Introduction

The EOS Workshop on Evaporation and Transpira- tion (ET Workshop) was held June 1-2 at Goddard Space Flight Center (GSFC). The aim of the workshop was to develop a working relationship among the various EOS Interdisciplinary Science (IDS) Teams, Instrument Teams (IT), and Distributed Active Archive Centers (DAACs) interested in land surface evapotranspiration by encouraging complementary activities and a better understanding of each other's needs.

The specific goals the conveners set for this workshop were the following:

1. review the current status of the application of satellite data to the determination of land surface evapotranspiration and delineate the pressing problems, limitations, uncertainties, and dilemmas that must be overcome;

2. summarize algorithms and subsequent data products, on the part of both IDS and Instrument Teams, that are currently being planned as a part of pre-launch and post-launch activities;

3. develop coordinated plans between IDS Teams and Instrument Teams in pre-launch studies designed to test proposed algorithms and data product generation; and

4. establish relationships and means for cooperative research among IDS Teams, Instrument Teams, DAACs, and EOSDIS in evapotranspiration algorithm and product development.

We want to continue to provide an open forum and discussion environment that will allow all interested parties to address the specific goals listed above and to continue building a consensus on future directions and coordination among the EOS elements represented. This report and several information supplements will be posted on the Physical Climate and Hydrology Panel's homepage: http://www.hwr.arizona. edu/pchhome.html, to facilitate communication. Below, the salient resolutions and outcomes of the meeting are listed in the Executive Summary. The issues addressed by the group in discussion are described and then summarized in tables.

Executive Summary

There was general agreement that process-based parameterization of land surface evaporation and transpiration is possible and appropriate. The estimation of these fields using remotely sensed and ground data, however, needs a more thorough evaluation across a wide range of scales and climates. There was concern about the lack of adequate observations and the poor accuracies of a number of variables.

A significant outcome of this meeting was the identification of those parameters or data sets that will be of most interest to EOS scientists working with evapotransporation (ET). This list is important for several reasons: it clearly identifies standard products that are required to calculate ET; it identifies this group as advocates for these products; and it represents a consensus view and thus carries more weight.

The parameters detailed in Table 1 are characterized by a wide range of space/time resolutions. A major effort will be required to properly assimilate these diverse data sets in order to produce a standard ET product. This synthesis must encompass in situ, satellite, and model data. It can also be seen that EOS-generation data resolutions will not be significantly different from those possible today. The expectation, however, is for data quality and availability to far exceed current norms.

Table 1. ET Parameter Requirements/Most-likely Products

ET Parameter Requirements: The estimation of ET requires the following data: precipitation, net radiation, surface meteorology (air temperature, wind speed, vapor pressure deficit), vegetation (cover, biomass, vegetation index, fractional photosynthetically active radiation [fPAR]), soils (moisture, class), surface temperature and emissivity, runoff, streamlines, Digital Elevation Models (DEMs) and atmospheric profiles. Where possible, pre-EOS data sets should be identified and developed to evaluate the accuracy, sensitivity, and robustness of potential EOS ET products.

Model and Product Development: ET models and algorithms can be broadly classified as physical, empirical, and boundary-layer. Variants exist for both the energy- and water-balance approaches. Every effort should be made to develop and share model concepts and procedures prior to the launch of EOS AM to better identify the strengths and weaknesses of our varied approaches.

Calibration/Validation Sites: The IDS teams recognize that validation test sites and calibration sites are important. It is now more likely that our IDS teams will coordinate with the instrument teams in choosing joint sites to maximize the benefits from our calibration/validation efforts. The workshop endorsed having a number of different sites that spanned a variety of climates and land surface types across the globe. Further, it was recognized that international sites would be beneficial to the EOS program.

EOSDIS and Other Collaborations: The EOSDIS Data Handling System (EDHS) is making a strong effort to interface with science groups such as ours but many details of the system remain ambiguous. Data issues that remain poorly resolved in the face of regular program restructuring include: a fully operational (HDF) standard, generation of Level 3 & 4 data products, mechanisms to interchange experiment data sets among EOS teams and the utility of Level 1 data visualization/access efforts.

