Improving Land-surface representation can enhance convection and - - PowerPoint PPT Presentation

Improving Land-surface representation can enhance convection and rainfall prediction

• ver the Indian Monsoon Region

Dev Niyogi climate@purdue.edu niyogi@gmail.com Landsurface.org

Also visiting Professor IIT Bombay (2015-2017), Visitor IITM Pune (2017) Additional collaborations with IIT BBS (UC Mohanty), IISc (PP Mujumdar), SAC (Chandra Kishtawal), JNU (AP Dimri) Work based on NSF CAREER, Monsoon Mission.

Paul Schmid and Dev Niyogi, 2017. "Modeling Urban Precipitation Modification by Spatially Heterogeneous Aerosols.“ in press, Journal of Applied Met. And Clim (JAMC-D-16-0320)

• Model results significantly improved for range
• f precipitation rates/ types with scale aware

Cumulus Parameterization and Aerosol Considerations.

• Results sensitive to (a)

microphysics option and (b) initial conditions as well as boundary layer feedback.

Land cover change due to

agriculture

intensification and

Urbanization is the new

global change underway.

Land feedback is causing

significant, and detectable,

changes in weather and climate through

temperature /rainfall

http://webpages.scu.edu/ftp/jready/family_urbanizationand modernization.html Sources: WHRC and Newslanc.com

Land effects are expected to be largest over India Monsoon region

POPULATION DENSITY MAP FROM FAO

Land Atmosphere Coupling globally dominant over Indian Region (GLACE study;

Koster et al. Science)

Why Does

Land Surface

matter?

Does land surface / heterogeneity really matter or is it a subgrid feature? (Hydrological studies suggest e.g. 10% of grid cases detectable change in response)

Land Surface: Bottom Boundary

• Where it all happens (land modeler’s perspective)

Courtesy Mike Ek (NOAA)

Slide courtesy: Dennis Baldocchi, UCB

Land Surface Model Development

• moving beyond energy and water flux source to the atmosphere
• focus on a more process-based approaches
• can now produce detailed surface information

– Vegetation temperature, Soil layer temperature and moisture – Snow depth and water, Vegetation, including crop growth – Upper soil – aquifer interactions

• important for forecasts of all time scales. Why?

– Memory – significant sources exist in soil (water and energy), snow and vegetation Our goal is to improve seasonal forecasts through

the inclusion of a better land model physics representation for the atmospheric model.

Figure from A. Pitman (2003)

Noah-MP: a community land model

• Computational power and process

understanding  finer grid spacings  more realism In land representation

• Corollary-

Finer the grid spacing, more would be the importance (and impact)

• f realistic land representation.

Model Uncertainty: Land Surface Model Structure

15

Noah LSM in NOAA Eta, NAM, GFS, CFS, MM5 and WRF Models and the LDAS/GLDAS

(Pan and Mahrt 1987, Chen and Dudhia 2001, Ek et al., 2003)

Noah-MP LSM in WRF and NOAA CFS (Yang et al., 2011; Niu et al., 2011, Barlage et al. 2014))

“Reality”

Tcan(x,y,z) Tbc(x,y,z) Tg(x,y)

Noah Noah-MP

Tcan Tbc Tg Tskin

Single surface temperature Multiple surface temperatures and distinct canopy

Tleaf Tleaf(x,y,z)

Detecting land

feedbacks from observations and model studies…..

Urbanization and land use change leads to regional temperature changes (warming= Urban Heat Island)

60 65 70 75 80 85 90 5 10 15 20 25 30

Average Temperatures in July for Urban & Rural Areas

http://earthobservatory.nasa.gov/Newsroom/N ewImages/images.php3?img_id=17489

ATLANTA UHI Kishtawal, Niyogi, Pielke, Shepherd, IJOC 2010

Observed urbanization and agriculture impacts

• ver Asia region

0.05 C/ decade ‘observed’ warming impact of urbanization

• ver China (Liming Zhou et al.

