Thursday, December 14, 2006
Joshua B. Fisher: PROGRESS REPORT
Quantifying Evapotranspiration in California Forest Ecosystems:
Remote Sensing vs. Flux Measurement
Proposal Recap
Evapotranspiration, a major component in terrestrial water balance and net primary productivity models that control larger general circulation and global climate change models, is difficult to measure and predict, especially on a landscape spatial scale. The objectives of my study are to estimate evapotranspiration from remotely sensed data by scaling ground-based measurements and ecosystem models to FLUXNET data and satellite remote sensing. Both NASA’s Earth Observing System and a global network of ecosystem flux measurement sites (FLUXNET) have produced water vapor estimates that will be validated and analyzed in this project at two study sites in California: Tonzi Ranch (P.I. Dennis Baldocchi) and Blodgett Forest (P.I. Allen Goldstein). I will examine uncertainties in remotely sensed data for evapotranspiration modeling, and offer methodological recommendations by documenting problems and solutions from working with such data and scaling models. It is critical that the observational and modeling activities be linked together, as observations that do not exist in the context of a model-based hypothesis can provide data but little insight, while models unconstrained by observations are frequently of little use in explaining the past or present, let alone for predicting the future. This project, as an integral part of water and energy cycle research, is part of an ultimate objective to close the water budget worldwide; the overall goal is to deliver reliable estimates of precipitation minus evapotranspiration over the whole surface of the earth, and success will depend on a combination of measurements and model estimates of evapotranspiration.
Current Work and Results
Plant Ecophysiology
As my project scales from plant to ecosystem to landscape, I shall commence at the plant ecophysiological level. No biophysical variable is directly measurable at a global scale; consequently, none can be comprehensively validated (Running et al. 2000). A coherent Earth science research strategy must combine findings from global observations with insight gained from specialized studies; in situ measurements made at the Earth surface or in the atmosphere provide “ground truth” against which space-based measurements are compared, thus increasing our knowledge of processes through comprehensive characterization of specific regions, which is not usually possible with remote sensing techniques. To gain understanding of plant-water relations, I studied plant ecophysiology under Professor Todd Dawson. Relevant topics included: the principle of limiting factors, plants and microclimates, radiation balance and leaf energy budgets, stomatal and biochemical control of leaf gas exchange, stable isotopes, water-use efficiency, variation in photosynthetic pathways and carbon allocation, photosynthetic adaptation to light and temperature, canopy architecture and productivity, water in plants and in soils and the atmosphere, root systems and water capture, water use and tissue water relations, plant architecture and hydraulic conductivity, and adaptation to water stress and nutrient availability. These plant-level characteristics will guide ecosystem and landscape responses, which will influence larger-scale modeling.
I have begun my field site and measurement preparation by logistically planning collaborations with the research groups and principal investigators at Blodgett Forest and Tonzi Ranch. Furthermore, I have been reading extensively on measurement methods and theory for sap flow and soil evaporation. Blodgett Forest Research Station (38°53´N, 120°37´W, 1315 m) is a research forest of the University of California, Berkeley (Goldstein et al., 2000). The forest was planted in 1990 and was dominated by ponderosa pine trees (Pinus ponderosa Doug. E. Laws), the most common conifer species in North America. The canopy also included individuals of Douglas fir (Pseudotsuga menziesii), white fir (Abies concolor), giant sequoia (Sequoiadendron giganteum), incense-cedar (Calocedrus decurrens) and California black oak (Quercus kelloggii). The major understory shrubs were manzanita (Arctostaphylos spp.) and Ceonothus spp. (Xu et al., 2001a). In 1997, about 25% of the ground area was covered by shrubs, 30% by conifer trees, 2% by deciduous trees, 7% by forbs, 3% by grass and 3% by stumps. The forest area was in a stage of rapid growth, as exhibited by the 10% increase in leaf area index (LAI) between the 1997 (2.9-4.2) and 1998 (3.2-4.5) growing seasons. The site is characterized by a Mediterranean climate with an average annual precipitation of 163 cm (180 cm in 1997 and 117 cm in 1998), the majority of which falls between September and May, and almost no rain in the summer. The soil is a fine-loamy, mixed, mesic, ultic haploxeralf in the Cohasset series whose parent material was andesitic lahar. Tonzi Ranch (38° 25.896' N, 120° 57.959' W, 177 