Prepared for presentation at the 2nd International Conference on Climate and Water, Espoo, Finland, 17 - 20 August 1998. Conference URL: http://ahti.hut.fi/wr/caw2
Regional Climate Scenarios for Water Resources Research
and Impact Assessment
E. S. Takle, Z. Pan , R. W. Arritt , and W. J. Gutowski, Jr.
Iowa State University
Ames, IA 50011
Regional climate modeling offers a physically consistent method of evaluating the impacts of global climate change on scales of relevance to water resource issues. We present simulations of extreme precipitation periods for the central US and conclude that regional climate models are capable of reproducing the mass, wind fields, and moisture movement critical for generating precipitation. The region being studied has a complex combination of dynamical and thermodynamical processes that lead to extreme precipitation events. The model successfully captures these processes, although some discrepancies in the timing and magnitude lead to precipitation fields that are close but not exact matches of observed patterns.
Understanding present and future characteristics and variability of the hydrological cycle requires that we develop physically consistent simulations of present and future climate on scales critical for management and long-term planning for water resources, agriculture, and natural ecosystems. Of particular interest is the ability to simulate extreme hydrological events such as floods and droughts. We are coordinating simulations of regional hydrological climates by use of limited-area climate models that are designed to (1) evaluate strengths and weaknesses of regional climate models, including improvement over results of global climate models, when compared with observations, and (2) develop high-resolution regional climates consistent with global model simulations of the present climate and future climates perturbed by increased greenhouse gases and sulfate aerosols.
The focus of our studies is on the central US where summertime precipitation is governed by a complex combination of mesoscale processes. Figure 1 (Takle, 1995 as adapted from Wallace and Hobbs, 1997) shows the diurnal pattern of precipitation in this region. Directions of arrows on this map indicate time of day (24-h clock) that summertime precipitation reaches its diurnal maximum. Most regions of the US have maximum between noon (northward pointing vector) and 6 PM LST (eastward pointing vector). The US Midwest, by contrast, has a distinct nocturnal (midnight, as indicated by southward pointing vector) maximum. Length of the arrow gives the amplitude of the maximum frequency divided by the mean hourly frequency. Details of precipitation regimes in this region are given by Takle (1995), but it well known that the nocturnal low-level jet (LLJ) plays a significant role in moisture transport leading to this precipitation anomaly. This complex combination of moisture availability and mesoscale dynamical processes make this a challenging region for simulation of the climatology of precipitation.
2. Simulations of Periods of Drought and Flood
Two 60-day simulations were performed for the US Midwest, one for the 1988 drought (15 May - 15 July 1988) and the other for the 1993 flood (1 June - 31 July 1993) with boundary conditions supplied from the reanalyzed data of the US National Center for Atmospheric Research and National Centers for Environmental Prediction and implemented over a boundary-condition fram beyond the continental US. Results shown herein were produced by the RegCM model (Giorgi et al 1993a, b).
Analysis of the precipitable water fields (not shown) produced by the model for the summer of 1988 revealed that the model simulated very well the large region of water vapor deficit over the western third of the US. The observed west-to-east gradient of precipitable water over the Great Plains, extending from the Canadian border well into Mexico, also is present in the simulated fields, although the model gradient is not as large. For the eastern half of the US, the observed fields contained a relatively smooth reduction in precipitable water from a high over the Gulf of Mexico to a minimum north of the Canadian border. The model tended to produce precipitable water values that were lower, more spatially irregular, and having lower north-to-south gradient over this region.
Figure 2a gives a cross-section of the simulated northward (meridional) component of the wind at 06 UTC at 35oN for the 60-day period in summer of 1988. The model simulates the coastal jets and a northward 8 m/s LLJ in the central US as compared with the observed 7 m/s wind for this period as shown in Figure 2b. For contrast, the observed (and simulated) 1993 LLJ shown later reaches a magnitude of 13 m/s. It is clear from these figures that the model is capturing the LLJ and its suppressed magnitude due to large-scale circulation during the 1988 drought period.
