Transcript Title Slide
An Overview of the NoahDistributed Land Surface Model David J. Gochis, Wei Yu, Fei Chen, Kevin Manning WRF Land Surface Modeling Workshop Sep. 13, 2005 Brief Rationale for NoahDistributed • Standard land surface parameterizations characterize exchanges of radiation, heat, mass and momentum between the land and atmosphere • Historically, treatment of terrestrial hydrology has been simplified 1-d formulations • With high resolution implementations/applications there is now a need to explicitly account for enhanced hydrological processes: Violating early assumptions… 1. Surface runoff can not assumed to be “captured” by a stream channel 2. Lateral transfers from one cell may form significant input to adjacent cell Brief Rationale for NoahDistributed • Standard land surface parameterizations characterize exchanges of radiation, heat, mass and momentum between the land and atmosphere • Historically, treatment of terrestrial hydrology has been simplified 1-d formulations • With high resolution implementations/applications there is now a need to explicitly account for enhanced hydrological processes: – Higher resolution capabilities land surface heterogeneity – Earth systems-biogeochemcial cycling – Mitigation of “high-impact” weather events (e.g. floods) Outline • Brief overview of the Noah LSM • Noah-distributed core features • Implementation of Noah-distributed into the NCAR/HRLDAS framework • Ongoing and planned upgrades to Noahdistributed Community Noah Land Surface Model – Recent Enhancements • Recent Enhancement of the Community Noah LSM (released in WRF V2.0, May 2004) ‘Noah-Unified’ – “Nearly”-identical implementations of Noah LSM development effort: NCAR, NCEP, U.S. Air Force Weather Agency, NASA, university community – Fully modularized, F90 code conventions – Seasonal surface emissivity • surface emissivity is introduced as function of landuse • Added surface emissivity in surface energy balance equation for both snow and non-snow surfaces – Urban model improvements (the simple approach) such as • Large roughness length • Low surface albedo • Large thermal capacity and thermal conductivity Overland Flow Processes in Noah-Router (NCAR Tech Note: Gochis and Chen, 2003) IF (Surface Head > Retention Depth) Route Water as Overland Flow • New Parameters: retention depth, surface roughness • Ponded water in excess of retention depth subject to overland flow • Overland flow: fully-unsteady, explicit, finite-difference, 2dimensional diffusive wave (generally applicable to length scales < 1km) 2-Dimensional Diffusive Wave Overland Flow Routing Ogden, 1997 dhdx (hi 1, j hi , j ) / gsize sfx SOX i, j dhdx 1E 30 ( sfx / ABS ( sfx)) h5 3dt qx dx Dynamic modeling of land-surface hydrology with ‘Noah-Router’: Ponded Water Processes (NCAR Tech Note: Gochis and Chen) • New Parameters: None Direct Evaporation Surface Runoff Surface Head • Currently no formulation for partial area coverage Issue: May need to revise infiltration formulation • Ponded water consists of: residual ‘infiltration excess’ when using routed runoff to ofcalibrate: from previous time step and Surface runoff can not assumed to be “captured” by water a stream channel routed surface Re-infiltration • Direct evaporation of ponded water reduces potential evaporation (no adj. for temp/albedo) Ponded Water Evaporation and Re-infiltration • Ponded water not evaporated is subject to infiltration Subsurface Flow Routing Noah-Router (NCAR Tech Note: Gochis and Chen) Surface Exfiltration from Saturated Soil Columns Lateral Flow from Saturated Soil Layers Saturated Subsurface Routing Wigmosta et. al, 1994 • New Parameters: Lateral Ksat, n – exponential decay coefficient • Critical initialization value: water table depth • 8-layer soil model (2m – depth, sealed bottom boundary) • Quasi steady-state saturated flow model, 2-d (x-,y-configuration) • Exfiltration from fully-saturated soil columns SOX i, j dzdx 1E 30 ( gsize LKSAT SOLDEP ) tan n z hh 1 SOLDEP qsubx hh n Noah-Distributed Core Features • Present issues in treatment of subsurface routing: – Frozen soil adjustment to soil water • Remove soil ice from total soil moisture and route only liquid component – Update of conductivity as a function of soil temp/fraction of frozen soil – Inclusion of variable depth soils Subgrid Routing • Noah LSM is run at a variety of grid spacings • Subsurface and overland flow routing need to be performed on a terrain grid (< 1 km) • Required fields are aggregated/disaggregated using a simple averaging scheme Noah land surface model grid Routing Subgrids • Soil water, infiltration excess, routing parameters • • Can offer significant computational savings compared to full resolution implementations of Noah LSM Sacrifice detail in current formulation AGGFACTR = 4 Noah-Distributed Core Features • Subgrid disaggregation: proposed new method carrying over weighting factors between LSM model executions • Eliminates the “loss” of distributed information between routing timesteps Noah land surface model grid Routing Subgrids Noah-Distributed Software Features • F90, up to date with recent version in HRLDAS • Routing routines (1-d and 2-d) are contained within a single module (all agg./disagg. Routines will be included into routing module) • Routing and sub-grid options are switch-activated though a namelist file • Options to output sub-grid state and flux fields to WRF consistent netcdf files • Basic Flow: LSM > Disagg. > Subsfc > Overland > Agg. > LSM > … Outline • Brief overview of the Noah LSM • Noah-distributed core features • Implementation of Noah-distributed into the NCAR/HRLDAS framework • Ongoing and planned upgrades to Noahdistributed NCAR-HRLDAS (High Res. Land Data Assim. System) • Rationale and basics of HRLDAS: – Create globally-deployable variable resolution equilibrated land surface conditions for NWP initializations • Current static & forcing data: Time Variable Interval Dataset Grid Resolution Reference Precipitation Hourly NEXRAD Stage IV 4 km Fulton et al, 1998 Surface Meteorology Hourly EDAS 40 km Rogers et al., 1995 Land Use Static USGS-24 Category 1 km Loveland et al., 1995 Greenness Fraction Monthly n/a 0.15 degree Gutman and Ignatov, 1998 Soil Classification Static STASGO-16 Category 1 km Miller and White, 1998 Overland Flow Roughness Coefficient Static n/a Mapped to Land Use Adapted from Vieux, 2001 HRLDAS • Recent tests over International H2O Project (IHOP) domain • 18 month execution 1 Jan, 2001 – 30 Jun, 2002 Top Layer Soil Moisture (fraction): Total Column Soil Moisture (mm): Total Surface Evapotranspiration (mm): Ponded Water Evaporation (mm): Outline • Brief overview of the Noah LSM • Noah-distributed core features • Implementation of Noah-distributed into the NCAR/HRLDAS framework • Ongoing and planned upgrades to Noahdistributed Future Upgrades to Noah Distributed • Improve runtime performance: – 2-d vs 1-d formulations • DEM-based steepest descent method is much faster • Strictly DEM based routing (kinematic) problematic in flat areas where change in sfc water influences flow direction (e.g. backwater) – Working on a compromise algorithm • default to DEM based routing • check for backwater • Perform search – 1-d: 129 model days/wall clock d vs. 2-d: 97 model days/wall clock day (~1/3 faster) Future Upgrades to Noah Distributed • Complete parallelization • Couple to stream channel model (via DESWAT project) • Develop better method to nudge/assimilate groundwater and river/stream stage into modeling system • Develop enhanced method to characterize stream-aquifer exchange