USING TIME-SERIES AIRBORNE MULTISPECTRAL IMAGERY TO CHARACTERIZE GRASSLAND COVER AND LAND MANAGEMENT PRACTICES INFLUENCING SOIL CARBON STOCKS Matthew E. Ramspott, Ph.D. Candidate, Department of Geography Kevin P. Price, Professor, Ph.D. Department of Geography and Associate Director, Kansas Applied Remote Sensing (KARS) Program Bryan L. Foster, Associate Professor, Ph.D. Cheryl Murphy, Ph.D. Student Department of Ecology and Evolutionary Biology University of Kansas Lawrence, KS 66046 [email protected] (corresponding author) [email protected] [email protected] [email protected] ABSTRACT Land use practices greatly influence soil carbon stocks, which in turn influences the potential for soils to store carbon. With a growing interest in the use of a carbon credit system to decrease atmospheric carbon dioxide, there is increasing interest in development of cost effective methods for identifying land cover types and land use practices that maximize carbon storage potential. This study focuses on the use of remotely sensed data for characterizing Central Great Plains grassland land types and their management practices. During this study, near-biweekly multispectral submeter resolution imagery was collected throughout two growing seasons over 30 grassland study sites. These measurements were related to field data collected within five grassland management types including cool-season (C3) hayed, C3 grazed, warm-season (C4) hayed, C4 grazed, and USDA Conservation Reserve Program (CRP) lands. Field variables measured included: green Leaf Area Index and Fraction of Intercepted PAR, live and senescent plant cover, and topsoil carbon. Our findings show significant differences in topsoil carbon and bulk density levels among management categories, with grazed fields exhibiting generally higher levels of both carbon and bulk density compared to hayed fields. The airborne multitemporal datasets showed optimal data collection periods, which varied considerably from conventional wisdom on this matter. Our findings show remotely sensed measurements useful for discriminating among management practices influencing grassland soil carbon stocks. We also examined the relationship between our high-resolution airborne imagery and Landsat Thematic Mapper data and found them to be highly correlated, suggesting that our finding can be scaled up to the regional scale. INTRODUCTION The global atmospheric concentration of carbon dioxide (CO2) has increased markedly since the onset of the industrial revolution, as shown by data collected from long-term monitoring stations and by glacial ice cores. Indications are that this change in atmospheric chemistry may be leading to significant increases in average global temperature (e.g., Mann et al., 1999), with as-yet unknown, but potentially devastating consequences for global food production capacity, biodiversity, sea level changes, and other areas of social and ecological concern. One way to approach this daunting global problem is to more closely examine the potential for carbon storage in the terrestrial biosphere, an idea that has garnered much attention recently in the scientific discourse (see, e.g., Burke et al., 1997; Mermut et al., 2001). It can be argued that terrestrial sequestration of atmospheric C has the potential to temporarily mitigate anthropogenic CO2 increases in the atmosphere, giving the scientific community additional time to find Pecora 16 “Global Priorities in Land Remote Sensing” October 23 – 27, 2005 * Sioux Falls, South Dakotamore long-term solutions to this environmental issue. Soil carbon sequestration also has the potential of increasing organic matter in soils that have been depleted by past land management practices (Lal et al., 1998). Grassland systems have been identified as a tremendous storehouse of terrestrial carbon (Seastedt, 1995; Burke et al., 1997; Johnson and Matchett, 2001), in part because of the large component of biomass that is found below the ground in these systems. Much of the grassland soil carbon stock worldwide has been compromised by the disturbances associated with land use conversion, primarily to agriculture. The conversion of natural grassland systems to cultivated agriculture has led to a significant loss of the soil carbon resource through processes of volatilization, increased rates of microbial respiration, and complex problems associated with soil erosion (Lal et al., 1998; Ellert et al., 2001); each of these problems can also be precipitated by excessive grazing (Follett, 2001; Lal, 2001). BACKGROUND In general, the importance of land use and management on soil C stocks has been noted in many studies (e.g. Sperow et al., 2001; Lal et al., 1998). Evidence in the literature with respect to the impacts of grassland management on the soil carbon resource, however, is largely incomplete and in many cases inconsistent (Milchunas and Lauenroth, 1993). Many studies have suggested some potential for sequestration of C in soils of re-established grasslands (e.g., Gebhart et al., 1994; Potter et al., 1999) although the time period required for recovery to pre-cultivation conditions is not yet known and may be very long compared to the time period required for depletion (Burke et al., 1995). Any systematic study of how grassland management might impact the soil carbon resource could potentially make a significant contribution to this debate. Any such study, however, should not focus solely on manipulation of management practices to sequester additional soil C on lands that have been depleted, but also on the protection of lands that still have much of their soil C resource intact, such as native prairies. Part of the inconsistency associated with studies of the impacts of management on grassland soils stems from the shortage of workable direct methods for measuring the relevant biophysical parameters of the carbon-fixing plant/soil interface, especially when the goal is to objectively monitor geographically extensive study areas. Thus it becomes necessary to identify more readily measured parameters for examining the carbon–fixing properties of grasslands and their relationship to soils. Previous work by many scientists has demonstrated great promise for the measurement and characterization of the above-ground component of the plant/soil system in grasslands using visible/near infrared (VNIR) remote sensing methods (e.g., Weiser et al., 1986; Bartlett et al., 1990; Lauver and