B41A-0153
Global and Regional Surface Albedo Changes due to Land Use Transformation: an Anthropogenic Source for Climate Change
For the past decades, cropland area has been slowly increasing while forests and woodlands diminished, leading to consequent changes in land use resulting from human behavior. Besides, desertification directly affects millions of people around the world and not a single year goes by without new reports of ice melting. More than being an economic issue, land use transformation can prove to have altered the energy balance, and therefore the climate, through surface albedo changes over the past decades. Each land category has its own surface albedo, defined as its solar back scatter and being only a function of the radiation field incident on it and the properties of the land category itself. Using a global surface albedo model (Hummel and Reck, 1979), involving 49 different types of surfaces for each quarter of the year, January-March, April-June, July-September and October-December, surface albedo maps are computed from land usage maps for the 1970s and 1990s. Regional changes in the surface albedo can cause variation in the energy budget of the earth-atmosphere system, specifically in the tropospheric distribution of temperature, and therefore can be an anthropogenic source for climate change at a global scale. Many feedbacks and teleconnections can be found between surface albedo, cloud coverage and CO2 fluxes leading to a potentially unstable energy budget system. In order to fully comprehend climate change, a extensive review on that system and its foundations is expected to be released in 2006.
B41A-0154
A Southern Washington Chronosequence Study: The Impact of Interannual Climate Variability on Ecosystem Exchange of Carbon, Water, and Energy in a Newly Established and Old-Growth Coniferous Forest
The T.T. Munger Research Natural Area of southern Washington provides a unique opportunity to study carbon exchange between coniferous forests and the atmosphere in a region that experiences a significant amount of forest harvesting disturbance and interannual climate variability. Here we present initial biometeorological measurements of carbon and water exchange at a 10 year old Douglas-fir stand with the goal of gaining information on how regional climate change will affect the carbon and hydrological budgets of a newly established forest. The young forest is 1.25 km from the Wind River Canopy Crane Research Facility, an AMERIFLUX site that has been continuously measuring carbon, water, and energy fluxes at an old-growth forest since 1998. Though still in its infancy, data from this chronosequence study will be used to quantify how sensitive net ecosystem exchange (NEE) of carbon is to interannual climate variability at different aged stands of the Washington western Cascades. Because the young stand is in close proximity to the old-growth forest, the climates at both forests will be identical, though the microclimates will not. The response in NEE at the young stand during the seasonal drought may be very different from that at the old-growth forest due to dissimilar canopy understory composition, which will lead to site differences in soil moisture and soil temperature. How this affects respiration rates and photosynthetic rates at both stands is one of the questions that will be addressed by this study. As the chronosequence study progresses, we hope to show any sensitivities that a newly established forest has to climate variability and in conjuncture with data from the old-growth stand, give the global carbon community important information on the forest carbon sequestration potential of the Pacific Northwest.
B41A-0155
Comparative Effects of Burning on Ecosystem Fluxes and Productivity in Two Adjacent Tall Grass Prairie Pastures
Management practices can have large effects on ecosystem productivity. In grasslands and prairies, one of the most common management tools is fire. While judicious burning generally increases grassland productivity, there have been few controlled studies that quantify the effect of fire on the fluxes of carbon, water, and energy. We have initiated an experiment in two adjacent pastures with similar native prairie vegetation, grazing, and burning histories located in the USDA, Agricultural Research Service's Grazinglands Research Laboratory in central Oklahoma. In March, 2005, we burned one pasture while leaving the other intact as a control. Identical eddy covariance and automated soil respiration systems and sampling protocols were installed and are being used in each pasture. Measurements include: net ecosystem exchange (NEE), soil carbon dioxide flux, soil moisture and temperature, and plant biomass by C3 and C4 functional group. Early in the growing season, there was more above ground green biomass in the unburned, control pasture. This was accompanied by higher NEE and latent heat fluxes. Later in the season, however, significantly greater productivity was observed (biomass and NEE) in the burned pasture compared to the unburned control.
B41A-0156
Application and Validation of a MODIS-based Vegetation Transpiration Model Over the Southern Great Plains Using IHOP 2002 Data
The International H2O Project 2002 (IHOP 2002), which was conducted during May and June 2002, sought to better understand spatial and temporal variations in the water vapor field in the Southern Great Plains (SGP) of the United States. One fundamental influence on this water vapor field is transpiration. Data collected at six IHOP 2002 surface sites, which are representative of the crop and grassland environments typical of the SPG as a whole, were used to validate a remote sensing based vegetation transpiration model (VTM). The model uses several vegetation indices derived from remotely sensed data in conjunction with surface observation data, for example air temperature and incident solar radiation, to estimate transpiration as a function of gross primary production and water use efficiency. While VTM has been tested over forest environments, this is the first evaluation of the model over crop and grassland environments. Since these land use types represent a significant proportion earth's terrestrial surface including nearly 40 percent of the land cover of the contiguous United States, this research marks an important step toward modeling transpiration over a region that plays a critical role in numerous biogeochemical cycles on both regional and global scales. By comparing model output with observations, it was found that the VTM represented temporal trends reasonable. It was found that the modeled values for transpiration were consistently less than the total observed moisture flux. This is to be expected since the VTM model currently considers only transpiration neglecting the moisture stream due to evaporation. Research is ongoing to develop an evaporation component for the VTM model so that it is able to accurately describe all streams of moisture transfer to the atmosphere.