2004 PNAS) 0.34C cooling during growing season due to agricultural ‘green revolution’ in India (Roy et al. 2007 JGR)

Scales of landscape interactions– particularly agriculture and urbanization – are becoming significant in affecting climate via feedbacks and more importantly possibly detectable teleconnections

Land feedbacks have an even more profound impact

• n rainfall (and

associated precursors)

Pielke Sr., R.A., G. Marland, R.A. Betts, T.N. Chase, J.L. Eastman, J.O. Niles, D. Niyogi, and

• S. Running, 2002, The influence of land-use change and landscape dynamics on the

climate system: Relevance to climate change policy beyond the radiative effect of greenhouse gases. Phil. Trans. Royal Soc. (London) A. 360 , 1705-1719.

Physical Changes

• Deforestation
• Replace/transform

natural landscape

• Urbanization
• Irrigation
• Harvesting
• Intensification
• Energy Balance Changes
• Net Radiation and

Partitioning Changes

• Boundary Layer Moisture

changes

• Surface temperature

changes

• Roughness change
• Albedo change
• Precipitation
• Basinscale

Hydrological changes

• CO2 changes

(storage/emission)

• snow cover

Feedbacks

Effects/Impacts

Teleconnections

Image: D. Baldocchi

Pielke, R A., A Pitman, D Niyogi, R Mahmood, C McAlpine, et al. "Land use/land cover changes and climate: modeling analysis and observational evidence." Wiley Interdisciplinary Reviews: Climate Change 2, no. 6 (2011): 828-850.

Urban modification of Surface energy and Radiation Balance Change in Temperature and Moisture Greater trapping of Infrared radiation (Warmer air holds more moisture) Increased CAPE, Stronger thermals, Modified regional convergence Modified location, intensity, duration of Rainfall

Urbanization Impacts Scale Beyond the Surface (temperature)

30 yr rainfall climatology shows the Urban – Rural impact

Niyogi et al. 2011 JAMC

The enhanced convection and rainfall is simulated only when urban heterogeneity/ flux boundary exists

0100 UTC 0200 UTC 0300 UTC CONTROL NOURBAN

Diagnosis of the land heterogeneity rainfall anomaly (M. Lei and D. Niyogi, 2012– extended study by P. Schmid and D. Niyogi GRL 2013- “RAIL” method)

• Combination of

– Thermal Properties – (Albedo) – Surface Roughness – (z0) – Landscape size – (sprawl) – Flux gradients create mesoscale convergence / divergence due to land heterogeneities

Triple Interaction Term (F123)

Example - Cloud convection – land surface feedback

• Southwest Australia,

approximately 13 million hectares of native vegetation cleared for agriculture

• A 750km vermin proof

fence demarcates the boundary between cleared and pristine areas

• 20% reduction in precip
• ver agricultural areas
• Ray et al. 2003
• US Nair, UAH

Summary from multiple studies and reviews

• Land surface feedback and

heterogeneity has a significant impact on the timing, location,

intensity, and magnitude of

regional convection and rainfall

– Tremendous improvements in satellite capabilities and physical parameterizations, land models need to keep up the pace to benefit from them.

Radar reflectivity (dbZ) valid 00 UTC 25 May 2002 24-h forecast

LSM impact in coupled model precipitation forecast (SLAB versus Noah LSM and land data assimilation)

Observed 2-km Mosaic

Noah LSM

40 60 20 80

SLAB

Effect of Land Surface Representation on Convection and Precipitation simulation

(Holt T., D. Niyogi , F. Chen, M. A. LeMone, K. Manning, A. L. Qureshi*, 2006, Effect of Land - Atmosphere

Interactions on the IHOP 24-25 May 2002 Convection Case, Monthly Weather Review, 134, 113 – 133)

00 UTC 24 May – 12 UTC 25 May 2002 Nest 2 (4-km)

LSM representation impact not just significant for great plains but also for coastal regions

LULC impact important not just for calm conditions – but also important for active synoptic conditions (e.g. TS Alison 2001)

Ensemble land surface response on tropical storm rains / track

a) Black – NHC best track observations Red – Noah LSM (dynamic soil moisture/temperature) Yellow -Simple Slab land model (constant soil moisture) c)

Is the land surface feedback significant for Indian monsoon region where synoptic weather patterns and oceans, are known to be important?

