Intercomparison of Precipitation and Soil Moisture Products in a Semi-Arid Area

Abstract:

Precipitation is the driving force of the terrestrial water cycle over land, and is also one of the most common indicators of climate change. As such, it is rigorously monitored by many organizations using an array of instruments and/or models. However, disparities between the various products and observations lead to the questions: what are the uncertainties of these products? and how do these uncertainties propagate to the estimation of soil moisture?

To address the first question, we have evaluated over 20 gauge-based, satellite-based, modeled, and reanalysis precipitation products using 57 years of observations over a highly instrumented 150 km² watershed in southern Arizona. Most of the products do a better job of estimating winter precipitation than summer or annual precipitation with higher daily correlation and lower root mean square differences compared with observations. All products with the exception of models estimate the timing of the peaks in the annual cycle well, but overestimate amount with respect to the observations. The gauge and satellite products estimate average intensity of precipitation well, while the models estimate drizzle (<0.01 mm/day) too frequently. Many products reveal positive or negative trends in seasonal and annual precipitation total, while observations do not show any significant trends over the same time period.

To answer the second question, several modeled and reanalysis soil moisture products relying on the above mentioned precipitation products are compared with 11 years of in situ soil moisture measurements as well as the modeled soil moisture of Stillman et al. (2014), which extends the measurement period of summer soil moisture to 57 years. Most of the soil moisture products are significantly correlated with observations except for the models, partly due to errors in precipitation estimation. Improvements in soil moisture estimation from updated versions of ERA and MERRA are also shown.