There are striking similarities and common objectives between this group and organizations such as the Global Energy and Water Cycle Experiment (GEWEX) and the International Geosphere-Biosphere Program (IGBP). We need to make greater efforts to develop collaborative research and stay abreast of each other's activities.

Issues and Discussions

Many issues were brought forward as part of background presentations and were then discussed in greater detail by all the attendees. We have tried to capture the gist of these discussion periods in the four sections below.

ET Parameter Requirements

• Scale: ET parameter requirements are driven by different objectives. These are related to the spatial and temporal scales of interest. It quickly became clear that there is a valid need to pursue model development and observations at a range of scales from local to global. There are IDS investigations across this range of scales, and this diversity should be encouraged.

• Precipitation: The overall importance of precipitation as an input and validation parameter was widely acknowledged. The usefulness of any global-scale precipitation estimates outside of the Tropical Rainfall Measuring Mission (TRMM) seems doubtful and was a concern to many of the IDS teams. Since direct observations from TRMM are infrequent, it will be important to have access to supplemental observations from GOES, gauge, and runoff data sources within EOSDIS. TRMM's operational product will be monthly precipitation totals at a 5 degree bin. Higher spatial and temporal products (1 deg, 5 day) are also being considered. Many participants seemed unaware of the current capabilities and limitations of these products and were encouraged to contact the TRMM program to evaluate pre-launch products in their models. The use of surface radar at regional scales, particularly NEXRAD data, should improve our knowledge of precipitation and hydrologic processes and should be an active link within the EOSDIS system.

• Radiation: Net radiation is another critical parameter. Assuming that estimates of solar radiation will never be practical at a resolution of 1 km/10 min, then resolutions of 50 km/60 min are desirable in that the space/time variability due to partly cloudy conditions is much reduced. The CERES radiation product at 1.25 deg/3 hr resolution is slightly coarser than desired but is acceptable. Hourly cloud cover from NOAAs GOES satellites is, and will continue to be, a high-priority data product.

• Soil Moisture: Soil moisture was repeatedly cited for its importance yet there seems to be some resignation that little can be done to improve our knowledge of its space-time distribution. A more-optimistic outlook is possible by considering that new 4D-data assimilation (4DDA) models and better land-surface parameterizations will greatly improve the representation of soil moisture in the coming generation of GCMs. Care must be exercised in recognizing the inherent model-derived limitations of these products. Any wide-spread, long-term surface monitoring programs should be encouraged for a variety of calibration and validation uses. This community should monitor and champion the need and development of sensors such as the Multifrequency Imaging Microwave Radiometer (MIMR), Advanced Microwave Scanning Radiometer (AMSR), and Global L-band Observations of the Earth (GLOE).

• Surface Meteorology: Near-surface winds, air temperature, and vapor-pressure deficit are required to estimate ET using robust physical models. This kind of information is not available from satellite, so non-EOS data sources must be identified and evaluated. Some useful work is being done to assess the utility of forecast-model-derived fields (from the National Center for Environmental Prediction [formerly NMC] or European Center for Medium Range Weather Forecasts [ECMWF], for example) for this purpose. Climatologically-derived fields, particularly with respect to wind and precipitation, are also of interest. None of these techniques provides data at better than 50 km/3 hr resolution at the global scale. Higher resolution data will be of use in areas of high relief and land-cover diversity.

• Surface Characteristics: Slowly-varying surface characteristics such as soil and vegetation parameters are required by many models as boundary conditions or are directly active in the partitioning of energy and mass at the surface. At global scales, the International Satellite Land Surface Climatology Project (ISLSCP) Initiative 1 soil data (texture, depth, slope) and vegetation data (NDVI, fPAR, land cover) represent a solid reference from which improvements can be made. At regional scales, improved data are desirable and may exist although more could be done to index available sources. MODIS is expected to provide improved estimates of fPAR and vegetation index, which need to be mapped to canopy conductance for estimating transpiration. The ability to quantify these parameters for different vegetation types and in different climates is critical and needs further testing. Soil temperature and emissivity are useful and recommended for model validation, data assimilation, and energy-balance calculations.