Image Source: S. Gadgil and wikipedia

Perspectives on the impact of LCLUC on the Indian monsoon region Hydroclimate Dev Niyogi, Subashini Subramanian, U.C. Mohanty, K. Osuri C. M. Kishtawal, Subimal Ghosh, U. S. Nair, M. Ek, and M. Rajeevan Book chapter for NASA LCLUC Volume

• Some perspectives/ assumptions..
• Monsoon is not a giant sea breeze with rain band
• Myriad of coherent clusters lead to

seasonal and interseasonal rainfall variability and amounts (and eventually trends)

– E.g. thunderstorms and monsoon depressions contribute to about third to half of heavy rains – significant spatio temporal variability in actual rains that gets blended in grid analyses

• Challenge is to capture these organized clusters

and coherent feedbacks

Example of agriculture irrigation impacts on weakening of monsoon rain

• ver northern India (Niyogi et al. 2010, WRR)

Rodell et al. (2009) groundwater changes in India (2002-08), GRACE estimated rate of depletion in NW India is 33 cm/yr

Shift in the NDVI peak greenness with ag intensification by 30 days

• ver 2 decades

Reduced rainfall over NW India as a causal response of April vegetation cover and ag intensification leading to weaker monsoon heat low.

Human activities like agriculture and irrigation could be changing monsoon rains

Similar results for China where agricultural intensification contributes/cause drought (Liu Y., ….., Niyogi et al. Nat. Scientific Reports 2015)---- additional effect of management practice.

Indian monsoon rainfall is becoming more extreme (Goswami et al., Science, 2006)

Is the increased heavy rainfall climatology trend over the Indian Monsoon Region due to urbanization?

Kishtawal, Niyogi, Tewari, Pielke, Shepherd 2010, IJOC

Heavy rainfall trend noted for urban grids both in the surface dataset and satellite (TRMM) data

Kishtawal, Niyogi, Tewari, Pielke, Shepherd 2010, IJOC

• The results are also true for Northeast Monsoon

rainfall

Land feedbacks over the Indian monsoon region coherently organize themselves to modify regional/ multiscale temperature, rainfall and regional climate Implication:

Improved representation of land state can

lead to improved prediction of weather and climate over Indian Monsoon region

Kumar S., P. A. Dirmeyer, V. Merwade, T. DelSole, J. M. Adams, and D. Niyogi, 2013: Land Use/Cover Change Impacts in CMIP5 Climate Simulations –A New Methodology and 21st Century Challenges, Journal of Geophysical Research- Atmospheres, 118, 6337-6353.

Suggested pathway for developing and using improved land models for IMR

(i) Realistic Physics + (ii) Data to create model parameters / initialization  Tests in offline mode (LDAS) Tests in regional (WRF) model  Translate to GLDAS  Translate to global (CFS) experiments

Observations (and analysis)

Processes Model representation Model Datasets Model Sensitivity Research

• r Operational

improvements

Creation of gridded, multi decadal land data assimilation system (LDAS) SM/ST, surface energy flux products;

4km grid spacing (sub daily scale) by

downscaling different reanalysis products and enhancement by assimilating additional Indian

data available (veg cover, radar/satellite products)

Calibration / Validation of the LDAS fields using available insitu and satellite estimates (will be released as a Monsoon Mission contribution to the it )

LDAS Soil moisture at Khargpur

SoilM 5cm RMSE Correlation BIAS 0.027 0.90 0.014

H P NAYAK

Example of Validation of SM (m3/m3) with IMD (AgroMet) OBS at different locations

(Case-2, 18UTC 18 June 2011)

RED – LDAS/NoahMP_3DVAR BLUE - CNTL

Deep layer soil moisture (m3/m3) at initial time

CNTL (No LDAS input) LDAS Difference

S Moisture S Temp (Case-2, 18UTC 18 June 2011)

Using In models - Improved Thunderstorm simulation (27 May 2007)

K. K. Osuri, R. Nadimpalli, U. C. Mohanty, F. Chen, M. Rajeevan, and D. Niyogi: 2017: Improved prediction of severe thunderstorms over the Indian Monsoon region using high resolution soil moisture and temperature initialization, Nature Scientific Reports, 7: 41377.

