Transcript Thrusts of Noah LSM Development and Land Data

Thrusts of Noah LSM Development and
Land Data Assimilation at NCEP:
recent ops upgrades, testbeds, validation, recent
lessons, physics changes, calibration,
new vegetation, albedo & snow cover data sets
Ken Mitchell
EMC Land Team and Collaborators
WRF Land Modeling Working Group Meetings
13-15 September 2005
Outline
• Noah LSM upgrades in ops Eta/NAM meso model of May 05
– Hi-res USGS vegetation types & STATSGO soil types
– Recalibrated canopy resistance parameters
• Noah LSM uncoupled testbeds at NCEP
– N. American & Global Land Data Assimilation Systems)
– Porting the NASA Land Information System (LIS)
– Revisiting the optimum LAI value for the Noah LSM
• JCSDA: Joint Center for Satellite Data Assimilation
– Joint NCEP-NASA-NESDIS-AFWA-NAVY
– New vegetation, snow cover and albedo databases
• Anticipated future upgrades to the Noah LSM
Noah LSM upgrades in NCEP operational
Eta/NAM mesoscale model of May 03
• Implement Noah LSM Version 2.7.1
–
–
–
–
Upgrade from Version 2.3.2
Includes new patchy snow cover treatment for sfc fluxes & skin T
Substantially eliminate early snowpack depletion bias
Surface emissivity less than unity over snowpack (0.95)
• Reduce Noah LAI value from 5 (Version 2.7.1) to 2
– Use of high LAI=5 in Noah 2.7.1 emerged from glitch in NLDAS
• Apply 1-km USGS global landuse database
– replaces 1-deg NASA/ISLSCP BiB global landuse
• Apply 1-km STATSGO landuse database
– Replaces 1-deg NASA/ISLSCP 1-deg Zober
• Tune canopy resistance and soil parameter tables
USGS landuse on 12-km Egrid
STATSGO soil texture on 12-km Egrid
SiB landuse on 1-deg base grid
Zobler soil texture on 1-deg base grid
EROS Data Center also offers this global 1-km SiB landuse database
Suggested Discussion Point for Thursday session:
Easier tuning of canopy resistance parameters when fewer landuse classes.
Aside: Boston U. group has just delivered
MODIS-based global 1-km SiB landuse map (not shown)
Ops Eta Model
Soil Moisture
Precipitation Analysis
Jul 04
T
Jul 05
T
Jul 05
RH
MDW
Jul 04
RH
NESDIS has new 12-year
Global 0.144-deg monthly
Green vegetation fraction (GVF)
Climatology to replace former
5-year climatology (same grid).
OLD: JULY
NEW: JULY
The new GVF product is also
available as a realtime weekly
update (Mondays) – not shown
Both climo and weekly update
based on AVHRR NDVI
Difference of new-minus-old NESDIS July greenness fraction climatology
New field has higher greenness fraction over central US
NESDIS IMS snow maps over North America. Apr 29, 2005
NESDIS has upgraded its operational daily
Interactive Multi-sensor Snow (IMS) analysis
from 24-km resolution to 4-km resolution.
The new 4-km analysis will be implemented
In the NCEP WFRF-NMM model.
4 km resolution (new)
24 km resolution (old)
From U.Arizona / X. Zeng
New MODIS-based
5-km global database of
maximum albedo for
Deep snow conditions.
From Robinson and Kukla
Older 1-degree global database
of maximum Albedo for
deep snow conditions.
Difference of Zeng minus Robinson max snow albedo over N. America
(on 12-km ops NAM Eta model grid)
NLDAS:
The N. American
Land Data Assimilation System
Recent lessons learned about:
1) the optimum LAI value for the Noah LSM
1) impact on transpiration in Oklahoma region
owing to change from 1-deg 12-class SiB to
1-km 27-class USGS landuse
NLDAS Design
(The Uncoupled Approach)
Project Leader: Dag Lohmann
1. Force models with 4DDA surface meteorology (Eta/EDAS),
except use actual observed precipitation (gage-only daily
precip analysis disaggregated to hourly by radar product)
and hourly downward solar insolation (derived from
GOES satellites).
