OLYMPEX Data Workshop: GPM View W. Petersen, NASA Marshall Space Flight Center GPM OLYMPEX Primary Objectives: Datasets to enable……… Direct validation at multiple scales, for liquid and frozen precip types, Do we capture orographic enhancements/transitions, critical range of precipitation intensity (e.g., light to heavy) and types, spatial variability? How well can we estimate accumulated precipitation over terrain (liquid + frozen) and at what scales? Physical validation in mid-latitude cold season frontal systems over ocean and complex terrain, Column properties of frozen, melting, liquid hydrometeors- including their relative contributions to surface precipitation, transition as a f(terrain character), and systematic variability as a function of synoptic regime Integrated hydrologic validation in complex terrain Combined observation-modeling approaches to developing area-time integrated "reference" products over complex topography What are capabilities and limitations for use of satellite estimates based on observed uncertainties? An end to end tracing of GPM products, algorithm physics and product errors to water resources applications at different scales

Specific Requests From GPM Algorithm Developers Coordinate remote sensing and in situ microphysical profiles along ground-based radar & disdrometer radials over both ocean and land (ice and liquid PSDs, type, qi,l-contents)- at least one stratiform case desired Provide estimates of seasonal/basin-integrated snow water equivalent Collect surface reference emissivity and s0 data over land/ocean in rain-free conditions Collect T/P/RH sounding profiles Coordinate aircraft/ground instruments with GPM overpasses …................. An end to end tracing of GPM products, algorithm physics and product errors to water resources applications at different scales

……….Translated to OLYMPEX Observations Airborne Well-stacked airborne microphysics (ice-liquid) and “simulator” (radar, radiometer) observations over land/ocean, coincident GPM core (DPR swath) Included patterns for over ocean radiometer/radar instrument calibration, collected rain/no-rain reference backscatter cross-sections and emission (land/sea) Ground Based Operational network radar for continuous/regular background sampling Research-grade scanning multi-frequency/polarimetric and vertically-pointing radars (rain mapping, hydrometeor profiling, "build the column") Reference gauge/disdrometer/sounding network(s) supporting ground-radar retrievals, environment and process characterization, hydrologic studies. Regional falling snow estimates/measurements in terrain Basin hydrologic measurement networks (stream flow, soil state etc.) An end to end tracing of GPM products, algorithm physics and product errors to water resources applications at different scales Houze et al. 2017 BAMS

Mission to OLYMPEX: The Bottom Line (Precipitation Rate) 12/3 Case DPR V4 2015-2016 Performance in broader stratiform (light/moderate) Ok….. GMI ITE111 2015-2016 OLYMPEX EOP Nov-May 15/16 Left side is just a composite for duration of OLYMPEX EOP. DPR top, GMI bottom both against MRMS. DPR patterns are conditioned on 0.2 mm/hr/ Magnitude of estimates I similar. GMI same thing and even a hint of rain shadow- but largely low over middle of peninsula (mountains and coastline). MRMS almost certainly biased low over mid-Peninsula. We are for sure getting the estimates right part of the time- but are we getting them right for the right reasons, and where are we missing? So one job here is to take this "direct" validation down to case types - for example this is the champion overpass of 12/3 Baroclinic system forcing southerly flow and orographic enhancement on the south side of the Olympics - most do ok relative to MRMS in this broad stratiform type event and there are many more examples………………

Mission to OLYMPEX: Key Observables to Key Retrievals Z Dm R Nw Where coincident volumes of ground radar and DPR are compared- results are encouraging. Relative to the radar- which retrieves more in the process of anchoring the database of the radiometer retrievals as well………………in the mean DPR Z. Dm and RR all matched up pretty well with NPOL and KLGX volume matched quantities. But this can be misleading- it tells you that you are doing things ok where you have coincident matching volumes- good to know (basic retrieval mechanics produce something consistent with ground estimates)…….we'll come back to this.

