Recent Accomplishments

through 2003-2004

David Briske

Title: Impact of altered precipitation distribution and warming on tree and grass life forms in Oak savanna

Results to Date:

Construction of the experimental infrastructure was completed in March 2004. The site consists of 100 2x2 m plots established in spring 2003 with defined species mixtures. The infrastructure includes eight permanent rainout shelters constructed from 9 x 18 m greenhouse frames covered with greenhouse poly (see photo). The ends and sides below 1.5 m are left open to maintain microclimatic conditions as near ambient as possible. Sheet metal barriers isolate each shelter from both surface and subsurface water flow. An overhead irrigation system (16 spray nozzles per shelter) simulates precipitation regimes and infrared lamps suspended over individual plots simulate global warming. This enables us to evaluate plant response to altered precipitation patterns and warming both independently and in combination. In addition, two similar blocks of plants were left uncovered to serve as a control of ambient conditions. Treatment effects on microclimate and soil moisture are monitored on site through a weather station and various sensors and data loggers installed in individual plots.

Climate simulations were initiated at the beginning of March 2004. Simulated precipitation regimes include two patterns that vary in season and rainfall event size, but not in total annual rainfall or total number of rainfall events (Fig. 1). The long-term precipitation pattern characteristic of the region is simulated beneath four of the shelters. The frequency and intensity of precipitation events are also simulated from the long-term precipitation record of the region. In the rainfall redistribution treatment beneath the other four shelters, 40% of the summer (May – September) precipitation is withheld from each event and evenly redistributed to the two preceding spring (March and April) and two subsequent autumn (October and November) months. The redistribution treatment effectively increases the intensity of the summer drought and the amount of rainfall that occurs during the cooler seasons of the year. At the end of the initial re-distribution period, a small (2%), but consistent, reduction in surface (12 cm) soil volumetric water content occurred between the two rainfall regimes. Large rainfall events had unexpectedly small effects on volumetric soil moisture content between rainfall regimes because a large proportion was lost as run-off. One-half of the experimental plots beneath each shelter are warmed with overhead infrared lamps. These lamps increase soil and plant temperatures by about 1o C (100 W m-2 from a height of 1.5 m).

Five species combinations are grown in 10 2x2 plots beneath each of the eight rainout shelters and two shelterless controls. The most abundant deciduous tree (Post oak; Quercus stellata), perennial grass (little bluestem; Schizachyrium scoparium), and an invasive evergreen tree (eastern red cedar; Juniperus virginiana), are grown in monoculture beneath each shelter. In addition, both of the woody plant species are grown in equal numbers with little bluestem to investigate grass-tree competition under these global change scenarios. We are evaluating physiological and growth responses over a range of scales, from leaf and root, to whole-plant and ecosystem. The response measures include shoot and root growth, photosynthesis and transpiration rates, and plant water stress. Soil respiration and mineralization are also monitored to determine the effect of global change scenarios on ecosystem processes.

Experimental data for the first spring redistribution period (March-April) indicate that both warming and precipitation distribution had independent effects on plant performance. All three growth forms showed increased growth in response to warming. However, species specific responses were observed in response to precipitation redistribution, in spite of the minimal effect on soil water content. The grass and juniper both showed a positive response to additional precipitation, but the oak showed a strong negative response in growth. Unexpectedly, plant species showed few interactions to warming and precipitation redistribution during the spring. Competition is occurring in the grass – tree mixtures, with the grass being the strongest competitor at this stage of plant development. The major hypothesis is that the plant species that most effectively track the redistribution of precipitation, with or without warming, will increase in abundance and whole-plant carbon gain under the altered climatic regime.

The project to date has contributed to training in global change science, including one undergraduate honors student, one summer intern, and one graduate student contributing to data collection. The students are funded from external grants. In addition, the Texas Agricultural Experiment Station supports this project by providing partial funding to two part-time technicians.

James Heilman

Title: Net Carbon Dioxide Exchange in Live Oak-Ashe Juniper Savanna and C4 Grassland Ecosystems on the Edwards Plateau, Texas: Effects of Seasonal and Interannual Changes in Climate and Phenology

RESULTS TO DATE:

Year 1 (1 Sept. 2003 – 31 Aug. 2004) was devoted mainly to design, construction and installation of eddy flux towers and sensors on the grassland, and in the woodland. Although this project began on 1 September 2003, funding was not received until December of 2003. Interim funding from the Texas A&M Research Foundation allowed us to bridge a portion of the funding gap and reduce delays in tower construction and installation. Table 1 summarizes physical and biological measurements at the two sites.