Monsoon Depressions tracks and postlandfall rainfall improvement with enhanced land representation in WRF model

CASE-1 CASE-2

Other findings related to land state and monsoon depressions / heavy rain events

higher probability of MDs dissipating post landfall and cause intense local heavy rain if landscape is climatologically dry If anomalously wet antecedent land conditions exist prior to landfall, the MDs can have higher probability

• f deeper inland penetration and widespread rain

Kishtawal, C. M., D. Niyogi, B. Rajagopalan, M. Rajeevan, N. Jaiswal, N., U.C. Mohanty, 2013: Enhancement of

inland penetration of monsoon depressions in the Bay of Bengal due to prestorm ground wetness.

Water Resources Research, 49(6), 3589-3600) Bozeman L.M.G, D. Niyogi, S. Gopalakrishnan, F. D. Marks Jr., X. Zhang, and V. Tallapragada, 2011:An HWRF- based Ensemble Assessment of the Land Surface Feedback on the Post–Landfall Intensification of Tropical Storm Fay (2008), Natural Hazards, DOI 10.1007/s11069-011-9841-5 Chang H.G, D. Niyogi, A. Kumar, C. Kishtawal, J. Dudhia, F. Chen, U.C. Mohanty, M. Shepherd, 2009:Possible

relation between land surface feedback and the post-landfall structure of monsoon depression,

Geophysical Research Letters, 36, L15826

Large number of studies emerging in literature beyond our group from India (UC Mohanty group; IITM work by Saha, Roxy Matthews; IIT Bombay Ghosh, Karmakar; IISc Arindam; COLA Dirmeyer; TN Krishnamurti; Dimri at JNU and more (summarized in book chapter)

Irrefutable evidence that at micro, meso, regional

and climatic scales land surface representation has a detectable and important impact on high impact monsoon weather events and climate

Improved land state can lead to improved predictions – movement of monsoon trough post

• nset; tracks; depressions; mesoscale heavy rain

events; thunderstorms; urban flood causing rains;…

So whats next?

Broader community needs regional, local information at

• Larger, longer scale (infrastructure design)
• Seasonal scales (hydromet, water resources, ag)
• Urban mesoscale (high impact weather local

effects- fogs, urban floods) Decision makers will use what ever can be deduced if we do not provide the information! How to gear our process scale understanding, modeling framework to be responsive to that (need for multiscale models)

When Data and Evidence Are (made) Available- Knowledge follows

When Data and Tools Are (made) Available- Knowledge-based Action or Decisions can follow

Examples of next steps

• Community workshop, coordinated activities?
• Create new data sets (gridded products and

new observations- good spatial representation but think few research grade Critical Zone Observatories also)

• Data fusion techniques to create analysis and

model input datasets (greater need to work with computational data science/graphics community)

Examples of next steps (continued)

• Possible Quantitative Precipitation Forecast

(QPF) improvements at regional scales (need to customize mesoscale/regional models; improved land state; PBL; CP schemes etc)

• Generate / identify golden cases with datasets

that community can test, calibrate against.

• Help a U2U framework (making weather and

climate information useful to usable)

Agclimate4u.org

Transforming Climate Variability and Change Information for Cereal Crop Producer

(Source: Agcimate4u.org)

Integrating the knowledge base and community approach into urban datasets and land models

AMS Board on Urban Environment ICUC10 will be in NYC in 2017

9th INTERNATIONAL CONFERENCE ON URBAN CLIMATE

Procedural modeling of coupled urban land atmospheric interaction

Source: Garcia-Dorado, I. et al (2017), ACM Trans. of Graphics, in press.

Current (Simulated) View

• Key notion is that model is generated automatically from GIS style

information of the city – it is a plausible model and not a brick-by- brick, but it is fully changeable (note: simulation initially for Bangalore style)

Daniel Aliaga, Purdue

Future (Simulated) View

• We automatically simulate a plausible

future and the resulting city within a few seconds!

