Principal Investigator: Christopher Kidd, University of Maryland
The measurement of precipitation in the form of rainfall and snowfall is of significant value to our society and the environment we live in, as well as vital for all life on Earth. Conventional means for measuring precipitation worldwide are generally inadequate due to spatial or temporal sampling issues. Consequently, satellite-based observations of precipitation are necessary to provide a global picture.
Much effort has been expended in developing and refining precipitation techniques that can be applied globally. However, the number of routinely available retrieval techniques has diminished significantly over the recent years, restricting the diversity of products available to the user. While it is certain that some products often have undesirable side effects, the techniques used to derive such products have a sound physical basis that can be used to identify changes or trends in precipitation that may be a result of our environment or from artefacts in data records; these changes may be hidden in other more complex precipitation records.
A number of climate-length satellite-based data records are being produced based upon passive microwave observations. However, at present there are no long-term satellite-based precipitation benchmark products against which improvements in retrieval schemes can be measured; complex retrieval techniques often rely upon external datasets which may change over time, sometimes producing unforeseen and unexpected results. Such changes may also be subtle, making their source difficult to ascertain; through comparisons with benchmark products such changes will be easier to identify, along with any step changes in the original input data for such products.
We will use these climate-length datasets to generate precipitation data records based upon a number of well-founded retrieval techniques. Inter-comparison of precipitation products through past studies will be used to identify suitable techniques, which are typically “operational,” if not implemented as such. Such techniques are known to perform well at the local scale and may be used to monitor the long-term performance of not only the more complex precipitation retrieval schemes, but also of the input data set as well.