Estimation of Areal Rainfall Using the Radar Echo Area Time Integral

View More View Less
  • a NOAA/ERL/National Severe Storms Laboratory Boulder, Colorado
  • b NASA Goddard Space Flight Center, Greenbelt, Maryland
  • c NOAA/ERL/Office of Weather Research and Modification, Boulder, Colorado
© Get Permissions
Full access

Abstract

This work extends the Area Time Integral (ATI) method of Doneaud et al., developed for the lifetime rainfall from an individual storm, and the instantaneous areawide rainfall method of Atlas et al., to the measurement of the cumulative areawide rainfall for periods up to 12 h. The database is the radar and rainfall network data for the three summers of the Florida Area Cumulus Experiment (FACE) II. For 12-h accumulations, V, over the area of 3.6 × 104 km2, we find correlations of 0.92 between radar deduced rainfall and ATI where the latter is computed at intervals from 5 min up to 1 h. The slope of the regression line V/(ATI) is 3.4 mm h−1. Using a gage network with density of 1/11 km2 over an area 1.5 × 104 km2 the correlation coefficient drops to 0.84, still sufficiently high to confirm the validity of the ATI approach. Also, with the gages the V/(ATI) slope decreases to 2.6 mm h−1. The decrease in the correlation is due largely to anomalous propagation which falsely increases storm areas, and partly to the poorer sampling by the gages. The decrease in the rain volume from radar to gage-determined values is probably due to: 1) underestimation of the rain cores by the spaced gages; 2) the use of the wide beam WSR-57 and low threshold for echo area measurements, which detects weak anvil and other precipitation debris to increase the effective echo area without a proportional increase in surface rainfall; and 3) an inappropriate ZR relation. A comparison of the V/(ATI) ratios using either radar or gage rainfall to the value expected theoretically on the basis of the probability distribution of rain rate at Miami shows that one should expect about twice the volume per unit echo area as those observed. This too is believed to be due to the wide beam and the low threshold which tends to enlarge the echo areas excessively. Improved correlations and better agreement with theory are expected at higher radar/rain rate thresholds and with narrower beams.

Abstract

This work extends the Area Time Integral (ATI) method of Doneaud et al., developed for the lifetime rainfall from an individual storm, and the instantaneous areawide rainfall method of Atlas et al., to the measurement of the cumulative areawide rainfall for periods up to 12 h. The database is the radar and rainfall network data for the three summers of the Florida Area Cumulus Experiment (FACE) II. For 12-h accumulations, V, over the area of 3.6 × 104 km2, we find correlations of 0.92 between radar deduced rainfall and ATI where the latter is computed at intervals from 5 min up to 1 h. The slope of the regression line V/(ATI) is 3.4 mm h−1. Using a gage network with density of 1/11 km2 over an area 1.5 × 104 km2 the correlation coefficient drops to 0.84, still sufficiently high to confirm the validity of the ATI approach. Also, with the gages the V/(ATI) slope decreases to 2.6 mm h−1. The decrease in the correlation is due largely to anomalous propagation which falsely increases storm areas, and partly to the poorer sampling by the gages. The decrease in the rain volume from radar to gage-determined values is probably due to: 1) underestimation of the rain cores by the spaced gages; 2) the use of the wide beam WSR-57 and low threshold for echo area measurements, which detects weak anvil and other precipitation debris to increase the effective echo area without a proportional increase in surface rainfall; and 3) an inappropriate ZR relation. A comparison of the V/(ATI) ratios using either radar or gage rainfall to the value expected theoretically on the basis of the probability distribution of rain rate at Miami shows that one should expect about twice the volume per unit echo area as those observed. This too is believed to be due to the wide beam and the low threshold which tends to enlarge the echo areas excessively. Improved correlations and better agreement with theory are expected at higher radar/rain rate thresholds and with narrower beams.

Save