MetStorm: Precision Precipitation Analytics

MetStorm Logo

MetStorm is the next generation of storm precipitation analysis software used to produce high-resolution precipitation maps/grids for a variety of purposes, including extreme precipitation studies (PMP and Precipitation Frequency), media inquiries, hydrologic modeling calibration/validation, forensic cases, insurance claims, emergency management and situational awareness. The analyzed storms go to building a national extreme storm database that will be leveraged for engineering design and operation applications. MetStorm integrates quality-controlled precipitation gauge data, dual-polarimetric (dual-pol) radar-estimated precipitation data, satellite-estimated precipitation data and innovative algorithms for computing precipitation analytics.

Locations of all previously completed MetStorm analyses (updated weekly). Please contact us for additional information on any of these storms.

MetStorm is a new Geographic Information System (GIS) based analysis system that produces gridded precipitation at 5-minute and/or 1-hour intervals over a specified domain (Laro, 2015; Parzybok, 2015). The relative spatial precipitation patterns are largely governed by Weather Decision Technologies’ Polarimetric Radar Identification System (POLARIS) quantitative precipitation estimates (QPE). The POLARIS QPE is a mosaic of dual-pol radar-estimated precipitation at a spatial resolution of 250 m.

Meanwhile, the precipitation magnitudes of MetStorm grids are influenced by quality-controlled rain gauge data from the NWS Cooperative Observer Network (COOP), Global Historical Climatology Network (GHCN), as well as from our strategic partner, Synoptic Data Corp. MetStorm has the ability to integrate hourly, daily, and irregularly measured precipitation data, thereby providing a high degree of gauge density for “ground truthing.” Satellite data, though at a coarser spatial resolution (4 km), influences areas void of rain gauge and/or radar data. Innovative algorithms blend the precipitation estimates from the different sources into a seamless GIS grid, which provides the basis for summary statistics, maps, tables, and plots. MetStorm was developed and is operated exclusively by MetStat, Inc.

MetStorm uses up to seven (7) key inputs to compute gridded precipitation across the analysis domain. While the type and amount of data may vary between analyses, MetStorm intelligently integrates all available data.

Flowchart of MetStorm input data, processing and quality control procedures, and output.

Precipitation gauge data

Measured precipitation from daily and hourly precipitation gauges are required input. MetStorm has the ability to utilize hourly, daily, and irregularly reported precipitation data, thereby providing a gauge density as comprehensive as possible for “ground truthing.” For historical storms, gauge data is easily added manually, however, MetStorm automatically accesses quality-controlled rain gauge data for newer (post-1997) storms from our strategic partner, Synoptic Data Corp, who aggregates, quality controls and archives hourly precipitation gauge data from over 200 networks across North America. Synoptic’s 1-hour precipitation data amounts to over 25,000 gauges, which are quality-checked using a multi-sensor quality assurance system designed and maintained by MetStat, Inc. MetStorm also accesses data from MetStat’s own in-house database, which is largely based on the Global Historical Climatological Network (GHCN). Spatially-based algorithms, which leverage nearby hourly gauges and radar data, convert the daily precipitation amounts into estimated hourly precipitation for use in the MetStorm precipitation analysis system.

Basemap

Basemaps are independent grids of spatially distributed weather or climate variables that are used to govern the spatial patterns of the hourly precipitation, particularly in areas where radar is either not available or of poor quality. The basemap provides a stable and spatially consistent reflection of how the precipitation may fall over a region. For MetStorm analyses over complex terrain, climatological basemaps, such as PRISM mean monthly precipitation or 2-year precipitation frequency grids, are often used given they resolve inherent orographic enhancement and micro-climates. Climatological basemaps in flat terrain, however, are not as effective given the weak precipitation gradients; therefore, in these cases, basemaps are often developed from pre-existing (hand-drawn) isohyetal patterns, independently-created radar-estimated storm totals, the summation of PRISM daily precipitation grids or the individual monthly (e.g., March 2013) PRISM precipitation grids available online.