B41A-0157
Relationships among vegetation properties related to their interactions with atmosphere from the analysis of satellite derived data
Vegetation is an important element to understand the complex interrelationship between atmosphere and land surface. In this study, we analyze Normalized Difference Vegetation Index (NDVI), Leaf Area Index (LAI), and Vegetation Water Content (VegWC). These variables selected in this study are closely related to vegetation water status with surface temperature (Ts) as a key factor of the interaction between vegetation and atmosphere. These satellite derived data sets are from Moderate Resolution Imaging Spectroradiometer (MODIS) data for NDVI, LAI, and Ts, and Advanced Microwave Scanning Radiometer (AMSR) for VegWC. Three different regions, each of which is climatically unique, are selected in North America for this study: North American Monsoon System region (NAMS), South Great Plain (SGP), and Tifton, Georgia. The relationship between NDVI and Ts (known as TvX relationship) which has been studied in many previous works is identified and validated from a weather forecasting model (WRF). The sensitivity of the TvX relationship to land surface roughness is analyzed to quantify the TvX relationship, and the limits of the relationship between NDVI, LAI, and VegWC are presented. A new variable, Normalized Vegetation Water Content (NVegWC) from the ratio of VegWC and LAI, and NDVI show significant relationship especially in relatively arid regions such as NAMS region. In order to investigate the local variation of the relationships between the variables in the three study regions, the land cover classification map is analyzed, and it is identified that different environments related to water status influence the physiological properties of vegetation.
B41A-0158
Dry Season Rainfall Anomalies due to Deforestation in Northern Mesoamerica: Implications for Forest Sustainability
In the region stretching between Mexico and Panama, the proposed Mesoamerican Biological Corridor (MBC) is an ambitious effort to stem and turn back the erosion of biodiversity in one of the world's biologically richest regions by connecting large existing parks and reserves with new protected areas by means of an extensive network of biological corridors. The success of this effort will depend in part on the ability of the connecting corridors to provide adequate habitats permitting the sustainability of some populations and the migratory movements of others. Ideally these connecting corridors would contain the biological communities which were originally present. Currently, however, many of these connecting corridors do not contain their original forest, but are instead occupied by agricultural landscapes containing croplands, grasslands and degraded woodlands. The forest types in northern Mesoamerica generally are those that require dry season rainfall for their survival, and it is not clear whether current environmental and climatological conditions are sufficient to maintain existing forests and regenerate the pristine forests in the deforested patches. Hourly climatological rainfall rates have been averaged for the time period of 1961 to 1997 at 266 stations in Guatemala and adjacent areas. These climatological rainfall rates have been segregated for forested and deforested regions of each of the major Holdridge life zones. Dry season cloud frequency of occurrences derived from GOES satellite imagery then are. correlated with the March climalogical data in order to generate regression estimates of current local rainfall. Differences between estimated current rainfall and historical values define regions under increased dry season water stress. In general dry season rainfall in March is markedly lower in deforested areas than in forested areas of the same life zone for most of the Holdridge life zones. In some deforested areas within the Holdridge wet forest life zones, estimated March rainfall deficits are >25 mm. Dry season deforested habitats tend to have higher daytime temperatures, are less cloudy, have lower estimated soil moisture and lower values of Normalized Difference Vegetation Index (NDVI) than do forested habitats in the same life zone. The result is hotter and drier air over deforested regions, with lower values of cloud formation and precipitation. The data suggest that deforestation is locally intensifying the dry season and increasing the risk of fire, especially for the long corridor connecting regions. In addition, forest regeneration in some parts of the MBC may not result in second-growth forest that is characteristic of that life zone but rather in forest regeneration more typical of drier conditions. The extent to which this would influence the conservation utility of any given corridor depends upon the ecological requirements of the organisms concerned.
B41A-0159
Spatial Variability of Soil CO2 Concentration from Forested and Clear Cut Sites in Nova Scotia, Canada
Spatial variability of CO2 concentration within forest soils is poorly documented because most CO2 concentration or production research has been based on point measurements representing a site or experimental treatment. This ongoing study was undertaken to examine the differences in CO2 concentration within soil profiles from two paired forested and clear-cut sites in Nova Scotia, Canada by accounting for the spatial variability caused by local topography. For each site, data were collected from four depths (0, 5, 20 and 35 cm) and ten micro sites separated by approximately 10 m and representing three local topographic features (level, trough and hump). Data collected on approximately monthly time interval from each site and depth indicated that CO2 concentration was highly positively skewed with high frequency of zero concentration values (up to 60 % of data at 20 cm depth for one of the clear-cut sites). The frequent zero concentration values were attributed to water filled soil pore spaces caused by frequent high water table at these sites. No consistent pattern could be found in CO2 concentrations among the three local topographic features. Median CO2 concentration increased up to the 20 cm depth for all sites. Box and Whisker plots of seasonally aggregated data indicated distinct seasonal pattern with higher median CO2 concentration during summer (June, July and August) and lower median CO2 concentration during winter (December, January and February). However the shape of the CO2 concentration profile were unaffected by seasonal and site differences. There was no significant difference in median CO2 concentration between forested and clear-cut sites due partly to the high variability within a site and increased variability at lower depths. Further, the CO2 concentration dynamics in these soils appear to be controlled by frequent water saturation of the soils by a shallow water table.
