Climate Impacts of Agriculture Related Land Use Change in the US Jimmy Adegoke 1, Roger Pielke Sr. 2, Andrew M. Carleton 3 1 Dept. Of Geosciences, University of Missouri-Kansas City 2 Dept. of Atmospheric Sciences, Colorado State University 3 Dept. of Geography, The Pennsylvania State University World Meteorological Organization (WMO) Committee on Agricultural Meteorology (CAgM( CAgM) ) Expert Team Meeting on the Contribution of Agriculture to the State of Climate Ottawa, Canada, 27-30 September 2004
Presentation Outline Cropland/Forest Impacts on Convective Cloud Development in the US Midwest : Empirical studies Agriculture-related land use change impacts on seasonal climate in the Central US: Modeling studies Crop-climate modeling Issues, challenges & questions
Current & Potential Natural Vegetation [Copeland et al., 1996]
Agriculture - Human Imprint on the Land Surface Spring Summer
Land Surface-Atmosphere Interactions Schematic of the differences in surface heat energy budget and planetary boundary layer over a forest and cropland.
Atmosphere Temp, Water Trace Gases Pollutants Heat, Moisture Radiation Light, Temp Moisture, Wind Community Composition & Structure Landscape Modification Trace Gases & Pollutants Surface Physiology & Hydrology Water & Nutrients Agriculture Deforestation, etc. Biogeochemical & Hydrological Cycles Anthropogenic Activities Sec - Hours >1yr - 100yrs >1000yrs
Focus of Land Surface-Climate Work: Empirical Studies Impacts of changes in US Midwest land cover parameters (e.g. land cover, surface roughness, zones of land cover transitions) on convective cloudiness (Carleton et al., 2001 GRL Vol. 28, 1679-1684) Sensitivity of the AVHRR derived Normalized Difference Vegetation Index (NDVI) and the Fractional Vegetation Cover (FVC) to growing season surface moisture conditions (Adegoke and Carleton, 2002 JHM 4, 24-41).
Land Use Changes in Illinois Wetland 0% Prairie 2% water 3% River 2% Jackson County Urban 3% Forest 25% Prairie 0% Wetland 2% Forest 93% Agriculture 68% Barren 1% Water 1% 1820 1980 Wetland 5% water 5% River 0% Lake County Forest 9% Urban 34% Prairie 0% Wetland 2% Water 2% Barren 1% Prairie 29% Forest 61% Agriculture 52% [From Iverson & Risser, 1987]
Land Use/Land Cover Map of The Midwest Showing Sampling Locations of Convective Cloud Parameters
GOES INFRA RED & VISIBLE IMAGES 11 JUNE 1997 16:00 UTC
Stratification of Case Study Days: June-Aug. 1981-98 Strong Flow / Weak Flow : 500 mb Vector Winds
a) Mean BWF VIS: Crop MI Mean BWF IR TEMP: Crop MI Albedo 1.00 0.80 0.60 0.40 0.20 0.00 295.00 290.00 285.00 280.00 275.00 270.00 265.00 8pm 6pm 4pm 2pm 12Noon 10am 8am 6am 5pm 3pm 1pm 11am 9am 7am b) Mean BWF VIS: Boundary MI Mean BWF IR TEMP: Boundary MI Albedo 1.00 0.80 0.60 0.40 0.20 0.00 300.00 290.00 280.00 270.00 260.00 250.00 8pm 6pm 4pm 2pm 12Noon 10am 8am 6am 5pm 3pm 1pm 11am 9am 7am c) Mean BWF VIS: Forest MI Mean BWF IR TEMP: Forest MI Albedo 1.00 0.80 0.60 0.40 0.20 0.00 295.00 290.00 285.00 280.00 275.00 270.00 265.00 260.00 8pm 6pm 4pm 2pm 12Noon 10am 8am 6am 5pm 3pm 1pm 11am 9am 7am
a) Mean BWF VIS: Crop IN d) Mean BWF VIS: Crop MO Albedo 1.00 0.80 0.60 0.40 0.20 0.00 Albedo 1.00 0.80 0.60 0.40 0.20 0.00 5pm 3pm 1pm 11am 9am 7am 5pm 3pm 1pm 11am 9am 7am b) Mean BWF VIS: Boundary IN e) Mean BWF VIS: Boundary MO Albedo 1.00 0.80 0.60 0.40 0.20 0.00 7am 9am 11am 1pm 3pm 5pm Albedo 1.00 0.80 0.60 0.40 0.20 0.00 7am 9am 11am 1pm 3pm 5pm c) Mean BWF VIS: Forest IN f) Mean BWF VIS: Forest MO Albedo 1.00 0.80 0.60 0.40 0.20 0.00 Albedo 1.00 0.80 0.60 0.40 0.20 0.00 5pm 3pm 1pm 11am 9am 7am 5pm 3pm 1pm 11am 9am 7am
