Applications of the Terrestrial Observation & Prediction System Forrest Melton CSU Monterey Bay, Seaside, CA Ecological Forecasting Lab NASA Ames Research Center, Moffett Field, CA With contributions from: Rama Nemani, Petr Votava, Andrew Michaelis, Christina Milesi, Hirofumi Hashimoto, Weile Wang William Reisen & Chris Barker, UC Davis Support from: NASA Applied Sciences Program: REASo. N Award, Decision Support through Earth Science Research Results Award MODIS NDVI, Mesoamerica, Jan. 2 -16 SERVIR Workshop Panama City, Panama, Mar. 1, 2007 1
Outline - Ecological monitoring and forecasting - Types of ecological forecasts - Modeling framework for producing EFs - the Terrestrial Observation and Prediction System (TOPS) - Examples of ecological forecasts and TOPS products - TOPS-SERVIR datasets - EF Applications for Mesoamerica MODIS Terra Image of Panama, February 24, 2004 (MODIS Rapid Response) - Anomaly detection and landscape monitoring - NPP and agricultural production monitoring and forecasting - Vector and disease risk mapping and ecological forecasting 2
What is Ecological Forecasting? Ecological Forecasting (EF) predicts the effects of changes in the physical, chemical, and biological environments on ecosystem state and activity. TOPS daily soil moisture forecast, Dec 30, 2006 3
Changing Surface Temperatures Why we need ecological forecasting? 4
Why are Ecological Forecasts Important? • Ecological forecasts offer decision makers estimates of ecological vulnerabilities and potential outcomes given specific natural events, and/or management or policy options. • Ecological forecasting is critical in understanding potential changes in ecological services, before they happen (early warning), and are critical in developing strategies to off-set or avoid catastrophic losses of services. • Goal is to develop management strategies and options to prevent or reverse declining trends, reduce risks, and to protect important ecological resources and associated processes. Bruce Jones, NCSE, Forecasting Environmental Changes, 2005 Foster interdisciplinary activity 5
Short-term Monitoring and Forecasting Sacramento river flooding, California Irrigation requirements Based on weather forecasts, conditioned on historical ecosystem state Days to a week 6
Mid-term/Seasonal Forecasts: Water resources, Fire risk, Phenology ENSO-Rainfall over U. S El Nino La Nina Based on ENSO forecasts Weeks to months 7
Long-term Projected Changes Rizzo & Wilken, Climatic Change, 21(1), pp. 37 -55, 1992 Based on GCM outputs Decades to centuries 8
TOPS: Common Modeling Framework Monitoring Modeling Forecasting Multiple scales Predictions are based on changes in biogeochemical cycles Nemani et al. , 2003, EOM White & Nemani, 2004, CJRS 9
Access to a variety of observing networks Weather network Streamflow network Fluxnet Soil moisture network 10
Access to a variety of remote sensing platforms Integration across: Platforms, Sensors, Products, DAACs is non-trivial 11
Satellites: MODIS on Terra & Aqua Terra Launch: Dec. 18, 1999 First Image: Feb. 24, 2000 Retrospective to real time Operational remote sensing Aqua Launch: May 04, 2002 First Image: June 24, 2002 12
Multiple Instruments per Mission Example: MODIS on Terra & Aqua Terra Satellite Launched Dec. 18, 1999 with five instruments (ASTER, CERES, MISR, MODIS, MOPITT) Aqua Satellite Launched May 4, 2002 with six instruments (AIRS, AMSR-E, AMSU, CERES, HSB, MODIS) MODerate resolution Imaging Spectroradiometer Orbit: 705 km, 10: 30 a. m. descending node (Terra) or 1: 30 p. m. ascending node (Aqua), sun-synchronous, near-polar, circular Swath Dimensions: 2330 km (cross track) by 10 km (along track at nadir) Data Rate: 10. 6 Mbps (peak daytime); 6. 1 Mbps (orbital average) Spatial Resolution: 250 m (bands 1 -2), 500 m (bands 3 -7), 1000 m (bands 8 -36) Design Life: 6 years 13