Challenges GMI: Regime, Land Surface Post frontal GMI hits and misses on these. GMI V4 2014-2016 ? But what about misses- or potential challenges- we need to root out the situations where problems might exist and resolve. E.g., Post Frontal situations might be one place to look right away because of what we know about NUBF impacts. There are some indications that Post-frontal smaller/isolated systems can be hard to detect properly and a few instances of coastal issues came up as well- (not necessarily a surprise- but orographics impact where/how precip occurs out to the coastline- then there may be some unique things about when we miss). Interesting to note the that NCFR was basically captured (DPR got this pretty well). But again- how do case specific events as a function of regime, terrain type etc. contribute to the gross behavior we see when we do longer term accumulations or averages? This is where Direct Validation can blend with Physical Validation- and that is something we should do for this campaign dataset. Coastal, Pre-Frontal/Frontal NCFR 89 GHz had little or no ice scattering offshore; DPR NS consistent with GV- GMI misses intensity and coastal/water boundary

Challenges DPR: Clutter and Orographic Processes How well can we detect the orographic enhancement- its impacts on the precipitation process and surface estimate? NPOL RHIs up River Valley, across DPR Coincident D3R 1512 UTC (OP-10 min) 1532 UTC (OP+10 min) Z VR Zdr ρhv 1528 UTC (OP+6 min) Ku Z Ka Z Ku Zdr Ka Zdr 3 Dec 2015 GPM OP 1523-25 UTC HS Z MS Z 0°C ? NPOL/D3R Of course the GMI is not the only instrument that has specific challenges……The DPR is an excellent instrument for doing this type of precipitation- but the lowest clutter free bin at lower levels is an issue when considering its location relative to potentially important locations of "process"- This is our 12/3 overpass again and you can see in DPR xsects on the left, along a radial to NPOL from the nadir track……we are missing out on the lowest 1-2 km and this is a key process region for low level orographic enhancement (which is obvious)- and also the region where the melting level often interacts with the terrain (another basic question for precipitation processes and retrievals over this region). In NPOL data the enhancements are obvious up terrain; the melting level dips into/down toward the terrain - there are indications of enhanced streaks and layers of dendritic growth aloft (not as well shown in this xsect- but others show it at different azimuth angles; AC microphysical data indicated light riming and trace amounts of liquid water -10 to -15 or so- consistent ob in many events. Orographic enhancement at low levels and in ice Melting layer interaction with the terrain surface Both impact precipitation character at the surface Ice-process- DPR will capture the enhanced dendritic growth zone- but, what do/can the algorithms do with this information- how important is the process? Extrapolating to the surface……..? R. Chase, U. Illinois Stephanie Wingo, NASA MSFC/NPP

Column Properties and Variability Ocean to Summit Alexis Hunzinger, UAH Randy Chase, U. Illinois -15oC Column properties are important here- again their character and variability as a f(regimes); What process variability do we see and expecially repeatable or systematic behavior. E.g., ocean to land ice properties and contribution to total rain/snowfall. Ubiquitous layers of preferential growth of ice types and densities; ice water path contribution to liquid water path. Can we capture systematic microphysical parameter and process variability from ocean to summit? What are systematic contribution(s) (are there?) of IWP to LWP; ocean to summit, frontal-system regime/environment?

Character and Impact of Regime DSDs GV Dmass-ZDR relationship Oro-strat stratiform conv 12/12 frontal impacts on DSD. 12/08 AR, orographic enh., shallow conv., plus synoptic lift Once you make rain……………..how do the characteristics of the rain- in particular parameters like the DSD that provide the actual signal we measure with the DPR, vary. Have to stress that the OLYMPEX dataset is interesting and unique from most if not all of the other DSD datasets we have collected. This I because of the copius numbers of small drops in the disdrometer database. For example on the left- we create a "global" Dmass-ZDR relationship from our disdrometer datasets that we can use with continental scale radar GV. if we used the OLYMPEX DSDs as the basis for our more global retrievals, we would end up sifnificantly underestimating the Dm of other populations. By including it, we at least ensure that we represent that portion of the DSD population. On the right- we saw systematic impacts on the DSD character based on the height of the freezing level in frontal surface transitions, with transtiions between deeper stratiform and shallow orographically forced rain in several situations. Task is to analyze the data in such a way as to draw out the largest and most systematic modes of variability and then relate those to DPR observables or some other observable that the algorithm can use to be "aware" of a situation. Orographic enhancements should happen in many different locations in the world to first order……so OLY dataset applies in that regard. Copius numbers of small drops and attendant ZDR in the OLYMPEX precip regime are unique and emphasize impact of "regime" when considering/parameterizing global behavior. Rain rate Dm N OLYMPEx relationship underestimates Dmass at ZDR > 1.5 dB significantly if its relationship were used for ALL locations Complexity of terrain gradient/type, ML height (regime) create systematic changes in DSD (e.g,. Dm behavior)