A 3-m tower (Fig. 1) was installed on the grassland site and measurements began in April of 2004. A 17-m tower (Figs. 2 and 3) was installed in the woodland with measurements beginning in June. The 17-m tower is equipped with a tram to lower the instrumentation package for maintenance and calibration. The suite of instrumentation at both sites includes Campbell Scientific CSAT-3 sonic anemometers and LI-COR LI-7500 open path gas analyzers (both operating at 10 Hz) for measurement of CO2, water vapor, sensible heat and momentum fluxes, LI-6262 closed-path gas analyzers for measurement of CO2 concentration profiles, REBS Q7.1 net radiometers, Vaisala HMP45C temperature and humidity probes, REBS HFT3 soil heat flux plates, LI-200SA and LI-190SA sensors for measuring global irradiance and PPFD, respectively, and Decagon Echo probes for measuring soil water content. Data logging is handled by CR23X and CR10X microloggers, and 10 Hz are downloaded to laptop computers using the ‘baler’ software of Campbell Sci. Power to the grassland site is provided by 12 V batteries charged by a solar panel array, and mainline power (installation paid for by Texas State University) is provided to the woodland site.

Although we have just begun the measurements, we have observed differences in CO2 and water vapor fluxes between the two sites. During the spring of 2004, rainfall was high, and Texas wintergrass (C3 species) was prevalent at the grassland site. NEE of the grassland during this period was higher than that of the woodland, due mainly to a lower ecosystem respiration, as indicated by nighttime NEE in Fig. 4. Water vapor flux, on the other hand, was lower in the grassland, suggesting that water use efficiency may be higher in the grassland when water is not limiting. Onset of higher temperatures and reduced rainfall in early summer caused the cool season Texas wintergrass to go dormant, but we have not yet analyzed eddy flux measurements obtained during this transition.

Soil respiration (surface efflux) measurements thus far have ranged from 1 ?mol m-2 s-1 when soils are dry to 8 ?mol m-2 s-1 when soils are wet, with no significant differences between the two ecosystems. Soil respiration measurements at the woodland site include contributions from both the soil and leaf litter of varying thickness and composition. At the woodland site, we have not observed any consistent differences in surface efflux related to litter thickness and composition.

PRODUCTS:
None at this time

Christopher Kucharik

Title: Evaluating Integrated Models of Natural and Managed Ecosystems over the Central and Southeastern US

RESULTS TO DATE

Kucharik, C.J., Evaluation of a process-based agro-ecosystem model (Agro-IBIS) across the U.S. cornbelt: simulations of the inter-annual variability in maize yield, Earth Interactions 7, 1-33, 2003.

Kucharik, C.J. and K.R. Brye, Integrated BIosphere Simulator (IBIS) yield and nitrate loss predictions for Wisconsin maize receiving varied amounts of Nitrogen fertilizer, Journal of Environmental Quality 32, 247-268, 2003.

Donner, S.D., C.J. Kucharik, and J.A. Foley, The impact of changing land use practices on nitrate export by the Mississippi River. Global Biogeochemical Cycles, 18, GB1028, doi:10.1029/2003GB002093, 2004.

Yiqi Luo

Title: Modeling Studies of Forest/Atmosphere Carbon Fluxes in a Colorado Subalpine Ecosystem

RESULTS TO DATE

In the past three years, NIGEC has financially supported our modeling research in association with the Niwot site in particular and other AmeriFlux sites in general. We have developed a process-based model to simulate C fluxes at the Niwot Ridge, a method of gas-filling missing data, a global database to link NEE with nitrogen deposition and soil N fertility, and a statistical approach to identification of causes of IAV in NEE, and a comparative study of canopy quantum yields among the Niwot, Duke, Harvard, and Oak Ridge flux sites to examine the regulation of NEE by canopy properties. Here is the brief description of the past research results.

Modeling study of C fluxes at the Niwot site: We have developed a process-based model TECOS, which was validated by flux data from the Niwot site. Our model generally predicts the time course of measured NEE using the eddy-covariance method (ECM) well in both 1999 and 2000. Predicted NEE is correlated with measured NEE with y = 0.95x, where y is the modeled NEE and x is the measured NEE with correlation coefficient r = 0.88 in 1999 and y = 1.02x with r = 0.82 in 2000. We estimated gross primary productivity (GPP) from measured daytime NEE plus ecosystem respiration, which was estimated from nighttime NEE with a Q10 value equaling 2.1. The estimated GPP from ECM is overall consistent with the modeled one. Modeled GPP generated larger variation partly due to representation of water stress in the model. In particular, the model overestimated GPP in comparison to the estimated one during August and September. During the summer, the Niwot eddy-flux site experiences dry air when wind blows from west and moist air when wind blows from east. Canopy photosynthesis is likely determined by both soil moisture and air humidity. In the current version of our model, water stress is mainly induced by soil moisture content. We also compared modeled ecosystem respiration with estimated one from ECM measurements. Modeled respiration fits with the estimated one well in the whole time course in both 1999 and 2000 except during the summer when water stress is strong. The mechanisms causing the deviation are not clear yet. In addition, we have partitioned the whole ecosystem respiration into both plant and microbial respiration. Our modeling study indicates that plant respiration is a dominant component in both 1999 and 2000. The model also captured the observed quick switch from C source to sink in the early spring well.