Multiple Products per Instrument: MODIS Measurements MOD 01 MOD 02 MOD 03 MOD 04 MOD 05 MOD 06 MOD 07 MOD 08 MOD 09 MOD 10 MOD 11 MOD 12 MOD 13 MOD 14 MOD 15 MOD 16 MOD 17 MOD 18 MOD 19 MOD 20 MOD 21 MOD 22 Level-1 A Radiance Counts Level-1 B Calibrated Relocated Radiances Relocation Data Set Aerosol Product Total Precipitable Water Cloud Product Atmospheric profiles Gridded Atmospheric Product (Level-3) Atmospherically-corrected Surface Reflectance Snow Cover Land Surface Temperature & Emissivity Land Cover/Land Cover Change Vegetation Indices Thermal Anomalies, Fires & Biomass Burning Leaf Area Index & FPAR Surface Resistance & Evapotranspiration Vegetation Production, Net Primary Productivity Normalized Water-leaving Radiance Pigment Concentration Chlorophyll Fluorescence Chlorophyll_a Pigment Concentration Photosynthetically Active Radiation (PAR) MOD 23 MOD 24 MOD 25 MOD 26 MOD 27 MOD 28 MOD 29 MOD 31 MOD 32 Suspended-Solids Conc, Ocean Water Organic Matter Concentration Coccolith Concentration Ocean Water Attenuation Coefficient Ocean Primary Productivity Sea Surface Temperature Sea Ice Cover Phycoerythrin Concentration Processing Framework & Match-up Database MOD 35 Cloud Mask MOD 36 Total Absorption Coefficient MOD 37 Ocean Aerosol Properties MOD 39 Clear Water Epsilon MOD 43 Albedo 16 -day L 3 MOD 44 Vegetation Cover Conversion MODISALB Snow and Sea Ice Albedo 14
Ability to integrate a variety of models Biogeochemical Cycling Crop growth/yield Pest/Disease Global carbon cycle Prognostic/diagnostic models 15
Ability to work across different scales of time and space Hours Days Years/Decades Weeks/Months Weather/Climate Forecasts at various lead times Downscaling 16
Standard TOPS Outputs MODIS PRODUCTS (8 days/Annual) TOPS-NOWCASTS (daily) 1 LAI 17 TOPS-SNOW 2 FPAR 18 TOPS-SOIL MOISTURE 3 GPP/NPP 19 TOPS-ET 4 LST-TERRA/AQUA 20 TOPS-OUTFLOW 5 NDVI 21 TOPS-GPP/NPP 6 EVI 22 TOPS-PHENOLOGY 7 LANDCOVER* 23 TOPS-VEG STRESS 8 ALBEDO 9 SNOW 10 FIRE METEOROLOGY (Daily) TOPS-FORECASTS (5 days to 180 days) 24 BGC-LAI/PHENOLOGY 25 BGC-SOIL MOISTURE 11 MAX TEMPERATURE 26 BGC-OUTFLOW 12 MIN TEMPERATURE 27 BGC-ET 13 RAINFALL 28 BGC-VEG STRESS 14 SOLAR RADIATION 29 BGC-SNOW 15 DEW POINT/VPD 30 Spatial Resolution: 30 m to 1 km Temporal Resolution: 1 to 30 days Data Presentations: Nowcast, forecast, anomaly, cumulative, current average Data Formats: Binary, Geo. TIFF, JPEG, PNG BGC-GPP/NPP 16 DATA PROPERTIES DEGREE DAYS Metadata: ESML & OGC compliant Delivery Mechanisms: FTP, WMS, Web * Once a year 17
Standard TOPS Outputs: Local to Global Scales Global NPP Anomalies U. S. Gross Primary Productivity California Daily Soil Moisture Estimates Yosemite Minimum Temperatures Napa Valley Forecasted Vineyard Irrigation Demands Spatial scales from 0. 5 degrees to 4 m. Temporal scales from yearly to daily. 18
TOPS/SERVIR Products for Mesoamerica 19
TOPS/SERVIR Products for Mesoamerica TOPS MODIS Ecosystem Products – 1 km spatial resolution products – 8 / 16 day composites – Land Surface Temperature (LST) – Leaf Area Index (LAI) – Fraction of Photosynthetically Active Radiation (FPAR) absorbed – Normalized Difference Vegetation Index (NDVI) – Enhanced Vegetation Index (EVI) – Gross Primary Productivity (GPP) 20
TOPS MODIS Products for Mesoamerica: LST Land Surface Temperature (LST) • Land surface temperature at time of satellite overpass • Degrees Kelvin • Composited from the MODIS MOD 11 A 1 daily LST values • Derived from MODIS bands: 31 (11. 03 µm) 32 (12. 02 µm) • MOD 11 algorithm incorporates information from MODIS cloud mask, atmospheric profile, land cover, and snow cover 21