B41A-0160
Application of MODIS for Monitoring Water Quality of a Large Oligotrophic Lake
Flathead Lake, located in northwest Montana, is one of the 300 largest natural freshwater lakes in the world, covering an area of 480 km$^{2}$ with a maximum depth of 113 m. The Lake is oligotrophic, yet experienced an increase in eutrophication from 1977 to 2001, and two lakewide blooms of macroalgae in 1984 and 1994 that represented anomalous declines in water quality likely due to increasing nutrient inputs from anthropogenic sources. Summer field surveys in 2004 and 2005 showed surface chlorophyll-a levels from 0.1 to 0.9 mg m$^{-3}$, Secchi depths of 1.5 to 17.0 m, and surface temperatures from 8.3 to 22.6 °C. Depth profiles from surface to lake bottom were also obtained using a fluorometer and transmissometer. We examined the potential utility of MODIS medium resolution (250m and 500m) data (bands 1-4) and 1km ocean bands (8-14) to monitor spatial and temporal fluctuations in lake productivity indicators including chlorophyll content and turbidity. Several alternative approaches for retrieving water quality parameters from the MODIS data were evaluated, including atmospherically corrected reflectance products, and single scattering corrected radiance data. The zone of peak chlorophyll content and turbidity is found to occur immediately above the thermocline at water depths from 15-20 m, but with statistically significant linkages to surface conditions. Initial results indicate that the single scattering corrected radiance data provide the best prediction of chlorophyll-a, Secchi depth, and turbidity of the first 5m depth (r$^{2}$ = 0.46 - 0.75), but these parameters often co-vary at specific times throughout the season, creating difficulties in applying a consistent algorithm. Two complete daily time series from May 1 to Sept 30, 2004 were created from the 500m reflectance product and the single scattering corrected data to assess the sensor's ability to track lake fluctuations in water quality indicators. Mean daily lake reflectance values from these time series are found to be sensitive to both atmospheric particulate deposition and river discharge inputs at weekly to monthly time scales. Preliminary results show the potential of MODIS for water quality monitoring, but also highlight the need for improved algorithms and products specific to large inland water bodies.
B41A-0161
Effects of Succession on Carbon and Water Fluxes from Sagebrush Steppe
Prescribed burning is a management tool applied to sagebrush rangelands in the western United States to reduce shrub cover, increase forage quality and improve wildlife habitat. The resulting mosaics of vegetation in different stages of recovery (succession) following fire, with patches ranging in size from ~10 to >1000 m$^{2}$, have unknown impacts on the carbon and water cycles. We quantified the impact of changing contributions of mountain big sagebrush and perennial grass fluxes in south-central Wyoming to ecosystem fluxes in response to environmental dynamics through two growing seasons. We used eddy covariance to evaluate the influence of different vegetation cover on the magnitude and variability of carbon dioxide and water vapor fluxes during growing seasons of 2004 and 2005. Carbon was taken up at rates of 1 to 3 g C m$^{-2}$ d$^{-1}$ in June, and the ecosystem became a C source by mid- to late-July. Net C uptake occurred again in September and October following late summer rains in 2004. Peak growing season rates of C uptake (6-8 μmol m$^{-2}$ s$^{-1}$) and evapotranspiration (5-7 μmol m$^{-2}$ s$^{-1}$) compare well with fluxes measured from pure sagebrush stands in a large (4 m diameter) ecosystem gas exchange chamber. The results of this research contribute to a larger project quantifying the effects of vegetation succession on carbon sequestration and water loss in sagebrush steppe.
B41A-0162
Isolating the Effect of Surface Roughness Changes on Climate
Recent work (Keith et al, 2004) on the climate impact of very large scale wind farms has demonstrated appreciable effects on surface temperature, including warming and cooling on the order of a degree Celsius. In order to clarify the physical mechanisms of this impact, we have run simplified simulations of wind farm climate impacts, running the NCAR Community Atmosphere Model version 3.0 in "Aquaplanet" mode, with specified surface roughness and fixed sea surface temperatures. A circular region of enhanced surface roughness is imposed on the model. Several experiments, varying the location, size, and roughness of this region are performed. Results exhibit a Rossby wave response to the surface roughness perturbation. Southerly wind anomalies are correlated with increased tropospheric temperatures, resulting in increased downwelling infrared radiation at the surface. However, they are also correlated with increased cloudiness and albedo, and thus with reduced solar heating. These competing effects of warm advection on infrared and solar heating at the surface result in a complicated response pattern as the amplitude of the response decays away from the roughened region. This pattern in turn sheds light on the complicated climate impacts of wind farms found by Keith et al (2004).
