Radar Analysis for Design Storm Application

This study presents a Depth Area Reduction Factor (DARF) analysis for characterizing the spatial distribution of rainfall used for design purposes by the Nevada Department of Transportation (NDOT). Assuming uniform rainfall over watersheds can lead to unreasonably large peak flow rates, especially in regions where rainfall is highly variable or isolated in areal extent.

Despite the arid to semi-arid nature of much of Nevada, both flooding and flash flooding occurs throughout. Significant flash flood events are often associated with a westward-moving, inverted trough tracking across the southwestern United States. Under these conditions, torrential downpours and resultant flash flooding have been documented, particularly in Southern Nevada. During the North American Monsoon (NAM), short-duration, high-intensity rainfall is produced by atmospheric moisture surging northward from the Gulf of California. Atmospheric rivers are another mechanism that can produce significant floods in winter, particularly in Northwestern Nevada. While flooding can occur across Nevada at any time of year, it mainly occurs from July-September, coinciding with the peak of the NAM.

Investigation of significant, flood-producing events was performed to identify DARF relationships for a range of durations and for areas across Nevada. High-resolution coverage over large areas makes radar an efficient means for capturing the spatial distribution of storms and for estimation of DARFs. This efficiency is important for the study area where rain gauge coverage is sparse, thus making radar the primary source for capturing the depth-area relationship statewide. Since the rain gauge network is very sparsely distributed in Nevada, many of the storms considered do not have rain gauge accumulations. Thus, it is not possible to select storms of known frequency or return period depth. Periods identified as flood-producing, are analyzed with the aid of radar.

Analysis of radar-based storm events from the 2005-2014 period was performed to determine representative DARFs for Nevada. This period is defined by availability of polygon-based NWS flash flood warnings/flood advisories and reliable radar data. Storms that qualified for consideration were either within a warning/advisory polygon issued by the NWS and/or included in the NCDC database, and therefore are considered extreme or at least with flood-producing rainfall depths and intensities.

 

Rainfall data used for determining the depth-area relationship were taken from the National Weather Service (NWS) NEXRAD Level III reflectivity data. Data from four radar stations were used in this analysis spanning a period from 2005 - 2014: KRGX (Reno, NV), KLRX (Elko, NV), KESX (Las Vegas, NV) and KICX (Cedar City, UT). Radar data quality could be detrimental to the accurate quantification of depth-area relationships without quality control procedures. A determination of useable data was focused on areas with minimal ground returns. Steps were taken to remove anomalous propagation (AP), which can be unpredictable and often contaminates precipitation measurements, potentially causing erroneous rainfall estimates. While the second lowest radar tilt was used to avoid the most severe AP, this choice can make detection of rainfall difficult at far range. Each radar image was then visually inspected to remove events that exhibited evidence of AP, as well as, those that did not indicate significant rainfall. Further, events were examined for bright-banding, snow, or mixed precipitation, and were excluded if present. Only storm events having an estimated total storm depth greater than 0.5” were considered as a further screening criterion. A total of 547 radar-based individual events passed through quality control, and were used to determine depth-area relationships across the state.

 

Radar storm totals were generated and analyzed for peak 1-, 2-, 3-, 6-, 12-, and 24-hour durations. For each event, the DARF was computed by normalizing the rainfall by the peak depth, and expressed as a percentage or decimal fraction relative to the peak depth. Rainfall area is related to the duration of the storm with longer duration events covering a larger area, thus making duration an important factor for consideration of the DARF relationship. Of the 547 individual events that passed quality assurance steps, there were 1,720 duration-based storm totals developed and used to determine DARF relationships at the median and 90^th percentile. Because there were very few 24-hour duration storms, DARF relationships were identified covering the state, for 1-, 2-, 3-, 6-, and 12-hour durations.  Figure 1 shows the semi-log plot of median and 90^th percentile relationships fitted to the data cloud of depth-area for each event, with 12-hr duration and for areas up to 1295 sq. km. (500 sq. mi.) for southern Nevada.

 

Summarizing the average DARF values statewide, most of the reduction occurs once the area increases to 259 sq. km. (100 sq. mi.), and thereafter, declines steadily to 1,295 sq. km. (500 sq. mi.). As the duration of the storm progresses, less reduction is expected due to a larger areal distribution of rainfall. For areas less than 259 sq. km. (100 sq. mi.), the difference in DARF values across the state is 0.15 (or less) for 1-hr duration storms. At this short duration, Lake Tahoe and Northeastern Nevada had the greatest depth reduction with increasing area, whereas, along the western border with California, the least reduction is observed for areas up to 1,295 sq.km. (500 sq. mi.). Because of the small difference in DARF values (areas < 259 sq. km.) across Nevada, a single set of DARF relationships for 1-, 3-, 6-, 12-, and 24-hr durations were identified, as seen in Figure 2. Also shown are DARF relationships published by NOAA Atlas 2, Clark County Regional Flood Control District (CCRFCD), and the Truckee Meadows Regional Drainage Manual (TMRDM).  Note the plateau in the NOAA Atlas 2 and TMRDM studies, which were performed with geographically-fixed rain gauge networks. Similarity is evident between the DARF relationships found by this study and CCRFCD for smaller drainage areas.

 

Figure 1 Observed DARF data points and fitted model for median and 90^th percentile for 12-hr duration in southern Nevada (1 sq. mi. = 2.59 sq. km.)

Figure 2 Statewide DARF relationships for 1, 3, 6, 12, and 24-hr duration in comparison with the mean of other studies (1 sq. mi. = 2.59 sq. km.)