Total precipitation produced by the model for the 60-day period in 1988 is shown in Figure 3a and observed values for the same period are shown in Figure 3b. Although the simulated values are slightly too large over the Great Basin, the model produced generally quite good representation of precipitation over the western half of the US as well as the maximum over Florida. Largest discrepancies between model and observations were in southern Louisiana (simulated values too low), along the eastern seaboard (simulated value high), and in the Iowa-Wisconsin region (simulated value high).
The precipitable water fields (not shown) for 1993, both modeled and observed, showed the same (as 1988) large region of water vapor deficit over the Great Basin in the western third of the US, the west-to-east gradient over the Great Plains, and the north-to-south gradient over the eastern half of the US. The only deficiency of the modeled precipitable water for 1993 was a high value over Mississippi and low value in the western Gulf of Mexico. It is noteworthy that the only major difference between the precipitable water fields between the drought year and the flood year was a northward penetration of the Gulf dome of precipitable water into the central US as far north as central Missouri. The flood region was to the north and west of this dome in the observed data and more directly north of the dome in the modeled precipitation. The northward displacement of the region of highest precipitation from the region of highest precipitable water confirms the importance of mesoscale dynamical processes in governing the spatial details of precipitation in this region during flood conditions.
The peak value of the modeled precipitation (see Figure 4a) was about 550 mm whereas the observed maximum (see Figure 4b) was somewhat in excess of 450 mm. The modeled maximum was displaced to the northeast of the observed location by about 300 km, which is still within about 1 GCM grid box. The model produced very good precipitation patterns in the western US but failed to capture the precipitation maximum along the Gulf coast and the minimum in the southeast US.
The enhanced LLJ simulated by the model (Figure 5a) and observed (Figure 5b) is consistent with dynamical processes known to accompany intense precipitation events in the region. The displacement by the model of the precipitation region to the northeast of the observed location suggests that the model yet lacks some details of the coupling of the dynamics and thermodynamics with the moisture fields. This is further confirmed by the lack of a strong diurnal cycle as required by Figure 1. As an aside, we also have simulated this period with a different model that, for this case, has much better placement of the precipitation maximum. This provides encouragement that models eventually will be able to simulate correctly the position, timing and interrelationships of complex processes leading to precipitation in this region.
Impact of Surface Wetness on Precipitation for Extreme Events
Pan et al (1995) artificially changed the surface wetness for these 1988 and 1993 extreme rainfall periods to examine the role of moisture recycling on precipitation. They found that an artificially moistened surface during a drought period would have little impact on subsequent rainfall, but that artificially dried surfaces during flood periods will reduce rainfall. Hence the relative contribution to water recycling by local evapotranspiration is much less important during drought as compared to flood conditions.
Specific Events during the Flood Period
Detailed analysis of individual events during the flood of 1993 by Pan et al (1996) showed that spatial details (e.g., orientation of the precipitation footprint, shape of the precipitation region, occurrence of multiple precipitation maxima) were well simulated in some but not all cases and depend on details of the model parameterization of precipitation. They further found that specific rainfall maxima within a contiguous precipitation footprint may have opposite sensitivities to underlying evapotranspiration. For the case studied, the total rainfall at one rainfall center diminished when the surface was artificially moistened whereas a nearby local maximum was enhanced. Detailed analysis revealed that rainfall will be enhanced (reduced) by surface wetness if the convective boundary layer lacks sufficient moisture (heating) to increase rainfall.
Effects of surface processes on rainfall in the US involve a complex interplay between thermodynamics and dynamics both locally and remotely. The LLJ was observed to strengthen (weaken) when surface soil becomes dry (wet). The intensified LLJ over a dry surface provides abundant moisture and instability for precipitation systems, thereby negating the impact of decreased evapotranspiration. On the other hand, in the western US, where the atmosphere is relatively dry and there are no prominent dynamic factors such as the LLJ, rainfall tends to be more dependent on local thermodynamic conditions.