Cloud Resolving Models to Bridge Observations and Scale Observed (NPOL) Simulated (WRF) From a modeling perspective- we collected datasets that should enable us to verify in a bulk sense the performance of cloud resolving models, and then turn around and use them for estimates of hard to observe, but algorithm-used quantities like cloud water, rime content etc. Moreover, if we convince ourselves that the models replicate the spatial variability of the precipitation at high spatial resolution- we can use them to improve our precipitation products by enabling a downscaling of satelllite datasets. Observed NPOL Simulated WRF Simulated WRF Z (fill) Precip rate(dash) qcw (fill), rime content (contour) Z Courtesy, Aaron Naeger, UAH

Seasonal Precipitation: Benchmarking Multi-Satellite Estimates First, create a "best" estimate! OBS + VIC Model And sort of working into a more integrated applications/approach- we can use models and observations as a means to derive a "best" estimate of the seasonal total snow water equivalent- partitioned into rain and snow, high and low latitudes. Which is essentially what Quian and Dennis are doing. Cao et al., 2016 AGU ? OLYMPEX IMERG Hydrologic Application? OLYMPEX gauge, radar, snow observations integrated with VIC snow model to create "best" estimate over orography

Multi-Platform Data Fusion Will be Necessary! (e.g. SIMBA, VN…) SIMBA Framework System for Integrating Multi-platform data to Build the Atmospheric column Define Column: grid center location, horizontal and vertical extent, spacing Platform-specific Modules: read in native data formats, process gridding and/or interpolation as needed to set coincident observations into single, 3D column grid SIMBA Column Data File: Write all available observations to a common 3D grid in NetCDF format. Attributes maintain key properties of original data (exact locations, operation modes, algorithm versions, etc.) Finally- a persistent analysis theme involves the use of multi-platform data to really isolate, in a consistent fashion, the physics and its variability in the column- still pushing the "build the column" idea. So, hoping we will have some fruitful discussions along those lines here (I know of at least two different efforts along these lines that are ongoing).- this SIMBA approach is but one. Existing SIMBA modules support several OLYMPEX sensors: NPOL 2DVDs D3R Pluvios DOW Gauges NEXRAD/88D GMI/GPROF APUs DPR/2ADPR Developed initially for NASA WFF Precipitation Research Facility Versatile SIMBA system can be applied to multi-sensor observations in any geographic domain Continuing work to incorporate: Soundings Aircraft-based platforms SIMBA’s data-fusion column product can accelerate precipitation process & satellite algorithm assessment studies Contact: Stephanie.M.Wingo@nasa.gov -- NPP/USRA at NASA MSFC

Summary Direct validation: Isolate bulk product uncertainties and their association with orographic enhancements, synoptic environment (and uncertainties in its spec) , precipitation intensity and type, area/time-accumulated precipitation over terrain (liquid + frozen)…..this should blend into physical validation where possible. Physical validation: Using multi-platform approaches, characterize- (consider context of what we can and cannot directly observe with GPM)- bulk column properties/processes of liquid, melting, frozen precipitation and any systematic interactions with terrain gradients and frontal regimes (observations and models). Build the column! Integrated hydrologic validation in complex terrain Satellite estimates combined with modeling and ground obs over complex topography to both verify and improve products (accumulation, location, downscaling…..) [vision toward Level IV products] Identify most robust uses over orographic watersheds An end to end tracing of GPM products, algorithm physics and product errors to water resources applications at different scales