Radar Representation of Rainfall Part 2 of 2 2. Deriving rainfall (the Z-R conversion) Power returned to the radar is related to the rainfall intensity, but not in a consistent, easily-modeled way. Offers a good quantitative approximation Preserves important qualitative precip structure Varies with drop size distribution and precip phase
Historical Background Marshall, J. S. and W. McKay Palmer, 1948: The distribution of raindrops with size, J. Meteor., 5, 165-166. Spilhaus, A. F., 1948: Drop size, intensity, and radar echo, J. Meteor., 5, 161-164. Chapman, G., 1948: Size of raindrops and their striking force at the soil surface in a red pine plantation, Trans. Amer. Geophys. Union, 29, 664-670. Horton, R. E., 1948: Statistical distribution of drop sizes and the occurrence of dominate drop sizes in rain, Trans. Amer. Geophys. Union, 29, 624-637. Atlas, D., 1990: Radar in Meteorology, pp. 577-618. Austin, P., 1987: Relation between measured radar reflectivity and surface rainfall, Mon. Wea. Rev., 115, 1053-1070.
Z-R Relationships WSR-88D, Marshall-Palmer (general), and Tropical
Stage I PPS Module 4: Precipitation Adjustment Automated rain gauges are polled at a set time interval (once per hour) and used to determine if a bias exists in the radar-derived accumulation field. Gauge reports are matched with radar estimate for that area (9 nearest radar bins) The method (using the Kalman Filter) is designed to identify a single, representative radar bias If a bias is determined to exist, the entire radar-derived accumulation field is adjusted accordingly. Accurate and representative gauge reports are essential
Radar-Rain Gauge Comparisons Radar samples a volume of the atmosphere At discrete intervals Up to several thousands feet AGL Over a surface area which may exceed 1 mi2 Accumulations are estimated from reflectivities using an empirical Z-R relationship Rain gauges sample Continuously At the surface Over an area less than 1 ft2 Accumulations are measurements with the error factors associated with the gauge type
Rain Gauges Automated gauges - Two main problems: (1) Data disruptions cause missing periods of 5-20 min during height of storm. Underestimations by gauge (2) Noise in communication lines from lightning cause false reports. Overestimations by gauge Summer 1988: Matt Kelsch, Denice Walker, Erik Rasmussen, Ken Heideman Wedge gauges at BOU, LVE, PTL, LGM, ERI False reports verified at PTL, ERI 18 out of 22 significant rain days (3 sites >0.1”) had data disruptions during storms
Stage I PPS Limitations Gauges provide inadequate representation of the mesoscale structure of precipitation The vertical distribution of precipitation sampled by radar may be inadequate… ~l km above the ground @ 75 km distance may “look over the top” of stratiform precipitation significant evaporation may occur beneath radar beam problem is greater where terrain blocking exists
Stage 1 PPS Limitation (cont.) Hail enhances radar reflectivity resulting in the derivation of anomalously high rainfall rates. Accumulation may be overestimated by more than an order of magnitude Threshold to correct anomalous rainfall rates may cause underestimation of atypical heavy rain events Problem varies with site, season, and ambient conditions Enhanced radar return occurs in melting layer/bright band. Radar bias adjustments only work for systematic errors when the bias is uniform across the radar domain.
Adaptation Parameters: The adaptation parameter philosophy assists with general variations associated with site and season (coastal plain versus semi-arid prairie), but cannot easily account for atypical events within the climatology of a particular site or season.
Radar-Derived Precip: When changing Z-R coefficients is not the real solution: Range degradation, overshooting low-levels Problem associated with propagation of beam, not Z-R. Snowfall More complexity than liquid hydrometeors. Phase changes and mixed phases exist over small space/time scales. Range degradation often co-exists. Phase change: hail, melting snow Radical storm-scale changes in Z to R relationship. Minimal proof that hail correction can be done with Z-R. Inconsistent relationship between Z-R and hail occurrence.
Radar-Derived Precip: When changing Z-R may help: Consistently different average DSD (climate) Tropical versus mid-latitude (warm vs. cold process) Maritime versus continental Consistently different average DSD (season) Convective versus stratiform Precip System character Identify Convective versus Stratiform signature Identify warm versus cold rain signature Identify maritime versus continental
Why can’t the adaptation parameters and bias adjustment procedure solve all the limitations? Radar bias adjustment is only one uniform adjustment. It depends on adequate representation of precip by the local gauge network. Adaptation parameters can greatly help the algorithm performance for a given site and/or season. The parameters “tune” the algorithm for the typical scenario. Atypical events, such as unusually high rainfall rates, may not be diagnosed well. The most effective use of PPS is to make it a function of meteorology, not the “normal” climatology.
Can we account for the important atypical events without degrading the guidance for the more common typical events? Meteorological information from soundings, profilers, satellite, lightning networks, other 88D algorithms, and surface reports are a few examples of data sources that can assist with real-time adjustment of adaptation parameters. The most effective use of PPS is to make it a function of currently observed conditions, not the “normal” climatology.
Radar-derived Precipitation: A Summary Of Major Points Radar provides excellent storm-scale information about the spatial and temporal evolution of precipitation systems. Radar provides very valuable input as part of a comprehensive, multi-sensor precipitation system. Quantitative reliability issues are related to the fact that radar samples some volume at some elevation to estimate precipitation at the ground. Radar-derived precipitation is most reliably modeled for liquid hydrometeors; hail and snow add complexity. The above two points are not effectively corrected by changing Z-R coefficients; Z-R changes should be related to Drop Size Distribution knowledge. Radars and rain gauges do not measure equal samples Rain gauges do not provide a good representation of precipitation distribution, especially convective precip. The PPS algorithm has the versatility to evolve into a more comprehensive system, taking into account the ambient environment.