John Tuttle

Associate Scientist
TIIMES - MMM
BEACHON - WCAS

Contact Information:
PO Box 3000, Boulder, CO 80307-3000
Office: FL3 - 2026
Telephone: 303-497-8979
Email: tuttle@ucar.edu

Project Summary:

Click on picture to view the entire figure.

Figure 1. June-August 1999-2006 RUC Shortwave/Radar Event Statistics

Carbone, R.E. and J.D. Tuttle, 2008: Rainfall occurrence in the United States warm season: The diurnal cycle. To be submitted Journal of Climate the week of Oct 1, 2007.

We have nearly completed a study examining the diurnal cycle of warm season precipitation over the United States, which may result either from local cumulus convection or remote forcing and the delayed-phase arrival of propagating rainfall systems. The main objective of the study is to extend and to clarify the findings of Carbone et al. (2002) which were based on a short period of record (four years).  The period of record is now 12 years, thus permitting us to reduce uncertainty and to place the results on firmer ground for climate science applications.   With the 12-year record we can begin to search for evidence of diurnal anomalies that might occur during phases of El Nino and the Southern Oscillation.

Slides 1 and 2 of the companion power point presentation summarize the diurnal character of convection over the U.S. Near the time of maximum solar heating (2100 UTC) convection is prevalent over the higher terrain of the Rocky Mountains, the Gulf Coast/Florida Peninsula and to a lesser degree over the Appalachians (Slide 1). In the Central Plains convection is at a minimum despite the peak in solar heating. During the evening hours convection originating over the Rockies propagates into the Central Plains resulting in a nocturnal maximum of precipitation in that region. Elsewhere convection is in the decaying stage except offshore in the Gulf and Southeast coastal areas where new development occurs because of land breeze circulations.

Click on picture to view the entire figure.

Figure 2. Diurnal Cycle of Radar Rainfall

The sharpness of the diurnal distribution of precipitation can be seen in Slide 2 showing the standard deviation (hours) of the diurnal maximum. As expected the standard deviation is lowest in the Rocky Mountain region and along the Gulf Coast where the precipitation distribution is strongly peaked.  Even though the distributions are very similar in these two regions, the forcing mechanisms responsible are very different. In the Rockies forcing is due to the elevated heat source of the higher terrain while along the Gulf Coast sea breeze circulations play an important role. Note that the sea breeze effects extend up to 250-300 km inland. Along the northern tier of states the standard deviations are generally large (broad distributions of precipitation) due in large part to forcing from synoptic scale systems that are not tied to the diurnal cycle. Over a large part of the Midwest and extending southwestward into Oklahoma and Texas the standard deviations are also large. This is the result of both local forcing near the time of maximum solar heating and long-lived propagating convection that originated over the Rockies.

To assess the significance of propagating systems, cumulative rainfall from these was calculated as a fraction of total rainfall vs. longitude (Slide 3).  An ambiguity arises at the local diurnal maximum, where one may attribute all, some, or none of the rainfall to a coincident propagating system, owing to the superposition with locally forced convection.  We account for this ambiguity by exhibiting the extremes, 0% and 100% attribution to the propagating component.

A characteristic fraction of total precipitation that results from propagating systems of 6 h duration or longer is ~60% between the Rockies and the Appalachians (Slide 3a).  This fraction decreases for propagating systems longer than 12 h and 24 h duration to ~50% and ~30%, respectively (Slide 3b, 3c).  While the precise fraction of total precipitation is uncertain, it is evident that propagating rainfall systems weigh heavily in the diurnal cycle over the central United States.

There are multiple factors underlying the unusual diurnal cycle in the central U.S., among these being the mountain-plains solenoidal circulation (MPS) and the Great Plains low level jet (GPLLJ).  The MPS is characterized by afternoon ascent of order 5-8 cm s-1 near the continental divide and descent of order 1-2 cm s-1 over a broad area from ~95W-100W (Slide 4).    At night the vertical motion reverses sign retaining a slightly diminished magnitude at each location.  The MPS is favored where there exists westerly shear of the zonal wind, which is the prevalent condition.

We have examined the seasonal characteristics of the diurnal cycle features in search of signals possibly linked with remote oceanic forcings.  The NOAA Multivariate ENSO Index  is based on six observed variables over the tropical Pacific (surface observations of P, u, v, SST, T, and total cloudiness fraction of the sky).  Slide 5 shows in plan view the correlation between the ENSO index and radar precipitation anomalies for a one-month lag. For each warm season (JJA) the percent precipitation anomaly from the twelve-year mean is calculated and correlated with a three-month average of the ENSO index. To account for possible time lags between variation in the ENOS index and the precipitation response in the northern latitudes, time lags are introduced into the analysis. At a zero-month lag the Jun-Aug precipitation is correlated with the Jun-Aug ENSO average, at a one-month lag the Jun-Aug precipitation is correlated with the May-Jul ENSO average, and so on. It was found that the correlations peaked at a one-month lag (r = 0.94) with the correlations decreasing to 0.90 and 0.85 at zero- and two-month lags, respectively. To assess the significance of the correlations, a simple t-score was
computed and the black contours in slide 5 enclose those correlation values which are considered significant at the 95% confidence level.

Trying to correlate 12 years of radar data with the slowly varying ENSO signal is marginal at best. For this reason most of the correlations shown in Slide 5 should not be considered significant. The one exception is in the northwest quarter of the U.S. (Idaho, Montana and Wyoming) where there is a relatively large area of significant positive correlations. This is generally in a region where synoptic scale forcings are more important. Based on this short 12-year record there is no clear evidence that any of the regions with strong diurnal forcings are affected by ENSO during the warm season.

Publications:

arbone, R. E., J. D. Tuttle, 2007: Rainfall Occurrence in the United States Warm Season: the diurnal cycle. J. Climate. (Submitted)

Hsu, H., M. Moncrieff, P. S. Sullivan, M. Dixon, J. D. Tuttle, 2007: Spatial statistical properties of multi-scale convective precipitation over North America. J. Climate. (Submitted)

Xiao, Q., X. Zhang, C. Davis, J. Tuttle, G. Holland, P. Fitzpatrick, 2007: Hurricane initialization using airborne doppler radar data for WRF. 8th WRF Users' Wkshp., Boulder, CO, US, UCAR.

Liu, C.-H., M. W. Moncrieff, J. D. Tuttle, 2007: Propagating rainfall episodes over the Bay of Bengal. Geophys. Res. Lett.. (Submitted)

Xiao, Q., C. Davis, X. Zhang, J. Tuttle, G. Holland, Y.-H. Kuo, 2007: Initialization of Hurricane Jeanne (2004) using airborne doppler radar data. 11th Symp. IOAS-AOLS, San Antonio, TX, US, American Meteorological Society.