B41A-0163
Observations of orographic Cloud Base Heights from satellite and in-situ measurements at the Monteverde Cloud Mist Forest Reserve, Costa Rica
Tropical montane cloud mist forests are among the most biologically rich and diverse ecosystems, providing habitats for many of the world's endangered species. Survival of these habitats depends strongly on regular and frequent immersion in orographic clouds. At the Monteverde Cloud Mist Forest Reserve in Costa Rica, the bases of the clouds have shifted upslope, leading to anuran population crashes, an increase in the upper elevation of bird ranges on the Pacific slope, and longer dry season mist-free intervals. Satellite remote sensing techniques have been developed to determine the orographic cloud base heights; these are tested for the dry season month of March 2003 over the Monteverde cloud forests. The approach derives MODIS cloud top pressures and then converts them to cloud top heights using geopotential height profiles. The NCAR Land Use and Cloud Interaction Experiment (LUCIE), consisting of paired mobile radiosonde systems deployed in Costa Rica, provided the means for validating the retrievals. Results show that the four MODIS CO2 slicing channels do not provide sufficiently accurate cloud top height values, although some of the differences are due to a mismatch in the observational periods. In order to improve the results, two alternative approaches are examined. Simulated geopotential height profiles from the CSU Regional Atmospheric Modeling System (RAMS) initialized with soundings provided superior results. Another approach investigated the utility of multiple combinations of channels in the CO2 slicing technique using Atmospheric Infrared Sounder (AIRS) data for cloud height assignment. Using AIRS a more accurate determination of cloud top height is achieved. Cloud thicknesses are estimated using three different approaches: 1) constant liquid water content (CLWC); 2) an empirical relationship; and 3) an adiabatic model. The CLWC approach provided the most consistent results. Cloud base heights are computed from subtracting cloud thickness from cloud top height. Orographic cloud base heights derived from the combined MODIS/RAMS approach and AIRS were then compared with values observed at the study sites. Differences between the observed and remotely sensed values were on the order of 200-300m for MODIS/RAMS. The results suggest that it is possible to monitor global cloud mist forest cloud base heights using the combination of MODIS satellite imagery combined with AIRS and model simulations.
B41A-0164
The legacy of forest harvest and burning on ecosystem carbon storage in the northern midwest, USA
Over 90 % of the forested area in the upper Great Lakes region was harvested by the early 20$^{th}$ century. In many cases, harvests were followed by uncontrolled burns, similar to current patterns of disturbance in many developing countries. While afforestation in the northern midwest has resulted in increased regional carbon (C) storage, the rate of C storage by forests will depend on the severity of prior disturbance and consequent changes in site quality. We were interested in how long the legacy of poor management practices from the early 20$^{th}$ century would be reflected in forest C storage rates. We investigated C cycling and storage following disturbance in mixed deciduous forests of northern lower Michigan, USA. Study plots ranged in age from 6 to 68 yrs and were created following experimental clear-cut harvesting and fire disturbance. Annual C storage was estimated biometrically from measurements of wood, leaf, fine root, and woody debris mass, mass losses to herbivory, soil carbon content, and soil respiration. Maximum annual carbon storage, or net ecosystem production (NEP), in the disturbed stands was 50 % lower than that of adjacent, undisturbed forest. This decrease was caused by a reduction in site quality following disturbance. However, during regrowth the cut and burned forest rapidly became a net C sink, storing 0.86 Mg C ha$^{-1}$ yr$^{-1}$ after six yrs. Carbon storage reached a peak of 1.00 Mg C ha$^{-1}$ yr$^{-1}$ after 50 yrs and declined to 0.57 Mg C ha$^{-1}$ yr$^{-1}$ after 68 yrs. Above- and below-ground net primary production (NPP) averaged 42 and 59 % of total NPP, respectively, with fine root litter production accounting for 57 % of total NPP. Soil heterotrophic respiration was high, ranging from 4.55 Mg C ha$^{-1}$ yr$^{-1}$ in the 6-yr-old stand to 5.74 Mg C ha$^{-1}$ yr$^{-1}$ in the 50-yr-old stand. Soil C and coarse woody debris pools exhibited a U-shaped trend over time following disturbance. Mineral soil and coarse woody debris pools lost C at a combined annual rate of 1.10 Mg C ha$^{-1}$ yr$^{-1}$ in the 6-yr-old stand, but these pools accrued C at a rate of 0.30 Mg C ha-$^{-1}$ yr$^{-1}$ in the 68-yr-old stand. Detritus inputs augmented soil C six years after harvest and this legacy C persisted in the oldest, 68-yr-old stand. This resulted in higher soil C than in an adjacent undisturbed mature forest. These results demonstrate that lasting decreases in site quality following disturbance result in long-term reductions in forest C storage.
