Computer considerations limited the highest resolution to 3 km. Using triple-nested interactive domains with 27, 9, and 3 km resolution, all default parameters, such as land use and topography, were left in tact. The results of the computer- smoothed topography, left the N-S oriented Dudh Kosi-Tughla Khola-Imja Khola valley system 60 km from south to north, 45 km from east to west at the north end and 30 km from east to west at the south end. A 7000 + meters ridge surround the valley on all sides, except at its southern entrance. Elevation rise from south to north is about 4000m. Further, the valley side walls on southwest and entire east section are extremely steep.
The 60 hours of half-hourly MM5 simulations included two full diurnal cycles of valley winds. A 60km south-to-north cross-section along the valley axis from 3000 to 7000 meters indicated the following features, many consistent with our surface meteorological observations and measurements: · Vigorous daytime valley flow from the south occurred along the entire axis, up to just above 5000m, with duration several hours longer at lower elevations. · Elevation affected nighttime directions considerably. While lower elevations had easterly and northeasterly drainage flows, the north and northwesterly direction at the higher (above 4500m) elevations indicated reaction to and interaction with the upper-air winds. · On the 10km long, snow and ice covered section along the valley axis above 6000m, a 2 to 5 m/sec southerly daytime flow occurred for about 4 hours during the afternoons. Flow here was westerly at all other times. Nighttime drainage flows appeared to be completely blocked by these westerly winds. · For locations below 4500m, the daytime layer of southerly flow reached a depth of close to 2000m at midday, with speeds of up to 10 m/sec. This depth decreases to several hundred meters for locations higher up the valley axis. At 6000m, the daytime up-valley flow was 200m deep. · At midday on May 10, the boundary layer extended to approximately 420mb (7000m ASL) along the first 40km of the entire south-to-north cross-section, with a depth of 4000m above the 3000m valley surface, gradually decreasing to several hundred meters over the higher elevations, above 5500m. Within the boundary layer, winds above the daytime up-valley flow were mostly calm, 0 to 5 m/sec, with variable directions, likely indicating both the sinking, return circulation above the valley and the turbulent daytime mixing resulting from changing wind speeds and directions. · The prevailing upper air and subtropical jetstream winds were westerly at all times with speeds of 60 to 80 knots. Along the cross-section, these winds displayed the expected diurnal coupling and decoupling pattern. At midday high-speed 60 knot winds were located just 1500m above slower winds within the boundary layer.
In spite of the 3km resolution, there was remarkable agreement between our measured wind data and the simulated surface wind data. The model, however, underestimated measured surface temperatures and surface dew points. For the two-day simulation, the results indicate that extremely cold air at low and mid levels, a super-adiabatic lapse rate, high surface dew points and convergence at mid levels, contributed to thunderstorm initiation. The model produced snowstorms at 1615 NST on May 10 and at 1445 NST on May 11, on and just below Everest, missing actual observations by 2 ½ hours! However, the model was unable to reproduce the observed thunderstorms.