Orographic Precipitation Studies
of the
Cascade Mountains of Washington

Modeling Studies, 2010

For the next step it was decided to see if a mesoscale model could reproduce the cold air pooling and the observed west slope/east slope precipitation ratios with their seasonal variation. The first modeling attempt was using MM-5. The choice of MM-5 over WRF was purely based on expediancy as my WRF version was suffering from segmentation faults following a OS update.

The main grid was 75x75 grid points with 18 km spacing. Within the main grid were two nested grids of 76x76 grid points with 6 km and 2 km spacing as shown in Figures 18 and 20.

MM5 Domains
Figure 18. MM5 Domains used for initial simulations for October 2005 and December 2001.

The terrain resolution for each grid can be seen in Figures 19, 21 and 22. In domain #1 with 18 km spacing only the large scale features of the Cascades are resolved. The individual mounain valleys do not show at all. With 6 km spacing the main valleys are shown, but in some cases are broken into several enclosed basins. With 2 km spacing all the major valleys are clearly resolved.

Terrain for Domain 1
Figure 19. Terrain contours in meters for MM5 Domain 1 with 18 km grid spacing.

Although the main ridge-valley features are resolved, considerable smoothing occurs and full depth of the valleys and height of the ridges is not accounted for. Since most of the climatological stations that I want to use for comparison with the model are located in valley bottoms, their elevations are overestimated by the model terrains. For 22 climatological stations within Domain #3, the mean elevation was 362 m above sea level. Within the Domain #2 grid these elevations are overestimated on the average by 372 m and within the Domain #3 grid by 304 m. For five of the stations in the Domain #2 grid the elevation was over estimated by more than 600 m. In the worst cases the elevation of three stations was over estimated by more than 750 m in Domain #2 grid and two of them exceeded that in the Domain #3 grid. Three stations in the upper Skagit Valley, Newhalem, Diablo Dam and Ross Dam (see Figure 23), all have elevation errors greater than 700 m both for the 6 km Domain #2 grid as well as the 2 km Domain #3 grid. Holden Village in the upper Lake Chelan drainage has more than an 800 m error in the 6 km grid, but this drops to a 566 m error in the 2 km grid. Similarly Stehekin 3 NW at the head of Lake Chelan has a 656 m error in the 6 km grid and a 417 m error in the 2 km grid (see Figure 24 for the location). All these stations are located in the bottom of deep narrow mountain valleys where considerable smoothing of the topography results from smoothing at both 6 km as well as the 2 km spacing.


MM5 Domains 2 & 3
Figure 20. MM5 Domains 2 and 3 used for initial simulations for October 2005 and December 2001.


Terrain for Domain 2
Figure 21. Terrain contours in meters for MM5 Domain 2 with 6 km grid spacing.

Terrain for Domain 3
Figure 22. Terrain contours in meters for MM5 Domain 3 with 2 km grid spacing.

Upper Skagit Valley
Figure 23. Location of Newhalem, Diablo Dam and Ross Dam in the upper Skagit River Valley.

Head of Lake Chelan
Figure 24. Location of Holden Village and Stehekin 3 NW in the Lake Chelan Valley.

Next Page

© 2011 by Don Burrows, All Rights Reserved