Calibration/Validation Sites

• Coordination: Many Instrument Teams (IT) have the need for and are planning to establish surface calibration/validation sites. There is much to be gained by co-locating some of these sites, and a clear willingness to increase this type of collaboration was expressed. In order to accommodate different levels of resolution, a nested-watershed approach is recommended.

• Potential Site List: Several potential IT/IDS test sites were identified across a wide range of environments. These are listed in Table 2. Important factors to consider include: site/model objectives, areal coverage, climate, vegetation, existing data base and data availability, existing and planned activities/commitments.

  Table 2. Potential Calibration/Validation Sites

• Exemplary Site: The roughly collocated DOE Cloud and Radiation Testbed (CART)/Atmospheric Radiation Measurement (ARM) site and GEWEX Continental-scale International Project (GCIP)-SW area is exemplary of the synergies we would like to have. Prominent and desirable features of this area are:
  • institutional commitment;
  • open data policy;
  • Baseline Surface Radiation Network (BSRN) NEXRAD precipitation coverage;
  • Oklahoma MESONET;
  • ARM extended facilities with soil moisture nested watersheds (ARS Little Washita, Walnut, Arkansas/Red); and
  • NWS/ARM Radio Acoustic Sounding System (RASS).

Model and Product Development

• Validity: The validity and ability to estimate regional-scale ET was questioned. This critical view is justifiable since even local-scale measurements are difficult and any estimate is by necessity indirect. Still, large-scale ET estimates are commonly being made and have shown real value in a range of hydrologic applications. One mitigating factor is that spatial and temporal averaging act to dampen much of the apparent contrast due to heterogeneity. Our goal should be a clear statement of the inherent parameter accuracy, sensitivity, and error structure as a function of model or observation and an effort to steadily improve these measures as models, instruments, and data assimilation techniques mature. We must guard against thinking we have the right answers. Rather, a logical approach that stresses space/time congruence is recommended‹products at large spatial scales must be validated against large-area average data sets.

• Accuracy: Calculations of ET are the most sensitive to precipitation, radiation, and soil moisture. Thus, the accuracy of ET is largely limited by our knowledge of these parameters. It was suggested that current ET accuracies are on the order of 10-20 W/m2 or 10-20% but there are (unknown) sensitivities to space/time scale.

• Instrument/Product Status: The Internet allows us all unparalleled access to up-to-date information concerning other components of the EOS project. (See Table 3.)

Table 3. Useful Internet URLs

EOSDIS and Other Collaborations

• EOSDIS: The Version 0 EOSDIS is operational, and a Web version is on its way. The DAACs at Marshall (hydrologic cycle), and Goddard (meteorology) are all producing and archiving data sets that are useful for ET estimation. However stronger involvement with the user community is required to ensure that the capabilities of the DAACs are fully exploited. The DAACs seek to better support the IDS teams through: 1) involvement of IDS and Instrument Team members on the DAAC User Working Groups, 2) supporting data set generation and compilation, e.g., ISLSCP Initiative I CD-ROM data set, 3) outreach activities conducted by the DAACs with the science community, and 4) user-driven functional and performance improvements to the Version 0 Information Management System, user search, order, and data manipulation capabilities.

• Related Efforts: Other groups such as GEWEX and the ARM program are conducting observation and analysis activities related to regional ET assessment and we must work to integrate better. This is particularly important over the Cloud and Radiation Testbed (CART)-ARM site, which is a common area of interest for us all.

• ISLSCP Initiative I: The results of the ISLSCP Initiative I GCM baseline data synthesis were presented. The experience gained in producing this and follow-on data sets is useful to better understand the relationships between mixtures of model, satellite, and in situ data.

[Table of Contents] [Previous] [Next]