Normalized Difference Vegetation Index (NDVI) • Provides a measure of vegetation density and health. • Used in studies of landscape change, crop monitoring, and risk mapping for vectorborne diseases. • Calculated from the visible and nearinfrared light reflected by vegetation • Healthy vegetation absorbs most of the visible light that hits it, and reflects a large portion of the near-infrared light. Unhealthy or sparse vegetation (right) reflects more visible light and less nearinfrared light. Credit: Robert Simmon • NDVI values for a given pixel always result in a number that ranges from minus one (-1) to plus one (+1) 22
TOPS MODIS Products for Mesoamerica: NDVI Normalized Difference Vegetation Index • 16 -day values composited from the MODIS MOD 13 A 1 daily NDVI values • Derived from MODIS bands: 1 (Red; 620 -670 nm) 2 (NIR; 841 - 876 nm) • The reflectance values are the surface bidirectional reflectance factors for MODIS bands 1 (620 670 nm) and 2 (841 - 876 nm) • Tends to saturate over high biomass regions; sensitive to atmosphere and canopy variations. 23
Enhanced Vegetation Index (EVI) • EVI developed to provide improved vegetation signal in high biomass regions. • De-couples the canopy background signal and corrects for residual atmospheric influences. • Input reflectances may be atmosphericallycorrected or partially atmospheric corrected for Rayleigh scattering and ozone absorption. Tracking vegetation condition with MODIS EVI in the Amazon Basin Credit: Huete et al. 2006. GRL 33. http: //tbrs. arizona. edu/project/MODIS/evi. php 24
TOPS MODIS Products for Mesoamerica: EVI Enhance Vegetation Index (EVI) • Composited from the MODIS MOD 13 A 1 daily EVI values • Derived from MODIS bands: 1 (Red; 620 -670 nm) 2 (NIR; 841 - 876 nm) 3 (Blue; 459 -479) nm • Better performance than NDVI in the tropics and other regions with high biomass. 25
TOPS MODIS Products for Mesoamerica: LAI & FPAR Leaf Area Index (LAI) • Measure of plant canopy structure. • One sided leaf area per unit ground area. • Unitless index; values of <1. 0 indicate incomplete canopy closure. • Typical values range from 0 to 7. • Highly related to a variety of canopy processes, such as water interception, evapotranspiration, photosythesis, respiration, and leaf litterfall. • Fraction of Photosyntethically Active Radiation (FPAR) absorbed • Measure of the proportion of available radiation in the photosynthetically active wavelengths of the spectrum (0. 4 - 0. 7 microns) that is absorbed by the canopy. • Radiation term; more directly related to remotely sensed variables (such as NDVI) than LAI. • Can be used to translate direct satellite data such as NDVI into simple Used in ecological and climate models as estimates of primary production. a representation of canopy structure. • Both FPAR and LAI are used in biogeochemical models to estimate primary productivity. http: //www. ntsg. umt. edu/remote_sensing/leafarea/ 26
TOPS MODIS Products for Mesoamerica: LAI & FPAR • 8 -day composites • Provide measures of canopy structure and photosynthetic activity 27
TOPS MODIS Products for Mesoamerica: GPP Gross Primary Productivity (GPP) • Measure of gross CO 2 assimilation in vegetation. • Estimates of GPP from satellite data based on the concept of radiation use efficiency (RUE) • RUE is a measure of how effective vegetation is in using PAR to converting solar radiation in the wavelength band from 0. 4 - 0. 7 micrometers to fix CO 2 from the atmosphere as carbohydrate for growth and respiration • Varies depending on vegetation condition and environmental conditions. • Net primary productivity (NPP) is difference between GPP and amount of CO 2 lost to respiration. http: //www. ntsg. umt. edu/remote_sensing/netprimary/ 28
TOPS MODIS Products for Mesoamerica: GPP Gross Primary Productivity (GPP) • Composited from the MODIS MOD 17 A 1 daily GPP values • Derived from MODIS FPAR and LAI, and utilizes GMAO surface meteorology and a biome properties look-up table to produce model-derived estimates. 29
Mesoamerican Anomalies 30
Mesoamerican Anomalies Using anomaly persistence to assess significance 31
Mesoamerican Anomalies 32