Gridded dual-pol precipitation

MetStorm uses Quantitative Precipitation Estimates (QPE) from state-of-the-science dual-pol precipitation estimates from Weather Decision Technology’s (WDT) Polarimetric Radar Identification System (POLARIS) (Porter, et. al, 2012). The POLARIS QPE grids are a mosaic of dual-pol estimated precipitation from all U.S. and Canadian Next Generation (NEXRAD) radar sites. Depending on the precipitation type (e.g. wet snow, light rain, heavy rain, etc.) determined by the dual-pol radar data, an optimized radar-to-precipitation rate algorithm is utilized to compute precipitation at 5-minute intervals and at a spatial resolution of 250 m.

Radar reflectivity

Level-II radar reflectivity is the native data provided by NEXRAD weather radars across the United States. MetStorm translates this into a rainfall rate using a standard Z-R algorithm. The Z-R (radar reflectivity, Z, and rainfall, R) relationship allows estimation of precipitation from reflectivity. Most current radar-derived precipitation techniques rely on a constant relationship between radar reflectivity and precipitation rate for a given storm type (e.g. tropical, convective), vertical structure of reflectivity and/or reflectivity magnitudes.

Beam Blockage Mask

Particularly in complex terrain, radar coverage is often compromised, causing areas to have poor or no radar coverage. In order to overcome this, the MetStorm Analyst will manually define areas with poor radar coverage based on a summation of radar reflectivity grids during the storms. A summation of radar grids tends to amplify the impact of terrain blockages, thereby making it clear where radar coverage is poor. The radar blocked areas are used as a mask in MetStorm and infilled from neighboring, valid radar pixels. In most cases this resolves the problem, but in severe cases the blockages are so large that the infilling is not sufficient; in these situations, MetStorm instead relies on the isopercental interpolation estimates to produce a seamless transition from areas without radar to areas with adequate radar coverage.

Radar Confidence

In order to quantify the quality of the radar data (both POLARIS and reflectivity) across the analysis domain, MetStorm uses the lowest altitude (above mean ground level) of the radar beam. Generally speaking, a radar beam sampling precipitation closest to the ground is more reliable. Therefore, a function between a radar weight (ranging from 0 to 1) and radar beam height is imposed to create a radar-weight grid; this provides MetStorm with an objective means for determining where gauge-adjusted radar-estimated precipitation can be relied upon more than a purely basemap-driven interpolation of precipitation.

Satellite Precipitation

Satellite-based estimates of rainfall have been used since the late 1970s, especially in areas where rain gauge or radar data are unreliable or unavailable. Similarly, MetStorm uses satellite-estimated precipitation data, though coarse spatial resolution, to influence areas void of rain gauge and/or radar data. MetStorm ingests 1-hour satellite rainfall estimates known as “Hydro-Estimator” from NOAA’s Center for Satellite Applications and Research (STAR). The Hydro-Estimator uses infrared (IR) satellite data from NOAA’s Geostationary Operational Environmental Satellites (GOES) to estimate rainfall rates. The estimated rainfall rates are most accurate during the warm season in areas of deep convection (thunderstorms). The magnitudes of satellite rainfall are used for aiding the spatial interpolation of gauge data in MetStorm.

 

Quality Assurance

MetStorm is designed to operate with a wide range of input data which makes it a flexible tool for producing gridded precipitation associated with very old storms (pre-1900) as well as storms that just occurred. To instill consistency among previous storm analyses conducted by others, MetStorm’s logic is built on similar techniques used in the past by the U.S. Army Corps of Engineers (USACE) and others, but includes numerous improvements (see section titled: MetStorm versus Storm Precipitation Analysis System). MetStorm intelligently integrates all available data for creating quality GIS grids of Quantitative Precipitation Estimates (QPE). At a the minimum,  MetStorm requires one hourly station and a basemap to produce a series of QPE grids. However, if available, additional gauge data (of various types; see below), dual-pol radar-estimated precipitation data, traditional radar reflectivity and/or satellite data can be used to create seamless grids of QPE across varied terrain.

The nominal temporal resolution of MetStorm is 1-hour given precipitation is standardly reported in 1-hour precipitation increments. However, if radar data is available, further disaggregation down to 5-minute intervals (snapped to even 5-minute intervals, e.g. :05, :10, :15, etc.) is possible by imposing the temporal distribution of precipitation derived from radar precipitation estimates together with MetStorm’s final 1-hour QPE.

