WMOD00562 - Laserfiche WebLink

Home Browse Search

0.6 dBZ greater reflectivities than Albany for a given snowfall rate. This small difference is well within the range of calibration error, especially because no special calibration of the Albany radar was made during the 1995-96 winter. Consequently, the level II data (recorded at 0.5-dBZ resolution) for Albany were not adjusted in post-season analysis as were the Cleveland and Denver data sets. The fact that all three regions have the same optimized exponent of 2.0 is in agreement with a recent theoretical prediction and earlier radar experiments with snow, as has been discussed. Similar fJ values are anticipated to result from the additional regions for which data were collected during the 1996-97 winter as described in section 11. A single fJ value for all climatic snowfall zones in the nation would have obvious practical advantages. Smith and Joss (1997) point out that the fJvalue has limited impact on rainfall accumulations calculated from reflectivity measurements. It appears likely that the situation is similar for calculated snowfall accumulations. 5.3 SAA Coefficient Recommendations for Minneapolis Real-time SAA testing was scheduled for the Minneapolis area for the 1996-97 winter, prior to collection of hourly snowfall observations in that region. It was assumed that the average of the Cleveland and Denver Ze-S relationships would provide a good "first approximation. Minneapolis is about an equal distance from the other two areas. Accordingly, the recommended a and fJ values for Minneapolis real- time testing were 240 and 2.0, respectively. Some adjustment of these values is likely once analysis is complete of Minneapolis data collected last winter. 6. SNOW DEPTH PREDICTION 6.1 Introduction Radar echo power returned by snow particles is related to S by equation (1), and SWE accumulation can be calculated by integrating S over any desired time period (e.g., 1 h, storm period). But SD estimation is also important for many purposes such as decisions regarding snow plowing, traffic management, school closures, etc. Estimates of SD can be made from radar Ze observations if snow density is known because snow density is numerically equal to SWE/SD when both are measured in the same units. But real-time observations of snow density are scarce, and radar does not provide such measurements. The main purposes of this section are to review what is known about the density of freshly-fallen snow, and to recommend procedures for estimating it in an operational forecast setting, specifically the Cleveland and Minneapolis areas where real-time SAA testing was accomplished during the 1996-97 winter. The simplest approach would be to use average snow density values and to divide SWE estimated by equation (1) by average density in order to estimate SD. But it will be seen that snow density for a given hour will often differ from the average by a factor of 2, sometimes by a factor of 3, and occasionally even more. So a means of predicting snow density for the next 1 to few hours would be desirable for improved SD estimation by radar. The practicality of such short-term prediction will be considered. Ideally, one would be able to accurately predict the density by the beginning of any given hour with snowfall so that real-time calculations of SD could be made. However, we have been unable to discover any satisfactory prediction scheme. This section discusses snow density observations and our attempts to relate them to surface air temperature and to the past hour's snow density. Statistically significant but weak relations were found which provided too little predictive power to have practical application. 19