Gap-filling missing data using multiple imputation (MI) Missing data is a ubiquitous problem in evaluating long-term experimental measurements, such as those associated with the FluxNet project, due to the equipment failures, system maintenance, power-failure, and lightning strikes among other things. To estimate annual values of net ecosystem carbon exchange (NEE), latent heat flux (LE) and sensible heat flux (H), such gaps in the measured data must be filled or imputed. So far, no standardized method has been accepted and the imputation methods used are largely dependent on the researchers’ choice. We used multiple imputation (MI) to gap-fill the missing data for annual estimations of NEE, LE and H at three flux sites associated with the FluxNet effort. MI is a Monte Carlo technique in which the missing values are replaced by several simulated values. Each data set imputed is a complete one where the observed values are the same as those in the original data set; only the missing values are different. Thus, the normal statistical analysis (e.g., annual total calculation) can be applied to each data set separately. The results of each analysis can be recombined into one summary. We applied the MI method to eddy covariance measurements collected from Niwot site (a subalpine forest), Walker Branch Watershed (WBW) site (a deciduous forest), and Duke site (a coniferous forest). Results showed that annual estimations of NEE, LE and H by MI were comparable to other imputation methods but MI was much easier to apply because of readily available software and standard algorithms. Besides the normal statistical analyses, MI also provided confidence intervals for each estimated parameter. This confidence interval is most useful when assessing energy, water, and carbon balance closures at a given tower site. Significant differences in annual NEE, LE and H were found among years at the three AmeriFlux sites. NEE at the Niwot Ridge site was lower and LE and H were higher than at the other two sites. With the available softwares and realistic gap-filling capability, MI has the potential to become a standardized method to gap-fill eddy covariance flux data for annual estimations and to improve the analysis of uncertainties associated with annual estimations of NEE, LE and H from regional and global flux networks.

Nitrogen deposition and NEE. Atmospheric deposition of N to terrestrial ecosystems has increased dramatically in the past few decades, contributing up to 20 times inputs via natural processes to the northeastern United States and Europe. Chronic additions of N to these predominantly N-limited ecosystems have been shown to alter forest productivity, to modify species composition and diversity, and to affect losses of N from forest soil to stream water and to the atmosphere. However, it is still controversial whether or not N deposition will stimulate C sequestration in terrestrial ecosystems. To resolve this controversy, it is imperative to develop direct evidence that explicitly links C sequestration with N deposition. We developed a global database that encompasses NEE and N deposition at 35 eddy-flux sites in four continents. Our results indicate that C sequestration has no statistically significant relationship with N deposition in low productivity ecosystems and follows a parabolic function with a peak at 1.76 g m-2 yr-1 of additional N in high productivity ecosystems. That is, N deposition has positive effects on C sequestration in highly productive but low N dose regions and negative effects in highly productive and high N dose regions.

Development of statistical approaches to partitioning IAV in NEE into various sources. IAV in NEE is one of the most critical issues in ecosystem ecology. We hypothesized that IAV in NEE is primarily caused by climatic variability and its induced functional changes in ecosystem processes. We employed a homogeneity-of-slopes (HOS) model to identify the two causes for IAV in NEE and nighttime ecosystem respiration (RE). The model is essentially based on multiple regression analysis to relate NEE and RE with climatic variables and compute slopes of the regression lines using the data in individual years (bi) as well as the data from all the years (b). We applied the model to a dataset collected at the Niwot Ridge and Duke Forest AmeriFlux sites. We found that bi was statistically different from zero for both NEE and RE for at least one of the climatic variables, leading to a conclusion that the functional changes significantly contributed to IAV in C fluxes in the Duke Forest. With the detected functional changes, we simplified the HOS model to a separate-slopes model to partition IAV in NEE and RE to direct effects of climatic variability, indirect effects by functional changes, and seasonal climatic variation. Our results indicated that 16.4% of variation in estimated RE was explained by the functional change, 1.3 % by interannual climatic variability, and 69.9% by seasonal climatic variation. The three components explained 11.7, 2.8, and 73.8% of the variation in estimated NEE, respectively.