m) is a savanna site dominated by Blue oak (Quercus doublasii) and grazed grassland (bromus, frescue, oat, medusa head, rose clover) with scattered gray pine (Pinus sabinianai). The climate is also Mediterranean, although the site receives far less precipitation (56 cm annual mean) than does Blodgett Forest. The soil is of the Auburn series and is loamy, mixed, superactive, rocky, silt. I plan to set up sap flow measurements this summer on 6 ponderosa pine trees, 3 manzanita and 3 Ceonothus shrubs, and set up mini-lysimeter measurements of soil evaporation at Blodgett Forest. The selected trees will be coupled with respiration measurements managed by a colleague within my research group. Sap flow of shrubs is unique to understanding the partition of sub-canopy water use; my measurements will also be used to validate and partition a newly installed sub-canopy flux tower. Analogously, I will set up parallel sap flow measurements on 6 Blue oak trees, and set up mini-lysimeter measurements of bare soil evaporation and soil + grass evaporation. Similarly, the selected trees will be coupled with respiration measurements and will be used to partition a sub-canopy flux tower.
Ecosystem Ecology
As I scale up from plant-level to ecosystem scales, the functioning of plants begin to act as interacting systems in a different functional context. To understand ecosystem function and responses to water, I studied ecosystem ecology under Professors John Battles and Whendee Silver. Relevant topics included: water balance and budgets in global patterns of ecosystem distribution, energy and carbon balance, soil development and biology, nutrient cycling and decomposition, net primary productivity, population dynamics, equilibrium and non-equilibrium models of ecosystem dynamics, ecosystem development during primary and secondary succession, resource gradients on landscape patterns in ecosystem structure and function, biodiversity, and climate change. Ecosystem characteristics will guide the scaling of plant level measurements via sap flow to stand level responses via leaf area index (LAI) and percent cover. These same characteristics will guide scaling to remote sensing based on biophysical variables that can be made from various band spectrums like the normalized difference vegetation index (NDVI) and LAI relationship, temperature and thermal bands connection, and soil moisture and microwave spectrum link.
In my original proposal, I had planned on comparing a suite of process based and empirical evapotranspiration models (e.g. Monteith, 1965; Priestley and Taylor, 1972) to the ground-truthed measurements from the tower and tree scale measurements because remote sensing cannot detect evapotranspiration directly and thus evapotranspiration must be derived from models driven by various vegetation and meteorological variables. Evapotranspiration models vary widely in complexity from simple temperature driven methods to multi-layer process-based methods. Very few studies have compared evapotranspiration at forest ecosystems or FLUXNET sites not only because of a general focus on agriculture, but also because of the difficulty of obtaining evapotranspiration measurements for such an ecosystem. I have completed two papers on this objective, one of which I am primary author on and has been submitted to Environmental Modeling and Software, and the other of which I am secondary author on and has been submitted to Journal of Hydrology. The first paper scrutinizes the models in-depth, examines the mathematics and environmental variables, and includes uncertainty analyses; data used are from two growing seasons at the Blodgett Forest FLUXNET site. The second paper extends the analysis to multiple FLUXNET sites and focuses on site differences and similarities governing the model outcomes.
Remote Sensing of Natural Resources
Scaling from ecosystem measurements and evapotranspiration models to remote sensing relies on a combination of image processing, analysis and modeling. To understand remote sensing data acquisition, analysis, information extraction, and integration into geographic information systems (GIS), I studied Advanced Remote Sensing of Natural Resources under Professor Peng Gong. Relevant topics included: image calibration (radiometric, geometric, topographic), image enhancement (filtering, data reduction and transformation, texture measures), interpretation, classification, accuracy assessment, linear feature extraction, change detection, and multiple data source integration. In line with my NASA Earth System Science Fellowship, I focused on how spectral values translate into biophysical variables such as evapotranspiration; and, if effective models are developed, how we can validate those estimates since we cannot measure evapotranspiration directly on such large scales. Although many models exist that estimate evapotranspiration using remote sensing, the majority of those require ground measurements or weather station inputs. These inputs are infeasible in relatively inaccessible areas of the globe.
