodness of fit of the profile to a straight line, and b) exactly the same algorithm has been used by 94GHz radars on the ground where it has been validated by co-located rapid response rain gauges.

Initial tests show that the rainfall derived from the new 'Gradient' method is significantly greater than values from the PIA technique, so the first task is to reconcile these differences by analyzing the assumptions made by the PIA method. Next, we will refine 'Gradient' method and its errors by analysing and validating more CloudSat and ground-based 94GHz rainfall observations.

A new global rainfall data set with quantified errors will be made available to the science community. In collaboration with climate and weather forecast modellers, the observed geographical and seasonal variations in rainfall statistics will be compared with their representation in the models to identify when and where the parameterization schemes have shortcomings. To predict any future global warming we need to understand the current balance of incoming solar and outgoing thermal infra-red radiation; this current balance is also sensitive to any changes in the energy transported by the mean global precipitation that should be revealed by the new CloudSat estimates.

The 94GHz radar on the EarthCARE satellite (launch 2022) has an additional Doppler capability. We will use the ground based 94GHz Doppler data to establish if the Doppler on EarthCARE can provide improved rain rate estimates. We will also examine how future scanning 94GHz radars could provide a larger sample of rainfall; potentially such data in near real-time could be assimilated in near real-time to further improve forecasts of heavy rainfall.

Precipitation is vital to our civilization. Agriculture and the food supply depend upon the global distribution and frequency of precipitation, so a knowledge of when where and how precipitation patterns may alter in a future warmer climate can inform policy decisions on how to adapt to such changes. Government, businesses and the general public will also benefit from improved short-term weather forecasts of heavy precipitation likely to cause flooding so that mitigating action can be taken to minimise damage to property and loss of life. This project will use CloudSat (2006-2009) radar data to derive rainfall rates with quantified errors over the globe that can be used to improve the forecasts produced by the two types of model:
a) Climate models that are run for many years or centuries in a statistical sense to predict long-term changes in climate and
b) Weather forecast models that are initialized by the current state of the atmosphere to provide specific short-term weather forecasts for a few days or weeks ahead.

Weather services such as the Met Office and ECMWF will benefit from improved observations of global rainfall that will enable them to validate that both their weather forecast and climate models are faithfully representing the statistics of rainfall in the current climate. The resolution of both climate and weather forecast models is at least several km so they cannot represent the individual interactions between cloud particles that causes them to become raindrops or large falling ice particles, instead the rate of conversion must be 'parameterised' in terms of large-scale variables such as the cloud water content. These parameterisation schemes are known to have imperfections. The first step is to verify that these schemes can, for the present climate, simulate the correct statistical distribution of precipitation occurrence and intensity, its spatial extent and evolution, and how this differs both geographically and seasonally over the globe. The global climatology of rainfall characteristics supplied by this project will provide a benchmark against which these forecasts can be compared and shortcomings in the parameterisation schemes can be identified and rectified.

Weather forecast models such as those at the Met Office and ECMWF will benefit from real time global observations of rainfall to initialize their forecast models with the best representation of the atmosphere at the present time, so that when the model is set running it will give a more reliable forecast of the weather in the next few hours and days. To achieve this, the real time global observations must have an associated error so they can be used to 'nudge' the model with an error-dependent weighting, so that, after 'assimilation' of the new data, the state of the atmosphere in the model better mirrors reality, and when run forward in time should provide a more reliable forecast of when and where heavy rain is likely to occur.

Emergency services and the Environment Agency (who are responsible for issuing flood warnings) will benefit from the improved weather forecasts so they can more precisely target mitigation activities on areas identified to be at risk. The public can then take action - such as moving furniture upstairs, with less risk of false alarms.

The European Space Agency will benefit from the algorithms developed in this proposal to derive accurate rain rates with quantified errors from the CloudSat mission as they can then be applied to EarthCARE, due for launch in 2022 with a Doppler capability. This project should establish how the extra Doppler constraint can improve the accuracy of the rain-rate retrievals. ESA are considering concepts with scanning 94GHz radars; this project should provide evidence of the quality of the rain retrievals to be expected for such broad swath concepts. Assimilation of data from such broad swath radars would have a further significant positive impact on rainfall forecasts