B41A-0165
New Developments in the Remote Estimation of the Fraction of Absorbed Photosynthetically Active Radiation in Crops
The fraction of absorbed photosynthetically active radiation, fAPAR, is an important biophysical characteristic in models of gas exchange between the terrestrial boundary layer and the atmosphere, as well as in the analysis of vegetation productivity. Synoptic estimation of fAPAR has been performed by using NDVI as a linear proxy of fAPAR, despite the saturation of NDVI at fAPAR beyond 0.7. This paper analyzes the NDVI/fAPAR relationship in row crops (i.e. maize and soybean), and evaluates alternative vegetation indices to overcome the loss of sensitivity of NDVI at moderate-to-high vegetation biomass. Red-edge NDVI, which uses NIR and a band around 700 nm, Green NDVI and the recently proposed Wide Dynamic Range Vegetation Index (WDRVI), which uses red and NIR bands only, were found to be sensitive to fAPAR variation along its entire range and exhibited significant increase in sensitivity to fAPAR. The choice of the alternative index depends on the spectral characteristics of the sensor system at hand. Red-edge NDVI appears to be the best index for such estimation. It can be used in satellite systems with spectral bands in the red edge region (e.g., ESA's MERIS, NASA's Hyperion). The WDRVI exhibited high sensitivity to fAPAR at its entire range of variation. It can be employed to estimate fAPAR using such sensors as MODIS, Landsat and AVHRR, among others. The implications of these findings are far-reaching. Diverse regional to global studies requiring synoptic data on terrestrial vegetation will benefit from the increased accuracy of fAPAR estimation through the WDRVI, the Red edge NDVI, and the Green NDVI.
B41A-0166
An Improved Spatial Data Set of Tropical Deforestation Rates for the 1980s and 1990s
Tropical land cover dynamics in the 1980s and 1990s are highly uncertain, with enormous implications for balancing the global carbon budget and understanding the impacts on ecosystem goods and services. Recent estimates of tropical deforestation during the 1980s and 1990s vary by +/-40 percent due in part to differences in domain, forest baselines, methods, and definitions. The 8km Advanced Very High Resolution Radiometer (AVHRR) satellite record provides the only spatially-explicit data with comprehensive global coverage for both the 1980s and 1990s. However, sensor calibration and degradation issues combined with the coarse spatial resolution of AVHRR data may mask more diffuse deforestation events and likely capture only net changes in forest cover, thereby underestimating both gross deforestation and forest regrowth. Higher resolution Landsat data can capture gross changes in forest cover, but the processed data products are currently limited to particular regions or sampling schemes and "wall-to-wall" coverage is not available for the total tropics during the 1980s and 1990s. We used 200+ classified Landsat scenes from the TREES project and the FAO's Forest Resources Assessment to develop a spatially-explicit regression tree model based largely on the AVHRR record. Inputs to the regression tree included demographic, biophysical, and land-Use predictor variables such as population, fires, soils, elevation, and distance from roads, rivers, and urban centers. We used the regression model to create an improved spatially-explicit estimate of tropical deforestation rates and locations that incorporates the strengths of key regional to global-scale data sets.
B41A-0167
Silica Deposition in Two Large, Natural Riverine Lakes Sheds Light on Global Silica Cycle
Two large, natural riverine lakes in the upper Mississippi River basin offer a unique perspective on how human activities have affected the global silica cycle. Conley et al. (1993, Mar. Ecol. Prog. Stud., 101:1-2) and others have demonstrated that the eutrophication of artificial impoundments on large rivers around the world has significantly increased the sequestration of biogenic silica (as diatom frustules, etc.) in impoundment sediments, thereby decreasing silica supply to the oceans. One difficulty has been in establishing the ''natural'' (pre-dam) silica concentrations in large rivers when water quality data does not extend beyond a few decades. A mass balance approach was used to characterize the modern silica budgets of Lake Pepin and Lake St. Croix, large lakes in the Mississippi and St. Croix Rivers, respectively, in Minnesota and Wisconsin. These lakes formed as glacial meltwaters subsided around 9000 years B.P., so they provide the relevant qualities of impoundments (continuous sediment deposition, lake-like conditions for primary productivity) without the inherent disruption caused by dam building. Silica concentrations were measured in the lake inflows, outflows, lake water columns and surface sediment porewaters during 2004-2005, and the in-lake diatom productivity was calculated. In addition, these two lakes provide a valuable long-term record; our understanding of the lakes' modern silica cycles can help constrain historical values inferred, in part, from biogenic silica in the lake sediments. The timeline of twentieth century eutrophication of each lake has been established in previous paleolimnological studies, providing further perspective on the interconnection of silica and phosphorus cycles and land-Use change.
B41A-0168
Impacts of land use changes on heavy precipitation over the Indian monsoonal regions
In the middle and southern regions of the Indian sub-continent, 80% of the yearly precipitation is caused by the Asian monsoon, an intensely rainy season occurs from June to September every year. Weather models, during the monsoon season, have varying degrees of success at predicting the locations and amount of rainfall to be expected. In late July 2005, a record-breaking precipitation event occurred in Mumbai, western India, but the weather forecasting models failed to identify the heavy rainfall event, and the flooding did major damage. Anomalies in precipitation in the monsoonal regions could be the consequence of anthropogenic activities, such as urbanization and the changes in forested to agricultural crops and associated land use land cover changes. We report results on the impact of landuse land cover changes on the heavy precipitation episode simulations in the western Indian monsoon region using a mesocale model system. The primary area of interest is focused on domains based on the Arabian Sea Monsoon Experiment (2002) and then the validated model is applied to the Mumbai heavy precipitation case.