I reviewed two models that estimate evapotranspiration solely from remote sensing digital numbers: the “Simplified Method” (Jackson et al., 1977) and the NDVI-DSTV Triangle Method (Chen et al., 2002). To validate these models, I applied them to the Blodgett Forest FLUXNET study site using EO-1 imagery (Figure 1). The “Simplified Method” is used to obtain the integrated daily evapotranspiration from surface radiant temperature over variable vegetation (Carlson and Buffum, 1989; Carlson et al., 1995; Lagouarde, 1991; Lagouarde and McAneney, 1992; Nieuenhuis et al., 1985; Sandholt and Anderson, 1993; Seguin and Itier, 1983). The “Simplified Method” was designed to estimate evapotranspiration from a very few easily obtainable measurements, these being the surface radiant temperature near the time of local maximum (about 1300 h local time), a corresponding air temperature, and the net radiation integrated over a 24-h period. A common form of the model is as follows: LE = RN – B(Ts – T50m)n (cm/day) where LE is evapotranspiration, RN is net radiation, Ts is surface temperature, T50m is the air temperature at 50-m (obtainable from some weather stations and remote sensing), and B and n are functions of NDVI. The NDVI-DSTV (diurnal surface temperature variation) Method is based on the hypothesis that a triangle shape would result from a plot of NDVI and the difference between the surface temperatures obtained at day and night. Presence of green vegetation is a major determinant of evapotranspiration from the land surface due to enhanced surface roughness increasing turbulent exchange of water vapor and plant roots extracting water from the soil more rapidly than the water can diffuse to the soil surface (Smith and Choudhury, 1991). A multiple-regression equation from the values reported by Chen et al. (2002) for seven major land-cover types in south Florida is: ET = 6 + 6.4NDVI – 0.22DSTV. Surface temperature for both models can be estimated from the thermal bands in remote sensing alone, as specified by Kerr et al. (1992): Ts = C x Tv + (1 – C)Tsoil; Tv = -2.4 + 3.6(Thermal 11mm) – 2.6(Thermal 12mm); Tsoil = 3.1 + 3.1(Thermal 11mm) – 2.1(Thermal 12mm).
Although the “Simplified Method” significantly overestimated evapotranspiration relative to the slight overestimation by the NDVI-DSTV Triangle Method, the “Simplified Method” is more physically based than is the NDVI-DSTV Triangle Method because the former is bounded by net radiation. The upper bound to potential evapotranspiration should be the net radiation (under relatively stable conditions). According to energy balance models, incoming net radiation is partitioned into latent heat, sensible heat, and heat absorbed by the ground—therefore, latent heat, as a fraction of net radiation, should not exceed net radiation. A major problem with both models is the lack of soil moisture information—this absence is notably evident in areas or times of drought-stress such as in California. In the “Simplified Method,” only the surface temperature might increase if there was no soil moisture, but the estimated evapotranspiration would still be heavily weighted towards the net radiation. If there is no soil moisture to evaporate or transpire, then the net radiation becomes completely unimportant no matter how high or low it is. The NDVI-DSTV Triangle Method begins with an initial evapotranspiration of 174 W/m2 to add to a value from the NDVI that would theoretically range from 0 to 185 W/m2 (maximum of 359 W/m2). To account for lower bound from the soil moisture-surface temperature relationship, the DSTV would have to be 56ºC to estimate a zero evapotranspiration flux. These values are unrealistic, but not impossible, and the relationship between surface temperature and soil moisture must be examined closer to assess the lower bound of these models. Soil moisture, which varies widely throughout a landscape, affects land evaporation and plant transpiration, two processes that link the fluxes of energy (sensible and latent heat), water and carbon between land and atmosphere. New research has shown that wavelengths in the microwave portion of the spectrum respond to the amount of water present in the soil. Recent developments in both science and associated technologies now make the exploitation of the microwave region for soil moisture mapping feasible.