a) Mean WF VIS: Crop MI d) Data Range WF VIS: Crop MI Albedo 1.00 0.80 0.60 0.40 0.20 0.00 Albedo 1.00 0.80 0.60 0.40 0.20 0.00 5pm 3pm 1pm 11am 9am 7am 5pm 3pm 1pm 11am 9am 7am b) Mean WF VIS: Boundary MI e) Data Range WF VIS: Boundary MI Albedo 1.00 0.80 0.60 0.40 0.20 0.00 Albedo 1.00 0.80 0.60 0.40 0.20 0.00 5pm 3pm 1pm 11am 9am 7am 5pm 3pm 1pm 11am 9am 7am c) Mean WF VIS: Forest MI f) Data Range WF VIS: Forest MI Albedo 1.00 0.80 0.60 0.40 0.20 0.00 7am 9am 11am 1pm 3pm 5pm Albedo 1.00 0.80 0.60 0.40 0.20 0.00 7am 9am 11am 1pm 3pm 5pm
Mean SF VIS Brightness: Crop MI Mean BWF VIS Brightness: Crop MI 160 140 120 100 80 60 40 20 0 120.00 100.00 80.00 60.00 40.00 20.00 0.00 8pm 6pm 4pm 2pm 12Noon 10am 8am 6am Mean SF VIS Brightness: Boundary MI Mean BWF VIS Brightness: Boundary MI 140 120 100 80 60 40 20 0 100.00 80.00 60.00 40.00 20.00 0.00 8pm 6pm 4pm 2pm 12Noon 10am 8am 6am Mean SF VIS Brightness: Forest MI Mean BWF VIS Brightness: Forest MI 140 120 100 80 60 40 20 0 100.00 80.00 60.00 40.00 20.00 0.00 6am 8am 10am 12Noon 2pm 4pm 6pm 8pm
Average Cloud Top Brightness Temperature: MI & MO BWF Average Maximun Brightness Temp (MI) 102.00 100.00 98.00 96.00 94.00 92.00 90.00 88.00 Crop Boundary Forest BWF Average Maximun Brightness Temp (MO) 110.00 108.00 106.00 104.00 102.00 100.00 98.00 Crop Boundary Forest
Thermodynamic indices for selected weak flow days: 12Z Radiosonde Data: MI vs MO CCL (mb) CCL-EL (mb) K-Index (mb) Tc ( o C) Mean Tv ( o C) Mean T-Tv ( o C) RH o / o Water Content Michigan 749 480 12 32 16 3.7 68.18 Missouri 827 528 29 26.8 20.5 2.7 78.33
Sensible and Latent Heating of the Atmosphere Required for Initiation of Convective Clouds vs. Bowen Ratio [Rabin et al., 1990]
Summary of Cloud Research Findings Analyses of Visible and IR GOES cloud data for contrasting circulation regimes indicate some cloud-land cover associations across major crop-forest boundaries. Land cover boundary zones are shown to be favored areas for enhanced cloud development under moderate mid-tropospheric (< 30 m/s) flow conditions. The boundary zones tend to behave like regions of differential vertical circulations (i.e., NCMCs)
Focus of Land Surface-Climate Work: Modeling Studies Improving the representation of land surface heterogeneity (land cover; soil moisture; soil type) in the Colorado State University Regional Atmospheric Modeling System (RAMS) (Adegoke et al. 2003; Strack et al. 2003; Rozoff et al., 2003) Developing protocols for a more realistic description of seasonally and interannually varying vegetation cover and growth rates in regional climate models (Adegoke et al. 2004; Eastman et al. 2001; Lixin Lu et al., 2002).
Lessons Learned Realistic representation of spatial heterogeneity of land surface parameters improves model simulation of regional-scale effects of agriculture-related land use changes on climate and terrestrial biophysical processes. Key Parameters: - Land Cover - Soil Moisture -LAI -Soil Type - Soil Temperature
Soil Moisture Impacts Over High Plains, if model soil moisture is low Forecast CAPE too low, about half the observed CAPE Forecast of instability insufficient
Soil Moisture Simulation with Different Soil (a); Matching Soil (b) (a) (b) LDAS Evaluation Team: Alan Robock et al., 2004
Soil Moisture LDAS Evaluation Team: Alan Robock et al., 2004
Recent Improvements in RAMS-LEAF2 1. Protocols for ingesting variable soil moisture 2. Incorporation of high-resolution land cover data (30 m) from the USGS NLCD database 3. Specification of variable soil type from the FAO soil type database 4. Protocols for ingesting NDVI and derivation of LAI from NDVI 5. RAMS-Century coupling for explicit modeling of the seasonal evolution of vegetation in the simulation of seasonal climate.
