Diurnal to decadal variability of precipitation intensity, frequency, and amount over three areas in North America

Precipitation is a crucial quantity for weather and climate studies. It can be measured by rain gauges, surface-based radar, and satellite remote sensing. It can also be estimated through reanalysis and numerical modeling. Because of its substantial temporal and spatial variability, it is challenging to compare various precipitation products. In this presentation, we will discuss relevant issues based on observational data over three areas in North America.

Using hourly precipitation data from 88 rain gauges for 57 years over a 150 km2 watershed in southern Arizona, the diurnal to decadal variability of precipitation intensity, frequency, and amount has been analyzed. For instance, precipitation intensity does not show any trend for the 57-year period, even though such a trend was found in an earlier study using 44-years of data that were available then. Furthermore, these observational data have been used to quantify uncertainties of more than 20 gauge-based, satellite-based, modeled, and reanalysis precipitation products.

Using these data (over an area representing a typical regional model grid box) and those from radars and 120 rain gauges over Ohio (representing a typical global model grid box), the different number of gauges or 4 km radar pixels in realistically representing area-averaged precipitation intensity, frequency, and amount has been identified. For instance, the number of gauges or radar pixels needed for realistically estimating area-averaged precipitation intensity is much greater than that for precipitation amount.

In situ precipitation data over the core area of the North American monsoon region has also been used to evaluate the capability of more than 20 CMIP5 climate models in simulating the monsoon precipitation seasonal cycle. For instance, a relative criterion would better facilitate the definition of monsoon onset and retreat in these models than an absolute criterion.