TOPS/SERVIR Climate Products for Mesoamerica TOPS Climate Products – Daily, 1 km spatial resolution demonstration products – Gridded meteorological surfaces derived from station observations using modified Daymet algorithm (Thornton et al. 1997, 2000) – Derived from 90 meteorological stations in Mesoamerica that report to NOAA Global Summary of the Day (GSOD) – Maximum temperature (C◦) – Minimum temperature (C◦) – Precipitation (mm) – Vapor Pressure Deficit (Pa) – Shortwave Radiation (watts/m 2) 33
TOPS/SERVIR Products for Mesoamerica Minimum Temperature Shortwave Radiation Precipitation Vapor Pressure Deficit 34
Reporting Stations used by TOPS for NA / MA • Accuracy of gridded meteorological surfaces directly related to density of meteorological staitons • 3000 – 6000 stations in U. S. (depending on time period) • 700 stations in California • 90 stations in Mesoamerica • Sample data set prepared for SERVIR for 2000 -2005 35
Examples of TOPS Applications • Landscape monitoring and trend analysis • Soil moisture estimates and irrigation demand forecasting • Ecosystem monitoring for protected area management • Mapping of insect vectors for vector-borne diseases 36
Long-Term Monitoring and Trend Analysis 37
Agricultural Management Applications of EF Short-term: Vineyard Irrigation Forecasts Irrigation Forecast for week of July 19 -26, 2005 Tokalon Vineyard, Oakville, CA CIMIS Measured Weather Data through July 18, 2005 NWS Forecast Weather Data July 19 -26, 2005 0 meters 1000 N 0 30 Forecast Irrigation (mm) Seasonal Fully automated web delivery to growers 38
Agricultural Management Applications of EF Mid-range: Forecasting the onset of growing season Based on White and Nemani, RSE, 2006 39
Agricultural Management Applications of EF Mid-range: Forecasting crop yields • Lobell, Cahill, and Field (2006) recently demonstrated the use of climate data (temperature and precipitation) to predict seasonal yields for 12 major crops in California • Lead time of weeks to months • Forecasts capture more than 50% of the variability in yield anomalies, and as much as 89% Forecasted versus observed yields for 12 California crops (from Lobell et al. , 2006, California Agriculture, 60(4): 211 -215. 40
Long-range: Net Primary Productivity Anomalies 56% of global population lives in regions where water availability strongly influences NPP. Significant correlations between MEI and NPP were found over 63% of the vegetated surface, inhabited by 3. 3 billion people. Population(millions) 12 Milesi et al. , Glob. Pl. Change, 2005 Hirofumi et al. , JGR-Atm, 2004 7 41
Key Questions for Protected Area Managers • What is the current status of ecosystems in and adjacent to the park/protected area? • How are they changing? • How will they change in the future (in response to changes in climate and land use)? • How do these changes impact resource management? MODIS Direct Broadcast image of a fire event in Yosemite National Park, September, 2005. 42
TOPS EF Tools for Protected Area Management • Monitoring and forecasting of ecosystem conditions • Automated event and anomaly detection • Monitoring and forecasting of summer streamflow, soil moisture, and vegetation stress conditions for fire risk forecasting Monitoring and modeling of GPP, total aboveground carbon, and current aerosol levels to assess potential air quality impact of management initiated burns Snowpack monitoring and forecasting for runoff prediction Long-term simulations for analysis of potential impacts of climate change on ecosystem conditions • • • Observed vs. predicted snow cover, Merced Watershed, Yosemite National Park, 2000 -2004 43
Anomaly Detection for Resource Monitoring Automated anomaly detection and trend analysis assist resource managers in identifying significant events and focusing ground-based monitoring and management efforts. 44