Map of U.S. High Plains Aquifer
Acreage of Rain fed & Irrigated Corn Farming in Nebraska (1950-1988) 3000000 2500000 2000000 1500000 1000000 Rainfed Irrigated 500000 0 1950-51 1954-55 1958-59 1962-63 1966-67 1970-71 1974-75 1978-79 1982-83 1986-87 1990-91 1994-95 1998-99 Area (ha) Year
Nebraska Irrigation Modeling Project Complex changes in the lower atmosphere (PBL) radiation budget can result from large-scale land use changes of this magnitude (e.g., vapor flux CAPE) This study was designed to evaluate the changes in the summertime surface energy budget & convective rainfall parameters due to irrigation in Nebraska using RAMS.
RAMS Modeling Domain Coarse Grid: 40 km ; Fine Grid:10 km; Domain Height: 20km
a) Kuchler Potential Vegetation b) OGE Dry Run c) OGE + Current Irrigation Control Run (a) (b) (c)
Summary of Model Results Significant inner domain area-averaged difference between the Control and Dry runs: - 36% increase in surface latent heat flux - 15% decrease in surface sensible heat flux - 28% increase in water vapor flux at 500m -2.6 o C elevation in dew point temperature -1.2 o C decrease in near surface temperature Greater differences observed between the Control and Natural Vegetation runs e.g., - Near ground temperature was 3.3 o C warmer & surface sensible heat 25% higher in the Natural run. [Adegoke et al., 2003 Monthly Weather Review 131(3), 556-564.]
ClimRAMS Coupled with CENTURY (Lu et al., 2001, Journal of Climate) RAMS{temp, swin, prcp, rh, (u,v,w)} CENTURY{LAI, roots, Zo, evap, transp, vegfrac, vegalb}
Satellite-derived Leaf Area Index Derived from AVHRR 10- day composite NDVI NDVI LAI following Sellers et al. (1996) and Nemani et al. (1996) Lu et al., 2001 Derived LAI for Central U.S. in dry (1988), average(1989), and wet (1993) years. Average JJA NDVI for Central U.S.
Comparison of LAI Forcing Lu et al., 2001 STRONG DIFFERENCES Magnitude of LAI Heterogeneity of LAI SOME DIFFERENCE Seasonality of LAI Default LAI in Inner (50 km) Grid NDVI-derived LAI in Inner (50 km) Grid
Good Agreement between Model Predictions & Observations e.g. Domain-average maximum and minimum air temperature and precipitation for inner grid during 1989 (selected as an average year) for the run with NDVIderived LAI Lu et al., 2001
Runs Broadly Agree With Observations e.g. Distribution of maximum and minimum temperature and precipitation for inner grid for the run with NDVI-derived LAI January-March 1989 June-August 1989 Lu et al., 2001
Physical Mechanism for Precipitation Increase Lower domainaveraged LAI allows more solar radiation to reach the surface, increasing CAPE Spatial variability in LAI triggers mesoscale circulations.
RAMS-Century Coupling Strategy and Design Differences in spatial and temporal resolutions: RAMS: 3-D, CENTURY: 1-D time step: minute vs. day Internet Stream Socket Client/Server mechanism Both atmospheric forcings and biospheric parameters are prognostic variables Lu et al., 2001
Coupled Model Captures 2-way Feedbacks Default Coupled LAI response of CENTURY is different after harvest when run in coupled mode. The coupled model gives a response in modeled precipitation. Lu et al., 2001
Coupled Model Simulated Climate Lu et al., 2001
Conclusions From Lu et al. 2001 Both satellite-derived and model-calculated LAI produce a significant impact on the modeled seasonal climate In both cases, the climate is cooler and produces more precipitation relative to using RAMS default LAI The effect of heterogeneity in LAI appears to be the dominant factor in producing these differences Including realistic description of heterogeneous vegetation phenology influences the prediction of seasonal climate.
Looking Ahead The challenge: Fully coupled crop-climate model capable of investigating the 2-way interactions of cropclimate system under a wide range of conditions. Must include feedbacks of crop growth on surface climate Will require much stronger cross-disciplinary interaction/collaboration between Agricultural and Atmospheric Sciences
Discussion Issues/Questions Crop models tend to operate at the field/plot spatial scale while climate models typically have horizontal resolutions of a few km to 100~200 km. Addressing this spatial scale disparity is not trivial. Are there additional local terrain and surface/vegetation characteristics that should be considered in crop-climate simulations that may not currently reflected in climate models? Crop growth parameterization issues: Century vs CERES-maize model vs General Large Area Model for annual crops (GLAM)