Spatial ModelData Comparison Project Conclusions Forward models are

  • Slides: 12
Download presentation
Spatial Model-Data Comparison Project Conclusions • Forward models are very different and do not

Spatial Model-Data Comparison Project Conclusions • Forward models are very different and do not agree on timing or spatial distribution of C sources/sinks. • Examination of NEE shows GPP is largest component of discrepancies. • Why?

Tower-Based Spectral Measures of LUE, GPP Provide Independent Data at Meter-Scale and Minute-Scale Resolution

Tower-Based Spectral Measures of LUE, GPP Provide Independent Data at Meter-Scale and Minute-Scale Resolution

Adding Spectral Measures of Vegetation Light Use Efficiency to Existing Towers and Tower Sites

Adding Spectral Measures of Vegetation Light Use Efficiency to Existing Towers and Tower Sites Gap-Filler for Tower Measurements Diagnostics For Process Models Scaling to Landscape, Region Globe? ?

Spectral Measures of Surface-Atmosphere Energy, Water, Carbon Exchange Carbon Assimilation Rate Directly From Satellite

Spectral Measures of Surface-Atmosphere Energy, Water, Carbon Exchange Carbon Assimilation Rate Directly From Satellite A= par radiative xfer x fapar NDVI x LUE PRI Latent heat LE = r. Cp [e*- ea] gcga g gc+ga gc = Ah/c Models compute gc gc = gc* x f([e*- ea] , Ta, soil moisture) 0≤f≤ 1 • f is a non-linear function of soil moisture gc* = max gc • Soil moisture is difficult to model • Precipitation & soil properties variability • Soil Moisture diffcult to measure in the presence of dense vegetation.

Photochemical Reflectance Index A Normalized Difference Reflectance Index (NDRI) Gamon et al. 1992, 1993,

Photochemical Reflectance Index A Normalized Difference Reflectance Index (NDRI) Gamon et al. 1992, 1993, 1997, Gamon & Surfus 1999 r 531 - r 570 r 531 + r 570 PRI = As LUE decreases r 531 and PRI decrease. Reference Band 570 nm Xanthophyll Band 531 nm At the landscape scale r 531 r 570 are also sensitive to variations in other pigments and conditions other than LUE; beta-carotene, lutein, variations in soil, litter background.

Fluxnet Canada Doug Fir British Columbia a b Figure 3: (a) A dual channel

Fluxnet Canada Doug Fir British Columbia a b Figure 3: (a) A dual channel radiometer is mounted on the DF 49 Fluxnet tower with a vertical zenith angle (VZA) of 62°. A motor moves the canopy sensor 360 o every 15 minutes. Data are recorded every 5 seconds, year round. (b) The instrument on the Fluxnet Canada Douglas fir research site

Measured Components For Douglas Fir

Measured Components For Douglas Fir

Slope of PRI vs LUE Doug Fir LUE ≈ (∂PRI/∂SF-0. 21)/0. 05

Slope of PRI vs LUE Doug Fir LUE ≈ (∂PRI/∂SF-0. 21)/0. 05

COMBINING MODEL, TOWER AND REMOTE SENSING MEASUREMENTS • • Remote sensing estimates of LUE,

COMBINING MODEL, TOWER AND REMOTE SENSING MEASUREMENTS • • Remote sensing estimates of LUE, in combination with process model estimates and tower-based estimates form a powerful set of independent measures for cross-validation of these independent methodologies. Process models calculate – components of gross primary production, e. g. LUE, GPP – autotrophic and heterotrophic respiration, NEP Compare process model predictions to spectral and eddy covariance (EC) measures of same – compare model outputs of these quantities to the minute-scale tower and remote sensing-inferred values – Spectral measures can help gap-fill tower EC flux data – flag and help diagnose problems in any of the methods when the diurnal or longer-term values diverge. Aircraft and satellite landscape-scale measures (available only under clear-sky conditions) could be compared with similar estimates from process models – examine divergences over the landscape – flag and diagnose problems • remote sensing algorithm • variables used by the process models such as soil texture, soil moisture, nutrient levels etc. – data assimilation

What needs to be added to the existing Canadian, U. S. , European and

What needs to be added to the existing Canadian, U. S. , European and International network of flux towers?

Tower Spectral Measurement Instruments Costs of sensors range from 25 K$ as in the

Tower Spectral Measurement Instruments Costs of sensors range from 25 K$ as in the AMSPEC from Hilker et al, to a full spectrum CCD at about 6 -10 K$ - (350 -1500 nm), to a simple PRI sensor using a light sensor and interference filter for about 2 K$ (which could be handmade). Scaling to Landscape Aircraft Spectral Measurements - Missions of opportunity? NASA AVIRIS, the Canadian CASI, EO-1, NASA LVIS Lidar, other Canadian Lidar flights and the NASA airborne SLICER instrument (an aircraft Lidar flown during BOREAS), ASAS, multi-angle along-track hyperspectral flown during BOREAS Scaling to Region and Globe -- Not Global yet, but… Chris Proba, EO-1, MERIS MODIS