B41A-0169
Quantifying Forest Cover Change in Paraguay Using Data From Different Landsat Instruments
During the last 30 years, Paraguay underwent massive changes in the forest cover. Changes include the removal of the majority of the interior Atlantic Forest, an ecosystem with the highest levels of recorded biodiversity. This study quantified three decades of forest cover (FC) change in Paraguay using Landsat data. To assess FC change between the 1990s and 2000s, wall-to-wall mapping was performed, using data from Landsat TM and ETM+ instruments. Multispectral and multitemporal data were integrated into an unsupervised clustering process. Procedures were developed and implemented automatically to assign meaningful class labels to individual spectral clusters, substantially reducing human input. The accuracy assessment on the thematic map of FC in the 2000s was performed by examining aerial photos and the fine-resolution imagery. Overall accuracy on sampled populations was 0.9512 on average at a 95% confidence level. It indicates that the method used in this study is highly reliable in dealing both with data acquired at different times and with data from different Landsat instruments. Because of the differences in spatial and spectral-resolution between Landsat MSS and TM data, FC change between the 1970s and 1990s was estimated by assessing FC from each epoch using a different approach. FC for the 1970s was obtained by applying a statistical sampling method to the MSS imagery. Ecoregion-based stratified systematic sampling was utilized to select samples in every spatial interval of 0.15 degree in both the latitude and longitude directions. This method yielded 342 samples within the Atlantic Forest region. Each of 9 MSS pixels in individual samples was interpreted as forest or non-forest, which allowed the calculation of the percentage of FC within a given sample. When the method was applied to TM imagery for 1990s, the two methods, mapping and sampling, yielded almost identical results for the 1990 epoch. FC change between the 1970s and 1990s was calculated by differencing FC between the two epochs. The results show that the extent and spatial patterns of FC change in two main ecoregions, the Chaco and Atlantic Forest are considerably different. Between the 1990s and 2000s, 6.42% of the initial 134,400 km$^2$ of the Chaco was lost, while the initial 34,800 km$^2$ of the Atlantic forest was reduced by 38.9%. Within the Atlantic Forest region, the total area of 85,500 km$^2$ was forested by 73.4% in the 1970s. The forest area was reduced by 40.7% during the 1990s. By the 2000s, only 24.8% of the total region was forested. Although most protected areas experienced little change within their borders, buffer zones were deforested at an alarming rate. For instance, a well-managed reserve, Mbaracayu National park suffered from 40.0% of the initial forest loss from 400 km$^2$ to 240 km$^2$ in a 5 km buffer zone between the 1990s and 2000s.
B41A-0170
Detecting Deforestation In Paraguay From Multi-temporal Landsat Imagery Using A Spatio-temporally Explicit Algorithm
Forests in Paraguay have undergone extensive loss in the last decades. Detecting deforestation in this area with the use of satellite remote sensing data has particular scientific interests in a broad range of research fields. Conventional methods addressing this issue in terms of change analysis of difference image or post-classification comparison are incapable of modeling both spatial and temporal contextual information. In this paper, we propose a spatio-temporally explicit algorithm using multi-temporal Landsat imagery to detect the deforestation in Paraguay during the period between 1990 and 2000. In this algorithm, change analysis of difference image and classification of multi-temporal images are combined in a spatio-temporal model. Specifically, this algorithm includes the following three steps. First, a machine learning algorithm, Support Vector Machines (SVM), is trained with spectral observations to initialize the classification and to estimate pixel-wise class conditional probabilities for each individual image. Second, a modified Markov Random Fields (MRF) model accounting for pixel-wise transition probability is used to model the spatio-temporal contextual prior probabilities of images. Finally, an iterative algorithm, Iterative Conditional Mode (ICM), is used to update the classification based on the combination of spectral class conditional probability and spatio-temporal contextual prior probability. The results showed that the proposed algorithm achieved significant improvements over traditional pixel-based single-date approaches. The improvement from the contributions of spatio-temporal contextual evidence indicated the importance of spatio-temporal modeling in multi-temporal remote sensing in general and deforestation in particular.
B41A-0171
he Influence of Pre-settlement and Current High Plains Land Use and Land Cover on Atmospheric, Soils, and Vegetation Properties
Changes in land use and land cover can alter local and regional weather, hydrology, and ecosystem function. In the High Plains Region of the central and western United States (the area immediately east of the Rocky Mountains from South Dakota to Texas), human landscape modification from native grasslands to intensively managed croplands is especially striking. Before this shift was initiated in the mid-19th century, relatively continuous short-, mixed-, and tallgrass prairies dominated the region. In contrast, the current landscape is a mosaic of croplands (irrigated and non-irrigated), grasslands, reservoirs, and urban areas. Associated with the observed land-cover conversion over the last 150 years, land-atmosphere interactions, water cycling, and ecosystem functions have also changed. Our objective is to quantify the role of land-Use and land-cover change in the High Plains in modifying precipitation, near-surface air temperature, soil moisture and temperature, and plant growth. To do this, pre-settlement land cover was derived from Küchler's (1964) map of potential native vegetation and present land cover was adapted from National Land Cover Data (NLCD, Vogelmann et al. 2001) and agricultural statistics. Further, the land-cover datasets were enriched with C4 vegetation fractions from Tieszen et al. (1997). The contrasting land-cover data were used to define the lower boundary conditions in a coupled plant, land-surface, and atmospheric model (GEMRAMS) running on a 10 km horizontal grid increment. To isolate the spatial and temporal effects of the altered land surface on atmospheric properties, precipitation, soil moisture and temperature, and plant growth, paired GEMRAMS simulations were performed using identical large-scale meteorological and soil conditions. In addition, dry, intermediate, and wet spring and summer meteorological conditions were used to examine sensitivity to precipitation variation among the simulations.