Population Jobs

Daniel Aliaga, Purdue

Future (Simulated) View

Daniel Aliaga, Purdue

Google Earth, Open Street Maps and Landsat Imagery based reclassification of Indian Cities, with Local Survey and Verification, and Rendering. Map released to broader community.

Bangalore New Urban Climate Zone map (also done for New Delhi; Chennai; Mumbai can do for all the 100 + cities)

devel eloping 3 3D u urban featur ures es f from m 2D m maps for WRF RF t to hydrolo logica cal m l models. . Objective : Use 3D reconstruction from land use information and integrate the same within the WRF framework Example of UCZ and satellite data based reconstruction of urban cluster height for a city across a river bank and inlets (white space between the two building clusters)

Improved Physics + Human Feedbacks (not just in ESMs) + Computational Tools and Approaches = Enhanced Land Representation  Improved monsoon rainfall variability prediction.

Scale

Context

Physics

Model

resolution

Definition

Processes

Address this…

Evaluation Metric

?? ADD HUMANS!

Thoughts..

• With improved land over monsoon region, case

specific success has been very good

• Translating success to operations is not 1:1

(success depends on land controls / evolutions allowed / constraints on the parameters…PBL coupling; convection scheme, resolution, depends

• n other parameters…)
• More coordinated model cal/val and development

with the Indian groups.

Interactions between Land Surface Processes & Monsoon Intra- Seasonal Oscillations (ISOs), Seasonal Rainfall

Saha S. K. , S. Halder, K. K. Kumar, B. N. Goswami (2011), Pre-onset land surface processes and internal interannual variabilities of the Indian summer monsoon, Climate Dynamics, 36, 2011-2089, DOI 10.1007/s00382-010-0886-z. Saha, S. K. , S. Halder, A. S. Rao, and B. N. Goswami (2012), Modulation of ISOs by Land Atmosphere Feedback and Contribution to the Interannual Variability of Indian Summer Monsoon, Journal of Geophysical Research, 117, D13101, doi:10.1029/2011JD017291. Rai, A., S. K. Saha, S. Pokhrel, K. Sujith, S. Halder (2015), Influence of pre-onset land-atmospheric conditions on the Indian summer monsoon rainfall variability, Journal of Geophysical Research, 120, DOI:10.1002/2015JD023159, 1-13

Non-ENSO Sources of Predictability of the ISMR in NCEP CFSv2 (Eurasian Snow)

Saha, S. K., S. Pokhrel, K. Salunke, A. Dhakate, H. S. Chaudhari, H. Rahaman, K. Sujith, A. Hazra, and

• D. R. Sikka (2016), Potential predictability of Indian summer monsoon rainfall in NCEP CFSv2, J. Adv.
• Model. Earth Syst., 8, 96–120, doi:10.1002/2015MS000542.

Saha, S. K., K. Sujith, S. Pokhrel, H. S. Chaudhari, and A. Hazra (2017), Effects of multilayer snow scheme on the simulation of snow: Offline Noah and coupled with NCEP CFSv2, J. Adv. Model. Earth Syst., 9, doi:10.1002/2016MS000845.

Slide courtesy: Dr Subodh Saha, IITM

“Overall, these results highlight that evaluating and improving land– atmosphere coupling could potentially improve model performance across the globe.”

Improvements in land have benefited hydrological (offline) model performance but the impact on coupled atmospheric model has been modest (compared to observations) This could be intimately linked to the land atmosphere coupling terms used in the model.

New / ongoing Collaborations with IITM, IITBBS

• Immediate. Land – Atmosphere Coupling

Coefficient calibrations for the IMR (using the new 4km LDAS fields as proxy for surface flux measurements)

• Next. Noah-Crop in CFS/ LDAS (impacts over India)
• Future. Urban Datasets and Feedbacks

Provide guidance on how finescale land processes coherently organize to create regional feedback and potentially improve model results

• Large growth in in situ/ DWR/ satellite data

products – benefits would be limited

unless models have the capability to use

them– imperative we develop the model processes.

• New multiscale land process and management

feedback possible

• The benefits of land representation will

become even more notable as multiresolution models become available.

Final (final) thoughts