1. Federal Department of Home Affairs FDHA Federal Office of Meteorology and Climatology MeteoSwiss Evaluating modifications of the soil module TERRA COSMO General Meeting, September 2007 Felix Ament, MeteoSwiss

2. 2 COSMO GM, Athens Felix.Ament@meteoswiss.ch Dry soil moisture bias OPRerational COSMO, two-layer version Testsuite, multi-layer version Strong dry out bias! T2m Soil moisture Negative effect on T2m forecast.

3. 3 COSMO GM, Athens Felix.Ament@meteoswiss.ch Handling the dry out problem Long term Model improvement: Model formulations Parameters Short range Soil moisture nudging: Insertion ECWMF soil moisture at layers below 9cm  Negative effect on T2m forecasts vanishes ECMWF.

4. 4 COSMO GM, Athens Felix.Ament@meteoswiss.ch Design of TERRA standalone experiments time Meteorological Forcing: T, p, u, q, Q down Precipitation RR Simulation of Energy balance Soil processes Annual cycle of vegetation SVAT „TERRA“ COSMO analysis Atmospheric Forcing: COSMO analysis data Domain: see left; 64x61 gridpoints at 7km resolution Period: year 2006 plus December 2005 for spin up Initialization: Operational COSMO analysis Working in the dark – nearly no or insufficient observations!

5. 5 COSMO GM, Athens Felix.Ament@meteoswiss.ch Nudged mulitlayer versus two layer Analysis of the water budget Features of “Nudged Multilayer”: Despite Nudging, LE is reduced in July/August and T max is higher. Most of the nudged water (=residuum) is converted into runoff. Remarkable: Less precipitation. Nudged multilayer Operational 2-layer Snow SM Evaporation Rain Surface Runfoff Intermediate Runfoff Ground Runfoff

6. 6 COSMO GM, Athens Felix.Ament@meteoswiss.ch CTL standalone versus OPR 2-layer (mm) RRRunoff_sRunoff_gEvapo.  SM  SNOW ResiduumRunoff_m E (JA) OPR 2-layer 1389334291699-11-37-113?111 W/m 2 Nudged ML 11485036787276-32734?95 W/m 2 CTL standalone 13974403596214-27019693 W/m 2 Features of “CTL standalone”: Again, reduced LE in July / August (no response in T_2m due to external forcing) Dry out in summer, but recovers until the end of the year. Higher runoff. Do we really have a dry-out problem? Probably, the T_2m diagnosis is misleading? Doubts

7. 7 COSMO GM, Athens Felix.Ament@meteoswiss.ch Sensitivity experiments RIGIDRigid lit boundary condition = Impermeable rock below lowest layer GWATERGround water boundary condition = Water below lowest layer VEGPARASpatially varying stomatal resistance limits and plant albedo ROOTDISTImplementation of a non-homogenous root density distribution ECOVEGAll vegetation parameters adopted from ECOCLIMAP NP89Alternative bare soil evaporation Noilhan and Platon (1989) Z0LOCUsage of local roughness length without gravity wave drag component BROOKS1 Drainage and capilary rise acording to Brooks and Corey formulation 6 type DWD classification; lookup table adopted from R. Grasselt (UBonn) BROOKS2 6 type DWD classification; lookup table from J. Helmert (DWD) adopted from Shao and Irannejad (1999) BROOKS3 11 type USDA soil classification; Shao et al lookup table PEDO Brooks and Corey parameters derived form Rawls and Brakensiek pedotransfer function MACROPOR Enhanced drainage by Macropores Formulation adopted from VEG3D (Braun, 2002 ) Lower boundary Drainage & diffusion Vegetation Exchange

8. 8 COSMO GM, Athens Felix.Ament@meteoswiss.ch Lower Boundary Condition I - concepts Free drainage rigid lid ground water wet dry medium RIGID GWATER

9. 9 COSMO GM, Athens Felix.Ament@meteoswiss.ch Lower Boundary Condition II Ground water condition Problem: Definition of soil moisture gradient at top of water GWATER Solution: Solve Darcy equation with these simplifications: F is constant below centre of lowest layer D is constant there, too K varies only linearly with  :

10. 10 COSMO GM, Athens Felix.Ament@meteoswiss.ch Drainage and capillary rise I BROOKS1 BROOKS2 CTL: Rijtema (1969), e.g. for drainage K: Brooks and Corey (1964) – much more popular However, Brooks and Corey formulation requires three parameters to derive drainage and capillary rise (depending on soil moisture) – they are not well defined.  BROOKS1: 6 type DWD soil classification; lookup table adopted from R. Grasselt (UBonn)  BROOKS2: 6 type DWD soil classification; lookup table from J. Helmert (DWD) adopted from Shao and Irannejad (1999)