B41A-0172
Mapping Human-Dominated Landscapes: the Distribution and Yield of Major Crops of the World
Croplands cover 18 million km2, an area the size of South America, and provide ecosystem goods and services essential to human well-being. Most global land-cover classifications group the diversity of croplands into a single or very few categories, thereby excluding critical information to answer key questions ranging from biodiversity conservation to food security to biogeochemical cycling. Information on land-Use practices is even more limited. The relative lack of information about agricultural landscapes results partly from difficulties in using satellite data to identify individual crop types and land-Use practices at a global scale. We address limitations common to remote-sensing classifications by distributing national, state, and county level statistics across a recently updated global dataset of cropland cover at 5 minute resolution. The resulting datasets depict the fractional harvested area and yield of twenty distinct crop types: maize, wheat, rice, sorghum, millet, barley, oats, soybeans, sunflower, rapeseed/canola, pulses, groundnuts/peanuts, oil palm, cassava, potatoes, sugar cane, sugar beets, tobacco, coffee, and cotton. These datasets represent the state of agriculture circa the year 2000 and will be made available for applications in ecological analysis, modeling, visualization, and education.
B41A-0173
Low-Dimensional Model for Estimating the Effects of Anthropogenic Heat Sources on Air Temperature
The effects of anthropogenic source on air temperature within urban environments continues to be a fundamental and practical problem in air pollution meteorology and energy research. The approaches employed in the Literature are empirical and utilize statistical comparisons between air temperature series collected within urban environments and outside, or high-resolution regional modeling activities that require significant computational resources and the determination of a large number of parameters. Here, we explore a simplified low-dimensional model that retains the essential physics of surface-atmosphere exchange, boundary layer dynamics, and radiative transfer. It has the decisive advantage of transferability and usability with minimal parameter input or computational resources. Also, because of its simplicity, this low-dimensional model can deconvolve the anthropogenic effects from surface characteristics on mean air temperature traces. The model is applied to the metropolitan area of Milano, Italy. It is able to reproduce well the ensemble of hourly air temperature traces inside the city and outside, and for a wide range of anthropogenic heat source and environmental conditions.
B41A-0174
Urban Sprawl Impacts on the Carbon and Water Cycles in the United States
As recurrent construction booms keep expanding the urban suburban areas in the continental United States with the typical low-density pattern often referred to as sprawl, so is the amount of urban vegetation. Irrigation and fertilization treatments common for most of the lawns and trees of the American urban and suburban landscape result in urban areas maintaining a significant role in the terrestrial carbon cycle, but at a high cost in resources, especially water. Here we analyze almost a decade (1992/93-2000) of urban land development in the conterminous US inferred from the nighttime citylights data from DMSP/OLS. We use EOS NDVI data and the NASA-CASA model to provide a spatially explicit assessment of the contribution of urban areas to the continental net primary productivity and compare it to pre-Urban and potential conditions. We also present an assessment of the amount of water resources required to maintain current levels of urban net primary productivity.
B41A-0175
Assessment of Septic System Performance Using Remote Sensing Technology
Failing and improperly managed septic systems can affect water quality and cause health problems for individuals, community residents, and wildlife. Early detection of septic system leakage and failure can limit the extent off-site contamination. State and county health agencies are typically responsible for permitting and regulating septic systems, and they rely on onsite inspection to identify malfunctioning systems. External symptoms which occur over an improperly functioning septic system can include lush or greener growth of vegetation, distress of vegetation, excessive soil moisture levels, or pooling of surface effluent. The use of remote sensing technologies coupled with attainable permit records to identify these features will enable the appropriate agencies to target problem areas without extensive field inspection. High-resolution thermal and color-infrared imagery were acquired in May 2005 for a study area in Jackson County, Mississippi, adjacent to the Gulf of Mexico. Within this coastal neighborhood known to have significant septic system failures, volunteers supplied information regarding the function of their systems by completing a survey and allowing access to their property. For each of 36 data locations, a septic system score was calculated to indicate the level of system performance. Potential predictors of system performance were derived from data obtained from installation records and data extracted from imagery. Linear regression analyses of the dataset identified the significant predictors of septic system performance, and two models have been developed and proposed for the prioritization of problem septic systems by regulatory agencies. The Drain Field Model was developed using linear regression. Vegetative Index and Normalized Differential Vegetative Index were identified as the best predictors of system performance. The model considers the maximum values of the VI and NDVI within each drain field and calculates a score for each system. The score is then, based on a threshold value, converted to 1 (suggests that system should be investigated) or 0. The efficiency of the model is 86% within the systems used in this study. The Hot Spot Model requires less input data from the user. This model highlights 'suspect' areas that should be investigated (hot spots) by recognizing clusters of pixels within a range of values for a particular vegetative index. Tree clusters, etc. may also be identified as suspect areas and must be recognized and rejected by the user. This model did not identify all of the problem areas within the study area; however, 92% of the areas identified as suspect were, in fact, observed problem areas.