Dependence on Land use
Pan et al (1998) investigated the impact of land use on precipitation by artificially changing surface vegetation conditions with all other conditions being unchanged for periods of flood, drought, and normal rainfall. They found that under normal-year conditions, current land use has increased rainfall by about 8% over the central US (due to agricultural crops which transpire more than native plants) and decreased rainfall slightly over the western US. In extreme years, current land use decreases rainfall over the central US and increases precipitation over the western US. The decreased rainfall in the central US is associated with a weakening of the LLJ due to reduced daytime sensible heat flux over the southern Great Plains. This also demonstrates the critical role of mesoscale dynamics in controlling precipitation in this region.
We have demonstrated that a regional climate model simulates well the mass and wind fields, including the LLJ that is responsible for movement of moisture within the domain, when supplied with boundary conditions far from the region of interest. Results also indicate that the model produces aggregate precipitation values well below normal, as observed, during the drought year although spatial distributions did not match exactly to observations. For the flood year, the model produced well above normal precipitation in the upper Midwest corresponding to periods of observed flooding, although precise timing, location and amounts did not match observations. Ability of the model to simulate dynamical processes known to be critical to precipitation in this region offer encouragement for future model simulations of precipitation details critical to water resources research.
Acknowledgments. This research was supported in part by the Electric Power Research Institute agreement WO4446-03, NASA Grant NAG 5-2491, and a grant from the Center for Global and Regional Environmental Research at the University of Iowa.
Giorgi, F., M. R. Marinucci, G. T. Bates, and G. De Canio, 1993a: Development of a second-generation regional climate model (RegCM2). Part I: Boundary-layer and radiative transfer. Mon. Wea. Rev., 121, 2794-2813.
Giorgi, F., M. R. Marinucci, G. T. Bates, and G. De Canio, 1993b: Development of a second-generation regional climate model (RegCM2). Part II: Convective processes and assimilation of boundary conditions Mon. Wea. Rev., 121, 2814-2832.
Pan, Z., M. Segal, R. Turner, and E. Takle, 1995: Model simulation of impacts of transient surface wetness on summer rainfall in the United States Midwest during drought and flood years. Mon. Wea. Rev. 123, 1575-1581.
Pan, Z., E. Takle, M. Segal, and R. Turner, 1996: Influences of model parameterization schemes on the response of rainfall to soil moisture in the central United States. Mon. Wea. Rev. 124, 1786-1802.
Pan, Z., E. Takle, M. Segal, and R. Arritt, 1998: Simulation of potential impacts of man-made land use changes on US summer climate under various synoptic regimes. (Submitted to Journal of Geophysical Research)
Takle, E. S., 1995: Variability of Midwest summertime precipitation. In G. R. Carmichael, G. E. Folk, and J. L. Schnoor, eds, Preparing for Global Change: A Midwestern Perspective.. Academic Publishing bv, Amsterdam, The Netherlands. pp 43-59.
Wallace, J. M., and P. V. Hobbs, 1977: Atmospheric Science: An Introductory Survey. Academic Press. New York. 467 pp.
Figure 1. Diurnal variability of hourly summertime precipitation for the United States (adapted from Takle, 1995).
Figure 2. Simulated (a) and observed (b) meridional wind (m/s) at 06 UTC at 35oN for 15 May - 15 July 1988.
Figure 3. Simulated (a) and observed (b) accumulated precipitation (mm) for 15 May - 15 July 1988.
Figure 4. Simulated (a) and observed (b) accumulated precipitation (mm) for 1 June - 31 July 1993.
Figure 5. Simulated (a) and observed (b) meridional wind (m/s) at 06 UTC at 35oN for 1 June - 31 July 1993.