Regulations of net ecosystem exchange by canopy photosynthetic properties. Understanding the causes and control of seasonal and interannual variability in net ecosystem exchange of CO2 (NEE) in terrestrial ecosystems is one of the primary goals of both experimental and modeling studies. We used sixteen years of data from four AmeriFlux forest sites with different vegetation types and climatic conditions to evaluate (1) the relationships of NEE with canopy quantum yield (QY) and maximum ecosystem photosynthesis (Pmax), (2) variations of canopy QY and Pmax within and across the sites, and (3) their relationships with climatic variables. Canopy QY and Pmax were estimated by a nonrectangular hyperbolic equation based on absorbed photosynthetically active radiation (aPAR) for each daily data averaged over a week. Variation in annual NEE across sites was largely explained by the canopy QY and Pmax, which were the properties of these forest ecosystems. With small canopy QY and Pmax, less carbon was sequestrated at the Niwort Ridge site. In contrast, the Oak Ridge site had large canopy QY and Pmax, thus it sequestrated more carbon than the other sites. Interannual variation in NEE within a site might be caused by the variation in climatic variables. Seasonal variations in daytime NEE within each site were largely explained by the seasonal changes in canopy QY and Pmax. Canopy QY and Pmax were mainly influenced by aPAR and air temperature within a site, but other variables such as plant nitrogen content may have significant effects on canopy QY and Pmax. Interannual variability (i.e., coefficient of variance, CV) varied from 7.74 to 14.09 for canopy QY and from 4.15 to 11.95 for Pmax among the four sites. These results significantly improved our understanding on the observed variations in NEE and the parameterization of biogeochemical NEE models.

PRODUCTS

1. A comprehensive terrestrial ecosystem simulation model
2. Statistical approaches to analysis of interannual variability in NEE
3. A global NEE versus nitrogen database
4. Multiple imputation method for gap-filling missing data

Russell Monson

Title: Forest-Atmosphere Carbon Fluxes in a Colorado Subalpine Forest

RESULTS TO DATE

Our NIGEC funding for the first year of this grant cycle only arrived in late-November 2003. Thus, this report formally addresses seven months of research, at the time of this writing,. Much of our research effort during the past seven months has focused on organizing the GPS measurements of individual trees in plots east and west of the tower flux site, setting up and running an under-snow experiment measuring soil respiration during the winter of 2004, and making observations on the biochemical and physiological controls over photosynthetic recovery during the snow melt period.

Hypothesis #1 was inspired by analysis of the first five years of continuous NEE data from our site. During years when the late winter (February-April) snowpack was deepest, the rate of wintertime CO2 loss (as ecosystem respiration), was highest (Fig. 1). Once again, given that between 60-90% of the CO2 that is assimilated in any given growing season is lost by respiration during the subsequent winter, an understanding of the wintertime dynamics that control this variability is critical to fully understanding interannual variation in carbon sequestration at this site.

Hypothesis #2 was inspired by analysis of the first three years of continuous NEE data from our site. In order to study ecosystem-specific patterns in NEE, we took advantage of the location of the site within the ecotone between the lodgepole pine-dominated community to the east of the tower and the spruce/fir-dominated community to the west of the tower. The 30-min flux record for the first three years of measurement was "binned" depending on wind direction. Measured CO2 fluxes under light-saturating conditions tended to be higher for the pine-dominated community, despite slightly lower total LAI. The time-averaged ratio between CO2 and H2O fluxes (Fc /Fw) is an index of ecosystem WUE. When Fc /Fw was averaged over ten-day intervals, the pine-dominated community exhibited higher WUE through the first half of the growing season (May-July); after July, WUE was similar for both communities. The differences in CO2 flux rate and WUE between the communities could not be explained by climatic differences, or analysis of wind characteristics (e.g., co-spectral composition or u*).

To further study these differences in ecosystem water-use efficiency, we have recently obtained funding from NASA (a collaborative project between Professor Dave Bowling, University of Utah, Russ Monson, University of Colorado, and John Miller, NOAA) to install and maintain a Campbell Scientific tunable diode laser to continuously measure the 12C/13C ratio of CO2 in air when the wind comes from the east versus west. These data coupled with Keeling plot data collected from flask sampling in both the pine and spruce-fir communities (an effort currently maintained by Professor Jim Ehleringer, University of Utah) will be used to independently confirm the difference in water-use efficiency and to resolve whether these differences are due to respiratory characteristics or photosynthetic characteristics. Although the NASA-funded study of the stable-isotope dynamics of CO2 is independent of the NIGEC effort, it is nonetheless complimentary. One of the aims of the NASA study is to determine if CO2 derived from soil respiration east of the tower reflects a different Keeling-plot intercept than CO2 derived from soil respiration west of the tower, and if such differences can be attributed to difference in soil moisture dynamics in the eastern pine-dominated stand versus the western spruce-fir dominated stand.

