Variability of raindrop size distributions in a squall line and implications for radar rainfall estimation

The intrastorm variability of raindrop size distributions as a source of uncertainty in single-parameter and dual-parameter radar rainfall estimates is studied using time series analyses of disdrometer observations. Two rain-rate (R) estimators are considered: the traditional single-parameter estimator using only the radar reflectivity factor (Z) and a dual-polarization estimator using a combination of radar reflectivity at horizontal polarization (Z_n) and differential reflectivity (Z_DR). A case study for a squall-line system passing over the Goodwin Creek experimental watershed in northern Mississippi is presented. Microphysically, the leading convective line is characterized by large raindrop concentrations (> 500 drops per cubic meter), large mean raindrop sizes ( > 1 mm), and wide raindrop size distributions (standard deviations >0.5 mm), as compared to the transition region and the trailing stratiform rain. The transition and stratiform phases have similar raindrop concentrations and mean raindrop sizes. Their main difference is that the distributions are wider in the latter. A scaling-law analysis reveals that the shapes of the sealed spectra are bent downward for small raindrop sizes in the leading convective line, slightly bent upward in the transition zone, and strongly bent upward in the trailing stratiform rain. The exponents of the resulting Z-R relationships are roughly the same for the leading convective line and the trailing stratiform rain ( ≈ 1.4) and slightly larger for the transition region ( ≈ 1.5), with prefactors increasing in this order: transition ( ≈ 200), convective ( ≈ 300), stratiform ( ≈ 450). In terms of rainfall estimation bias, the best-fit mean R(Z_H, Z_DR) relationship outperforms the best-fit mean R(Z) relationship, both for each storm phase separately and for the event as a whole.