Current Operational Data Capabilities Issues and Perspectives Chandra

Current Operational Data Capabilities, Issues and Perspectives Chandra Kondragunta Hydrometeorology Group Hydrology Laboratory Office of Hydrologic Development NOAA/National Weather Service Q 2 Workshop Norman, OK June 28, 2005

Outline . . . QPE requirements for NWS operations Current operational data capabilities and issues Potential other data sets for QPE Field perspectives Summary

QPE Requirements Priority Category : “ 1” = Mission Critical : Cannot meet operational mission objectives without this data set Threshold Spatial Res. Temporal Res. Accuracy Data Latency Mapping Accuracy 1 km 6 min. 1 mm/hour 3 min. 0. 2 km Objective 0. 5 km 1 min. 0. 25 mm/hour 1 min. 0. 1 km

Current operational data capabilities and issues

Data Sources Current sources of data for QPE in NWS: 1. 2. 3. 4. Rain gauge data WSR-88 D radar rainfall estimates Satellite Precipitation Estimates NWP model output

Rain Gauge Data Rain gauge data for NWS operations come from several different sources: HADS: . Federal & State Wildland Fire Programs --- 2, 400 rain gages. USGS --- 1, 734 rain gages. USACE --- 1, 637 rain gages. NWS --- 222 rain gages. 117 other DCS Platform operators (USBR, TVA etc. ) Other: . State and local government funded agencies (Mesonets). Automated Surface Observing System. Cooperative rain gauge network. Other NWS supported gauges (IFLOWS, ALERT etc. ) Spatial resolution : Non-uniform Temporal resolution : hourly and daily (few 1 min. gauges)

HADS Hydrometeorological Automated Data System: . An integrator of in situ data. Acquires non-standard raw data relayed via GOES Data Collection System (DCS). More than 1. 7 million observational values processed each day. 11, 500 data reporting locations. 97% of data network is non-NOAA. A future component of NOAA’s Integrated Surface Observing System (ISOS) Program

Issues with rain gauge data There are several issues with rain gauge data: Coverage : Uneven spatial and temporal coverage, Sparse network density for some regions Quality : Gauge data quality is a big problem Examples: Transmission errors, staggered reporting times, frozen gauges, outliers, missing data etc Timeliness : Reports arriving late Errors: . Wind effects --- Under catch. Gauge exposure blockages (trees, buildings etc. ) --- Under catch. Solid precipitation --- under catch. Heavy rain rates --- under catch. Strong wind --- over catch

Density is uneven and poor in Nevada

Radar Rainfall Estimates Current radar rainfall estimates come from WSR 88 D radar network Spatial resolution : 2 km x 1 Deg. Temporal resolution : 6 min. Issues: Beam blockage, under estimation, over estimation, detection problem, Anomalous propagation etc.

Effective CNRFC Radar Coverage Effective NWRFC Radar Coverage

Satellite Precipitation Estimates Current Satellite Precipitation Estimates (SPE) come from GOES satellite. They are generated by an algorithm called the Hydro. Estimator. Spatial resolution : 4 km Temporal resolution : 15 min. Issues: Under estimation, over estimation, detection problem, mis-location of precipitation

NWP output Several NWP model outputs such as RUC, MM 5, MOS, NDFD etc. are used in operations Spatial resolution : 5 km Temporal resolution : 1 hr Issue: Accuracy of model output

MPE Multi-sensor Precipitation Estimator (MPE) is an operational software currently being used at several NWS field offices to generate QPE. It uses rain gauge, radar and satellite precipitation estimates to generate multi-sensor QPE The main features of MPE are: . Delineation of effective radar coverage. Mosaicking based on radar sampling geometry. Service area-wide precipitation analysis. Mean field bias correction of radar rainfall estimates. Local bias correction of radar and satellite precipitation estimates. Semi-automated rain gauge QC tools. Several GUI tools to interactively modify the point values or gridded fields

DPA MPE WSR-88 D ORPG/PPS Rain Gauges Mean Field/local Bias correction Multi-Sensor Precipitation Estimator (MPE) Hydro-Estimator Local Bias correction RFC

MEAN FIELD BIAS (MFB) ADJUSTMENT Before Correction After Correction

MULTISENSOR (GAUGE+RADAR) ESTIMATION FILLS MISSING AREAS Bias Corrected Multi-sensor

CNRFC 24 -Hour Precipitation, 17 Dec 2002 Hydroestimator (mm) Local Bias-Corrected Hydroestimator

Gauge QC in MPE Spatial Consistency Check (semi-automated): . Checks for consistency of a gauge value with the neighboring gauge values in all four quadrants. Lightning data is used to screen the gauges received rainfall from convective systems before flagging the outliers Multi-Sensor Check (semi-automated): . Compares the rain gauge values with radar estimates and points out the stuck gauges Display 7 X 7: . Ability to display 7 X 7 HRAP bins centered on a gauge to aid manual gauge QC

Spatial Consistency Check

Multi-sensor check

Locally Grown Capabilities Some of the locally grown software are. Mountain Mapper : To generate gridded QPE, gauge QC (mostly in the western region). XNAV, XDAT to QC gauge data

Potential other data sets for QPE. Reflectivity data from the Terminal Doppler Weather Radar. Canadian radar data (NMQ). Microwave satellite precipitation estimates from SSM/I sensors. Precipitation estimates from the TRMM. Lightning data

Field Perspectives

Rain Gauge Data. “There always issues with rain gauge data. Missing data, Zero reports, transmission errors, tipping bucket errors, poorly maintained equipment (particularly with IFLOWS) staggered reporting times etc. ” --- OHRFC (Several other RFCs expressed similar view point) . “High elevation data, such as SNOTEL has problems because of the freezing of the gauge” --- NWRFC . “WGRFC has numerous gauge – sparse areas over roughly the western half of our region. Gauges are densely clustered in our largest cities due to ALERT systems. There can be issues with data quality and timeliness within these systems” --- WGRFC

Radar Data. “Over and under estimation, significant gaps in coverage, lack of coverage of basins in Canada, gross underestimation in winter, inconsistent Z/R relationships in adjoining radars” --- NCRFC (Several RFC expressed similar view point). “ Radar data in our area is of no use in generating QPE. Beam blockage, inadequate coverage, melting level bright band etc. ” --- CNRFC. “Radar is useless in the NWRFC area” --- NWRFC. “We use MPE and have the usual radar issue: under/over estimation of rainfall, radar coverage, bright banding …” LMRFC

Satellite Precipitation Estimates. “We don’t use satellite precipitation estimates because of poor quality” --- LMRFC (Most of the RFCs expressed similar view point) . “ We use it EXTREMELY rarely, when there is no other data. Maybe 1 time in 10000 cases. It has proven to be of poor quality for the most part ” --- ABRFC . “WGRFC supplements the radar void regions with SPE. ” --WGRFC

In summary, Summary . Rain gauge data quality is an issue for current NWS hydrologic operations. Need to develop automated gauge QC techniques to satisfy the next generation QPE algorithm demands. Need to improve the rain gauge network density to improve the data coverage. Need to address the radar coverage gap issues by bringing in alternate data sets, such as satellite precipitation estimates and NWP model output. Need to address the radar rainfall estimation issues (over/under estimation)

Summary (Contd. ) . Need to improve the satellite precipitation quality by developing multi-platform, multi-sensor (IR+MW) techniques. Need to make better use of NWP model output in QPE estimation

Questions?

COOP network

Rain gauge in snow

Rain gauge site