Sociology of Natural Resources
In addition, I have worked on furthering the exposure and application of GIS and spatial statistics into the social sciences. I have submitted an abstract to the conference of American Public Health Association on using GIS and spatial statistics for Environmental Justice and Air Toxics. I have a manuscript in preparation that I plan to submit following the conference.
Plans for the coming year
This summer I plan on working extensively in the field to gather data, processing satellite imagery for both field sites, and coding and testing some terrestrial ecosystem models such as BIOME-BGC, which is designed to simulate hydrologic processes across multiple scales (Running and Hunt, 1993). Remote sensing data integrated within an ecological process model framework provides an efficient mechanism to evaluate scaling behavior, interpret patterns in coarse resolution data, and identify appropriate scales of operation for various processes (Kimball et al., 1999). BIOME-BGC has been used to compare estimates of hydrologic processes to observed data for different boreal forest stands (Kimball et al., 1997), used to simulate water balance and evapotranspiration over a historical 88-year record for a broadleaf forest (White et al., 1999), and coupled with remote sensing information to evaluate the sensitivity of boreal forest regional evapotranspiration (Kimball et al., 1999). In addition to BIOME-BGC, I will investigate the Boreal Ecosystems Productivity Simulator (BEPS), which is based on the same principles as BIOME-BGC, but BEPS is modified to better represent canopy radiation processes (Chen et al., 1996; Liu et al., 1997). BEPS outputs spatial fields of evapotranspiration, and it has been used to upscale tower measurements of NPP to the Boreal Ecosystem-Atmosphere Study (BOREAS) study region by means of remote sensing and modeling (Liu et al., 1999). Further, I will consider a simple model of energy exchange between the land surface and the atmospheric boundary layer, driven primarily by remote sensing data, that partitions surface flux into latent and sensible heat (Mecikalski et al., 1999). I plan on taking my Ph.D. qualifying exam in the Fall, and will continue to expand my knowledge breadth with classes in soils and statistics.
Support
I will work closely with my UC Berkeley graduate advisor, Dr. Greg Biging, whose research is focused on remote sensing of natural resources. I will be working in conjunction with the Center for the Assessment and Monitoring of Forest and Environmental Resources (CAMFER) at the University of California, Berkeley, led by Drs. Greg Biging and Peng Gong, who both specialize in remote sensing of natural resources; CAMFER is a NASA-supported Center for Excellence in Remote Sensing Applications. At the Blodgett Forest field site, I will work with Dr. Allen Goldstein and his research group; at the Tonzi Ranch field site, I will work with Dr. Dennis Baldocchi and his research group. Furthermore, I will be collaborating with the Numerical Terradynamic Simulation Group (NTSG) at the University of Montana, led by Dr. Steven Running.
References
Carlson, T.N. and Buffum, M.J., 1989. On estimating total daily evapotranspiration from remote surface temperature measurements. Remote Sensing of Environment, 29: 197-207.
Carlson, T.N., Capehart, W.J. and Gillies, R.R., 1995. A new look at the Simplified Method for remote sensing of daily evapotranspiration. Remote Sensing of Environment, 54: 161-167.
Chen, J.H., Kan, C.E., Tan, C.H. and Shih, S.F., 2002. Use of spectral information for wetland evapotranspiration assessment. Agricultural Water Management, 55: 239-248.
Chen, J.M., Liu, J. and Cihlar, J., 1996. Boreal ecosystems productivity simulator (BEPS) using remote sensing, meteorological and soil data, 1996 Annual Combined Meeting of the Ecological Society of America on Ecologists/Biologists as Problem Solvers. Bulletin of the Ecological Society of America, Providence, Rhode Island, USA.
Goldstein, A.H., Hultman, N.E., Fracheboud, J.M., Bauer, M.R., Panek, J.A., Xu, M., Qi, Y., Guenther, A.B. and Baugh, W., 2000. Effects of climate variability on the carbon dioxide, water, and sensible heat fluxes above a ponderosa pine plantation in the Sierra Nevada (CA). Agricultural and Forest Meteorology, 101: 113-129.