Quality control is an on-going and critical element of a storm precipitation analysis to ensure the highest data quality and integrity. Given the plethora of data available in each storm analysis, manual inspection is required to parse through data errors such as: incorrect observation times, previously undetected accumulation periods, beam blockage issues, radar anomalies, co-located gauges with different storm total amounts, undercatch of precipitation from gauges, etc.

An example of a MetStorm quality control interface used for performing QC on individual hourly grids.

MetStorm Products and Output

MetStorm’s gridded FQPE is the basis for a variety of precipitation analytics.  The complete list of potential deliverables include the following, which are described in more detail in the subsequent subsections:

  • High-resolution gridded QPE
  • Average Recurrence Interval (ARI) maps/grids
  • Depth-Area-Duration (DAD) plots/tables
  • Complete gauge data catalog
  • Validation plots
  • Error statistics
  • Mass curve tables/plots for any location at 5-minute or 1-hour intervals
  • Storm report, including the above elements plus a total storm map, a brief meteorological discussion, and the analyst’s assessment of the storm analysis reliability.

Below are examples of each MetStorm output product. Click on each image to view in fullscreen.

Example storm total map image from MetStorm.

An Average Recurrence Interval (ARI) map produced by MetStorm, highlighting the relative rarity of the precipitation event at a specified duration.

Spreadsheet of example station data. These files are output as .csv files, which can be read into programs such as Excel for easy viewing. It contains station metadata as well as storm-total precipitation.

Example of a 5-minute validation plot, wherein the gridded 5-minute precipitation as analyzed by MetStorm is compared to observed 5-minute gauge data.

Mass curve analysis, illustrating the incremental (shaded) and total (curve) precipitation accumulations at the analyzed point maximum rainfall location. These can also be created for user-specified locations.

Depth-area-duration curves, illustrating the maximum average depth of precipitation for a given duration and area size.

Example page from a Storm Precipitation Report. Reports are written by the storm analysts and describe the nature of the storm, data sources and quality, and confidence in the analysis results.

Scatterplot illustrating the relationship between observed gauge data and the MetStorm-analyzed grid cell corresponding to that location.

MetStorm versus Storm Precipitation Analysis System (SPAS)

Both SPAS and MetStorm are designed to achieve the same objective: produce high-resolution, gridded precipitation at 5-minute or 1-hour time steps for storm events. SPAS was designed, developed and used widely by MetStat from 2002-2015, but newer data sets and better algorithms motivated the development of MetStorm, which has a number significant enhancements (illustrated below) over SPAS.

 

FeatureSPASMetStorm
Integrates daily, hourly and irregularly reported precipitation gauge data
Integrates site-specific or mosaicked radar reflectivity data
Creates and applies on-the-fly radar-precipitation relationships
Utilizes “basemaps” for resolving precipitation without radar data and/or in complex terrain; dynamic basemaps are used when radar data is available
Creates hourly (5-min, if radar is available) grids of precipitation
Includes graphical user interface for thorough QA/QC
Addresses radar beam blockages
Includes an objective error/uncertainty analysis
Does not require a minimum storm analysis duration
Storm identification numbers based on storm date & location(uses indexing)
Provides control of Inverse Distance Weighing (IDW) function
Utilizes radar confidence grid for weighting
Includes automatic station co-location check and resolution logic
Automatically integrates quality-controlled precipitation gauge data
Utilizes quality control flags for hourly measurements
Uses quality-controlled, mosaicked Dual-Polarization precipitation estimates based on 13 hydrometeor classifications
Leverages satellite-estimate precipitation
Generates multi-duration Average Recurrence Interval (ARI) analyses
Automatic total storm precipitation map creation
Auto contribution into a national extreme storms PostGIS/PostGreSQL database
Spatial resolution as high as 250m2
Output/input clearly named with UTC date/time(uses indexing)
Disaggregation of daily to hourly data based on radar data
Automatically creates scatter plots and time series comparisons for station validation