11. 11 COSMO GM, Athens Felix.Ament@meteoswiss.ch Drainage and capillary rise II PEDO fields of soil pro- perties (e.g. pore volume) USDA classification ECOSOIL 6 classes Lookup table by DWD BROOKS3 11 classes Lookup by Shao not fully done! DWD classification Rawls and Brakensiek, 1989 Ecoclimap

12. 12 COSMO GM, Athens Felix.Ament@meteoswiss.ch Drainage and capillary rise III Runoff_g MACROPOR Marcopores help to infiltrate water rapidly during rainfall might avoid runoff generation of saturated top layer Parameterization (adopted from VEG3d, e.g. Braun 2002) mit F max =10 und  min =0.5.

13. 13 COSMO GM, Athens Felix.Ament@meteoswiss.ch Vegetation I Minimal / maximal stomatal resistance as well as plant albedo have constant value in TERRA CTL VEGPARA uses spatially varying values depending on land-use VEGPARA CTL

14. 14 COSMO GM, Athens Felix.Ament@meteoswiss.ch Vegetation II ECOVEG External vegetation parameters prescribed by ECOCLIMAP dataset (Mason et al., 2002): Exhibits more variabilty Systematic higher root depth More detailed seasonal cycle (not shown) (all maps are valid for July)

15. 15 COSMO GM, Athens Felix.Ament@meteoswiss.ch Vegetation III CTL Uniform root depth ROOTDIST Linear root depth distribution ROOTDIST Recipe Diagnose soil moisture stress function f SM,loc for each layer separately Determine mean SM stress by average weighted by layer thickness  z and root density  root Extract transpired water proportional to f SM,loc  z  root

16. 16 COSMO GM, Athens Felix.Ament@meteoswiss.ch Atmospheric exchange I ZOLOC Local roughness length z 0,local CTL roughness depends not only on local conditions, but also on variance of orography to account for gravity wave drag.  Very high roughness length over mountainous areas.

17. 17 COSMO GM, Athens Felix.Ament@meteoswiss.ch Atmospheric exchange II Dickinson, 1984 : BATS scheme Designed for a two layer soil module! Noilhan and Platon, 1989 (NP89) : ISBA scheme, Meso-NH Top Layer SM at Lindenberg NP89

18. 18 COSMO GM, Athens Felix.Ament@meteoswiss.ch Result I - bare soil evaporation NP89 Snow SM Evaporation Rain Surface Runfoff Intermediate Runfoff Ground Runfoff Significant reduction of Evaporation during spring and fall, … … but no effect during summer!

19. 19 COSMO GM, Athens Felix.Ament@meteoswiss.ch Result II – Budget Summary Runoff_s mm Runoff_g mm Runoff_m mm Evapo. mm  SM mm E (JA) W/m 2 CTL440359196621493 Snow SM Evaporation Rain Surface Runfoff Intermediate Runfoff Ground Runfoff RIGID-97722755 GWATER0-2937221 BROOKS1-2120-116-67-2 BROOKS2-2061-128-30-11-4 BROOKS3------ ECOSOIL29-22-2-1410-2 MACROPOR-122-18461 PEDO-954-65-34-10-4 VEGPARA13213-4412-6 ROOTDIST0-212-2 ECOVEG29-1145101-11817 NP8916622-8418-2 Z0LOC2592-374-3 Deviations in mm Reduced Runoff RIGID, GWATER Reduced interm. Runoff BROOKS1, BROOKS2 Reduced Evapo. (sustainable) NP89, VEGPARA, Z0LOC Little impact ROOTDIST, MACROPOR Problematic ECOVEG (dry out!)

20. 20 COSMO GM, Athens Felix.Ament@meteoswiss.ch Conclusions COSMO TERRA-ML is very robust; modifications have in general surprisingly small impact TERRA-ML standalone has proven to be useful tool to asses the midterm effect of model modification. However, objective decisions about implementation of modification is difficult, due to lack of observational data. Scientifically the following modification can reasonably be recommended: NP89 (removes high evaporation in spring & fall) VEGPARA (better representation of forest) (GWATER (counteracting dry-out)) (BROOKSX (being state-of-the-art)) Outlook: Cross studies (e.g. BROOKS and GWATER) Long term integration to reach model balance. Combination with improved T_2m diagnosis.