Jackson, R.D., Reginato, R.J. and Idso, S.B., 1977. Wheat canopy temperature: a practical tool for evaluating water requirements. Water Resources Research, 13: 651-656.
Kerr, Y.H., Lagouarde, J.P. and Imbernon, J., 1992. Accurate land surface temperature retrieval from AVHRR data with use of an improved split window. Remote Sensing of Environment, 41: 197-209.
Kimball, J.S., Running, S.W. and Saatchi, S.S., 1999. Sensitivity of boreal forest regional water flux and net primary production simulations to sub-grid-scale land cover complexity. Journal of Geophysical Research-Atmospheres, 104(D22): 27789-27801.
Kimball, J.S., White, M.A. and Running, S.W., 1997. BIOME-BGC simulations of stand hydrologic processes for BOREAS. Journal of Geophysical Research-Atmospheres, 102(D24): 29043-29051.
Lagouarde, J.P., 1991. Use of NOAA-AVHRR data combined with an agrometeorological model for evaporation mapping. International Journal of Remote Sensing: 1853-1864.
Lagouarde, J.P. and McAneney, K.J., 1992. Daily sensible heat flux estimation from a single measurement of surface temperature and maximum air temperature. Boundary-Layer Meteorology, 59: 341-362.
Liu, J., Chen, J.M., Cihlar, J. and Chen, W., 1999. Net primary productivity distribution in the BOREAS region from a process model using satellite and surface data. Journal of Geophysical Research-Atmospheres, 104(D22): 27735-27754.
Liu, J., Chen, J.M., Cihlar, J. and Park, W.M., 1997. A process-based boreal ecosystem productivity simulator using remote sensing inputs. Remote Sensing of Environment, 62(2): 158-175.
Mecikalski, J.R., Diak, G.R., Anderson, M.C. and Norman, J.M., 1999. Estimating fluxes on continental scales using remotely sensed data in an atmospheric-land exchange model. Journal of Applied Meteorology, 38(9): 1352-1369.
Monteith, J.L., 1965. Evaporation and the environment. Symposium of the Society of Exploratory Biology, 19: 205-234.
Nieuenhuis, G.J.A., Schmidt, E.A. and Tunnissen, H.A.M., 1985. Estimation of regional evapotranspiration of arable crops from thermal infrared images. Journal of Remote Sensing, 6(1319-1334).
Ostrom, E., 1991. Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press, Cambridge.
Priestley, C.H.B. and Taylor, R.J., 1972. On the assessment of surface heat flux and evaporation using large scale parameters. Monthly Weather Review, 100: 81-92.
Running, S.W. and Hunt, E.R., Jr., 1993. Generalization of a forest ecosystem process model for other biomes, BIOME-BGC, and an application for global-scale models. In: J.R. Ehrlinger and C. Field (Editors), Scaling Processes between Leaf and Landscape Levels. Academic Press, San Diego, CA, pp. 141-158.
Sandholt, I. and Anderson, H.S., 1993. Derivation of actual evapotranspiration in the Senegalese Sahel, using NOAA-AVHRR data during the 1987 growing season. Remote Sensing of Environment, 46: 164-172.
Seguin, B. and Itier, B., 1983. Using midday surface temperature to estimate daily evaporation from satellite thermal IR data. International Journal of Remote Sensing, 4: 371-383.
Smith, R.C.G. and Choudhury, B.J., 1991. Analysis of normalized difference and surface temperature observations over southeastern Australia. International Journal of Remote Sensing, 12(10): 2021-2044.
White, M.A., Running, S.W. and Thornton, P.E., 1999. The impact of growing-season length variability on carbon assimilation and evapotranspiration over 88 years in the eastern US deciduous forest. International Journal of Biometeorology, 42(3): 139-145.
Xu, M., Debiase, T.A., Qi, Y., Goldstein, A. and Liu, Z., 2001a. Ecosystem respiration in a young ponderosa pine plantation in the Sierra Nevada Mountains, California. Tree Physiology, 21: 309-318.