B41A-0176
The Hydropatterns of the Florida Everglades: Mapping Flooding Using MODIS
Determining the spatial distribution and timing of flooding in the world's major wetlands would be highly valuable for global methane models, water management, and biodiversity assessments. The Florida Everglades complex is one of the largest wetlands in the US, and is subject to substantial development and water scarcity pressures that require intensive hydrological modeling and monitoring. MODIS is a high-quality daily-repeat sensor with underutilized potential for mapping wetlands at moderate resolutions. We calibrated our model using water stage data from the South Florida Water Management District for the calendar year 2004, and found that hydropatterns in the Florida Everglades are strongly correlated to a Tasseled Cap wetness index of MODIS Nadir BRDF-Adjusted Reflectance data. Although we tested several indexes, including NDWI and a land-surface temperature index, the Tasseled Cap Wetness index showed the closest correlation to the annual hydrographs across a range of surface vegetation types. Other factors incorporated into our model included elevation, fire events, and continuous tree cover percent within a pixel. Our results suggest that MODIS may be useful for dynamic monitoring of flooding, particularly in areas with sparse tree cover. Future work will focus on exploring the utility of this approach in other types of wetlands around the world.
B41A-0177
The Influence of Thinning on Components of Stand Water Balance in a Ponderosa Pine Forest Stand During and After Extreme Drought
To understand the effect of restoration thinning of ponderosa pine forests of the southwestern U.S, we compared the components of water balance between an unthinned plot and a thinned plot using a paired water balance approach. Forest overstory transpiration was estimated from tree sapflow scaled to the plot level. Understory herbaceous evapotranspiration was estimated from throughfall precipitation and changes in soil water content measured in trenched plots that excluded tree roots. The thinning treatment in 2001 reduced plot basal area by 82% and leaf area index by 41%. Stand-level evapotranspiration (ET), which includes overstory transpiraion and understory evaporation and transpiratio, was 72% higher in the thinned than the unthinned plot in 2002 when soil moisture was low due to drought. In spring 2003, when soil moisture was high due to recharge from late fall and winter precipitation throughfall (PT), ET of the thinned plot was 56% lower than the unthinned plot. Thinning also influenced the partitioning of ET between understory herbaceous evapotranspiration (ETU) and overstory transpiration (TO). During periods of low soil moisture (July 2002), ETU accounted for 92% of stand-level ET in the thinned plot compared to 74% in the unthinned plot. When soil moisture was higher (August 2002), ETU accounted for 75% of stand-level ET in both the thinned and unthinned plots due to increased TO. The following spring under wetter conditions (May 2003) ETU accounted for 48% of stand-level ET in the thinned plot compared to 28% in the unthinned plot. Our results highlight the importance of moisture availability and climate as factors determining the impact of thinning on water balance in southwestern ponderosa pine forests.
B41A-0178 INVITED
Remote Sensing of Soil Surface Texture, Carbon and Water Contents using Bare Soil Imagery
Knowledge of spatial soil diversity and landscape dynamics is fundamental to understanding of global biogeochemical cycles. Remote sensing data are increasingly being used for large-scale quantification of land-based measurements such as soil texture, carbon and water content. These regional estimates of surface soil properties through remote sensing can be used as input for global biogeochemical models. The objective of this study was to explore the relationship between bare soil reflectance and surface soil texture (sand, silt, and clay), organic matter, and soil moisture. High spatial (2 m) and spectral resolution (414-920 nm) hyperspectral /multispectral aerial imageries were collected over the Mississippi Delta and Mississippi Blackland Prairie Regions. Major soils included Commerce (fine-silty, mixed, superactive, nonacid, thermic Fluvaquentic Endoaquepts), Robinsonville (coarse-loamy, mixed, superactive, nonacid, thermic Typic Udifluvents), and Convent (coarse-silty, mixed, superactive, nonacid, thermic Fluvaquentic Endoaquepts) and Brooksville (Fine, smectitic, thermic Aquic Hapluderts). Over three hundred surface soil samples were collected within the study area and analyzed for particle size analysis, organic matter, moisture and hydraulic properties. ArcView GIS was used to generate sampling locations which included random, transect, and target soil sampling. Each soil sample represented a composite of six sub-samples collected within a two meter square area. These sample sites were selected to represent the range of aspect, slope, elevation, and parent materials within the site. To reduce the dimensionality of the hyperspectral data set, PCA analysis was applied. The selected bands were used in generating the statistical relationships between spectral reflectance and surface soil properties data. Stepwise (backward & forward) and partial least square statistical methods were used to generate surface maps of soil texture, organic matter, and surface soil moisture. The multivariate analysis including partial least squares and stepwise linear regression reveal that the near infrared band (NIR950 nm) was the best predictor of percent clay (R2 = 0.683) and silt (R2 = 0.634), while the combination of Red band (RED650 nm) and Green band (Green550 nm) were the best predictors of organic matter. Surface soil moisture dynamic was highly spatially correlated with soil texture maps. Once these relationships were established, ERDAS Imagine Spatial Module was used to generate surface maps for percent clay, percent silt and percent organic matter. These final products not only could be used for management purposes but also to quantify the spatial patterns and temporal dynamics of soils and their impact on climate change.