Toward the Improved Use of Remote Sensing and Process Modeling in California's Premium Wine Industry Lee F. Johnson California State University NASA Ames Research Center, MS 242-4 Moffett Field, CA, USA 94035-1000 650-605-3331 (tel), -4680 (fax), Ljohnson@mail.arc.nasa.gov Ramakrishna R. Nemani Lars L. Pierce University of Montana California State University Missoula, MT, USA 59812 Seaside, CA, USA 93955 Matthew R. Bobo Daniel Bosch JCWS, Inc., NASA Ames Research Center Robert Mondavi Winery Moffett Field, CA, USA 94035 Oakville, CA, USA 94562 INTRODUCTION Winegrape quality is influenced by such factors as ratio of fruit to vine leaf area, amount of sunlight directly intercepted by grape clusters, and water stress levels. Vineyard canopy density (leaf area index, LAI) is thus a key variable of interest. California premium winegrowers are making increasing use of optical remote sensing as an additional tool for monitoring canopy density and managing vineyards [1]. In particular, high spatial resolution (2m) vegetation index imagery has been shown to be useful for subdividing individual fields ("blocks") for harvest based upon end-of-season vigor, as inferred by canopy density [2]. Block segmentation can result in more uniformly mature wine "lots" and, in some cases, ultimately in improved wine quality. In partnership with the wine and commercial remote sensing industries, NASA/Earth Science Enterprise investigators continue to examine relationships among vine stress, canopy development, and resulting wine quality by combining remote sensing with an agro-ecosystem process model adapted from BIOME-BGC [3]. The model, which predicts fluxes of water and carbon, uses remotely sensed LAI to modulate photosynthesis and transpiration across the landscape. The modeling framework potentially enables improved specification of irrigation and nutritional requirements for greater block uniformity. Landscape analysis represents a departure from much of the vineyard remote sensing application to-date, which has tended to emphasize relative canopy differences within a particular block. As an initial step in this direction, we seek to determine the robustness of remote sensing for retrieving LAI across different blocks (ie, regionally) in the face of such potential confusion factors as trellis system, sun/view angle, topography, grape variety, soil type/brightness, and atmosphere. METHODS The LAI-2000 Plant Canopy Analyzer (LI-COR Inc., Lincoln, NE, USA) was used to make indirect measurements of LAI (LAI i ) at 28 sites in California's Napa Valley (~38°26'N, 122°24'W), and at 22 sites in the nearby Carneros District (~38°14'N, 122°22'W). The sites were selected to represent the main trellis systems (split, vertical, sprawl) and variation in other factors mentioned above. At each site, a measurement of ambient light was made by holding the instrument at arm's length above the canopy. Two measurements were then made below canopy, one at row center and one at the midpoint between rows, then averaged and compared with ambient to derive LAI i . The measurement period was 2-3 September 1999. All observations were made under diffuse light conditions, either with the sun below the horizon or during fog. GPS readings were taken at each site. Direct LAI (LAI d ) measurements were made on 31-Aug-99 and 01-Sep-99 to calibrate the instrument readings. For this, three sites were chosen in the Napa Valley location, each on a different trellis system and with widely different row/vine spacing. LAI-2000 measurements were made under cloudcover at four nearby vines per site by the above method. All leaves were removed from each vine and weighed. A leaf subsample was then weighed and its area measured with an optical leaf area meter. Total vine leaf