3. Results

a. Precipitation

Comparison of the observed precipitation for the three simulated periods (Fig. 3) illustrates that the 1973 and 1994 periods were relatively wet, whereas the 1989 period was relatively dry, as discussed above. During the 1973 period, totals exceeded 450 mm over many areas of the western and southern peninsula. Totals for 1994 exceeded this threshold at several locations, with a few areas above 550 mm. Areas receiving more than 450 mm during the 1989 period are confined to the southwestern portion of the peninsula, with appreciably less over northern and central portions in comparison to 1994 and 1973. The July-August accumulated precipitation fields from the RAMS simulations (Figs. 4,5,6) illustrate that the model was able to capture the observed magnitudes with a reasonable degree of success. The overall magnitudes (with either land-cover scenario) are also relatively dry for the 1989 period, in accordance with the relative magnitudes in the observational fields. For all three periods, larger totals were realized with the use of the pre-1900 land cover. The difference fields (bottom panels of Figs. 4,5,6; 1993 minus pre-1900 scenario) show decreases over much of the domain when the 1993 land use was employed. South of the latitude of Tampa Bay, decreases are noted along parallel axes that run roughly north to south, between the coasts and the Kissimmee River basin. Interestingly, when the precipitation fields are averaged in space over all land grid points in the domain, all three of the resulting difference field values correspond to a 10%-12% decrease of their respective pre-1900 value.

		Fig. 3. Accumulated precipitation (mm) from the CPC analysis of cooperative observation network rain gauge data (analysis on 0.258 grid) for (top left) Jul-Aug 1973, (top right) Jul-Aug 1989, and (left) Jul-Aug 1994. [click on each of the images to view a larger version]

		Fig. 4. Accumulated convective rainfall (mm) from the model simulations of Jul-Aug 1973 with (top left) pre-1900 land cover, (top right) 1993 land use, and (left) the difference field for the two (1993 minus pre-1900 case). [click on each of the images to view a larger version]

		Fig. 5. Same as Fig. 4, except for Jul-Aug 1989. [click on each of the images to view a larger version]

		Fig. 6. Same as Fig. 4, except for Jul-Aug 1994. [click on each of the images to view a larger version]

Despite the consistent decrease in the grid-average value, precipitation increased along a north-south axis in the interior peninsula in all the three simulated periods when the 1993 land-use dataset was employed. Comparison with Fig. 1 shows that the area where these rainfall increases are found corresponds spatially to an axis of wetlands that was drained and converted to agriculture during the twentieth century. Note that the parallel axes of decrease discussed above that are juxtaposed to either side of this central axis of increase are consistent with the position of the afternoon sea-breeze fronts (Michaels et al. 1987). More evidence will be presented in later sections to suggest that these patterns are associated with marked changes in the climatological characteristics of mesoscale circulations that are associated with the land cover in this region.

In comparing these results and those presented in the study by Pielke et al. (1999) with the observational data, it is apparent that the new model configuration, with updated land-cover datasets and the Kain-Fritsch convective scheme, provided improved rainfall totals for the 1973 period. The model results shown in Pielke et al. (1999) illustrate that the old configuration yielded July-August 1973 rainfall totals that were under 300 mm for the 1993 land-cover case, with totals less than 320 mm for the pre-1900 scenario. The newer magnitudes are generally larger and closer to the observed. As noted above, the new model results also capture the general magnitudes for the 1989 and 1994 periods. However, there are some differences in the spatial details between the observations and the new model results. It should not be expected that the model would exactly resolve the finescale details of the accumulated field over the course of a 2-month integration. It is also worth noting that the observational dataset shown here may not perfectly represent the spatial distribution of the rainfall that actually fell. For example, the greater values in the observations over the open ocean could be an artifact of the nature of the Cressman analysis scheme (Cressman 1959) that is used by the Climate Prediction Center (CPC) to produce these objectively analyzed rainfall fields. The model results indicate that relatively little precipitation fell over the open ocean. The Cressman scheme extrapolates the rain gauge point values over land to the nearby ocean, because no gauge observations are available over the ocean to influence the analysis value. Despite discrepancies of this nature with the model results, and some disagreement between the model and observed spatial distributions, the simulations appear to provide a reasonable distribution of rainfall totals.

Go back to Experimental Design | Go ahead to Surface Fluxes