Interpreting Anomalies Ground-based observations key to validating and interpreting anomalies. 45
TOPS Data Fusion: Trend Analysis for Features of Interest 46
Ecological Forecasting and Public Health Potential Areas of Contribution: • Air quality (fire frequency, land cover change / desertification and particulates) • Water quality (flooding, drought) • Food security PATHOGEN HOST VECTOR CLIMATE HYDROLOGY HABITAT • Vector-borne disease 47
Ecological Forecasting: Lyme Disease CDC National Lyme Disease Risk Map Predictive risk map of habitat suitability for Ixodes scapularis in Wisconsin and Illinois. Fish & Howard. Morbidity and Mortality Weekly Report, 48, pp 21 -24, 1999 Guerra et al, Emerging Infectious Diseases, Vol. 8(3), 2002 48
Ecological Forecasting: Malaria Soil moisture • Patz, J et al. Tropical Medicine & International Health, 3. 10, (1998): 818 -827 – Modeled soil moisture / surface-water availability in Kenya to predict biting rates (climate, land cover, and soil type as inputs to model) – Soil moisture was a better predictor than precipitation, and comparable to NDVI from AVHRR Land cover change • Vittor, A. et al. Am. J. Trop. Med. Hyg. , 74. 1, (2006): 3 -11 – In deforested sites in Peruvian Amazon, A. darlingi had a biting rate > 278 times higher than the rate determined for areas that were predominantly forested. Regression of the log of An. Gambiae and An. Funestus HBR and modeled soil moisture. Source: Patz et. al. 1998 49
Regional Nowcasts: California Tracking parameters related to mosquito abundance: Meteorology Hydrology Vegetation Ecosystem 50
Land use and seasonal patterns of mosquito abundance Average seasonal profiles for Cx. tarsalis counts per New Jersey light trap-week by bioregion, 1981 -2000. Cx. tarsalis per trap-week Effect of land use on seasonal patterns of mosquito abundance in Sacramento, CA. 600 500 400 300 200 100 0 May Jun Jul Aug Sep Oct Month Figures courtesy of CM Barker 51
Summary of TOPS Applications for Mesoamerica Landscape Monitoring • Daily / weekly / monthly satellite- and model-based measures of ecosystem condition • Identification of anomalies and trends • Potential use as inputs to annual ‘state of the nation’ assessments Ecosystem Modeling • Soil moisture • Evapotranspiration • Watershed outflow • Accuracy and spatial resolution determined by availability of groundbased observations (soil classification map, hydrologic data, meteorological data) Research & Application Development • Inputs to other models • Crop yield monitoring • Irrigation demand forecasting • Fire risk mapping • Flood forecasting • Requires collaboration with MA research teams 52
Summary • TOPS is a modeling framework that uses ecosystem models to ingest satellite observations, meteorological observations and forecasts, and ancillary data to monitor ecosystem conditions and produce ecological forecasts. TOPS has been used to develop an initial suite of ecosystem products for SERVIR. • Remote sensing & ecological forecasts provide an important supplement to ground-based monitoring and climate forecasts for protected area management, agricultural management, and public health decision support / disease risk mapping. EF can assist with translation of climate variables into measures of ecosystem conditions associated with disturbance events, agricultural productivity, and pathogen-vector-host interactions. • Rapidly growing number of successful examples of applications that utilize ecosystem models to integrate satellite, climate, and ground-based observations to develop predictive models. Characterizing and communicating uncertainty remains a key issue. • Further progress depends on… – Improved in-situ monitoring networks. – Better linkages among models. – Comprehensive framework for data access and management. 53
54
Скачать презентацию Applications of the Terrestrial Observation Prediction System
cc8c8bacd2cf74515b00d71014329a54.ppt