2. Use 4 different land surface models:
–
–
–
–
NOAH (NOAA/NWS/NCEP)
MOSAIC (NASA/GSFC)
VIC (Princeton U./ U. Washington)
Sacramento (NOAA/OHD)
3. Evaluate results with all available observations, including
soil moisture, soil temperature, surface fluxes, satellite skin
temperature, snow cover and runoff.
NLDAS Lesson 1
Wide disparity among land models
in warm season evaporation over
non-sparse green vegetation
Top row: Map of NLDAS predominant class of vegetation type (A) and soil type (B)
Bottom row: Climo Jul green veg (C) from Gutman et al and avg annual precip total in mm (D)
NLDAS Model Mean Annual Evaporation (mm) over Oct 97 – Sep 99
Seasonal change of total column soil moisture: 30Apr minus 30Sep, 99, at 23Z:
For the four NLDAS land models
*
*
*
*
Monthly Surface Water Budget
dS = P – R – E
Annual Surface Water Budget
P=E+R
July 1999
April 1999
Fig. 22 SGP ARM/CART Monthly Mean Diurnal Cycle of Surface Energy Fluxes
Mean Diurnal Cycle of Sensible Heat
Fluxes
From OU Mesonet/OASIS: summer 2000 (Basar/Nemunaitis
M e an Se ns ible He at Flux at the OASIS Supe r Site s
300
OASIS Observ ations
Mosaic LSM
Noah LSM
250
Sensible Heat Flux (W m-2)
200
150
100
50
0
-50
-100
0
400
800
1200
Time (UTC)
1600
2000
Why differences in warm season
evaporation over non-sparse vegetation?
•
•
•
•
•
Canopy resistance specification
Canopy interception of rainfall
Rooting depths
Green vegetation cover and LAI
Upward diffusion of soil water from below
root zone
• Interaction of infiltration and evaporation
over the annual cycle
Evapotranspiration Treatment
E  Edir  Et  Ec
WHERE:
E = total evapotranspiration from combined soil/vegetation
Edir = direct evaporation from soil
Et = transpiration through plant canopy
Ec = evaporation from canopy-intercepted rainfall
Canopy Resistance Issues
 Canopy transpiration determined by:
– Amount of photosynthetically active (green) vegetation.
Green vegetation fraction (f) partitions direct (bare soil)
evaporation from canopy transpiration:
Et/Edir ≈ f(f)
– Green vegetation in Eta based on 5 year NDVI climatology of
monthly values
– Not only the amount, but the TYPE of vegetation determines
canopy resistance (Rc):
R c min
Rc 
LAI F1F 2 F3 F 4
Noah LSM in NLDAS used UMD landuse map
transformed to corresponding SiB classes
(to allow use of Noah’s traditional SiB-based tables for the vegetation parameters)
Correct NLDAS transformation of UMD landuse class to SiB landuse class
12
Incorrect Noah LSM transformation of UMD landuse class to SiB landuse class
mistakenly used in the NLDAS 4-model intercomparison production phase
(Note assignment of semi-arid “broadleaf shrubs with bare soil” class
over the cropland region of midwest, northern and southern central plains)
12
Tuning the Noah LSM in the NLDAS testbed with above mistake led to
adoption of overly high LAI=5 value in Noah 2.7.1
With correct veg type mapping
LAI = 5
With incorrect veg type mapping
LAI = 5
NLDAS-Noah-2.7.1 Mean Jul 98 18Z Surface Latent Heat Flux
With USGS landuse and STATSGO soils and their respective parameter tables
LAI = 1
LAI = 3
LAI = 2
LAI = 4
Impacts in NLDAS/Noah-2.7.1 of
switching from
SiB-based veg parameter tables
to
USGS-base veg parameter tables
Using correct landuse map in
both cases.
USGS / STATSGO: LAI = 4
SiB / Zobler: LAI = 4
Difference of
top two panels.
Anticipated Upgrades in
Unified Noah LSM
• Temporally varying annual cycle of LAI
• Dynamic vegetation model
– Multi-layer vegetation canopy treatment
• Groundwater component
• More and thinner soil layers
• Multi-layer snowpack model