In other studies conducted during the past year (conducted during the spring of 2004), we addressed Hypothesis #1 with regard to the controls that regulate the response of the forest to springtime snowmelt. We examined the capacity for needles that were in a state of wintertime downregulation of photosynthesis to regain photosynthetic competence after warming versus rehydration. Basically, we asked the question, “what environmental changes are required during the spring to trigger the upregulation of photosynthesis and which components of the photosynthetic apparatus control the upregulation?” We studied the processes in fir and pine needles. Our results revealed that in both species it is water availability, not warm temperature which triggers the upregulation of springtime photosynthesis.

Our studies have further shown that the first biochemical step to be affected during the upregulation of springtime photosynthesis is relaxation of sustained non-photochemical quenching of photosynthesis and thus improvement in the quantum yield (Fv/Fm) of photosynthesis. These processes are affected within hours after springtime rehydration. Resumption of diurnal opening of stomata is the next process to be affected, occurring within 36 hours of springtime rehydration. The slowest, and most rate-limiting process to be affected is the recovery of the carboxylation capacity of photosynthesis (increase in slope of the A:Ci relationship). This probably reflects the need for reconstitution of the Rubisco carboxylating enzyme, a step that takes 2-3 days to occur in pine needles and 7-8 days in fir needles. We are currently at the stage of analyzing these data and preparing them for publication.

We spent a lot of time over the seven months in a collaboration to develop a modeling context within which we can interpret our long-term eddy covariance measurements of NEE and evaluate seasonal differences in the controls over photosynthesis and respiration. As discussed above, we conducted an analysis of five years of NEE data from our site using an inverse approach developed by Professor Yiqi Luo and a Bayesian approach developed by Drs. Rob Braswell and Dave Schmiel.

PRODUCTS

None to report as yet.

William Pockman

Title: Impact of rainfall variability and woody encroachment on productivity in a semiarid grassland in New Mexico.

Results to Date:

After unanticipated delays related to the federal budget process, NIGEC funds for year 1 were received in mid-December 2003. Pre-award funds available through UNM were insufficient to cover the initially high costs and labor-intensive task of establishing new plots. In the first 8 months, 2 technicians were hired with leveraged support of the Sevilleta LTER program, water addition plots have been established, our irrigation systems have been designed, tested and constructed at all plots and installation of our water treatment system at UNM is nearly complete. Because of delays at UNM establishing the water source, we will begin water-addition treatments in October (instead of July). Although this is an unexpected delay from our original schedule, starting treatment in the fall is advantageous because it will allow us to collect pre-treatment productivity data from all plots and will avoid confounding current year productivity measurements with partial treatments during the summer monsoon. These efforts, our findings, publications and presentations to date are described below.

Water-addition plots: design and construction
With the construction of our water-addition plots, we now have an experimental system consisting of 27 plots (~10 m x ~15 m) with nine plots each in grassland, shrubland and the mixed grass-shrub ecotone. Like the existing rainout and control plots, the water-addition plot perimeters were trenched to 1.2m, vertical profiles of soil moisture sensors have been installed under canopy and bare soil cover and measurements have begun to provide a pre-treatment period of comparison with control plots. Vegetation measurements have also been completed to document the cover and abundance of all species in 3 1 m wide transects in each plot. Overhead photos will be collected in September 2004 immediately prior to the start of water-addition treatment and at 6 month intervals thereafter.

Irrigation system and water source
Our irrigation system consists of a series of square-pattern nozzles attached to overhead pipes that can be suspended from steel wires permanently mounted above each plot. These irrigation arms are put in place before each irrigation event and stored in shelters near the plots between events. Storage tanks (1100 gal) have been placed at each water-addition plot to allow us to move sufficient water for each treatment to the site in advance. By storing all necessary water supplies at the site, we can position all irrigation systems in advance and rapidly move between plots in grassland, shrubland and ecotone to apply all water additions in a short period of time, facilitating comparisons among plots in different vegetation communities.

We use application rates of 50 mm/hour, a reasonable intensity for NM storms that yields a spatially homogenous pattern with realistic drop sizes. As mentioned above, we have elected to delay the start of treatment until the start of the new water year in early October 2004. This slight change is a result of delays in the process of locating space for the water treatment facility on the UNM campus. By waiting until October, the physiological and productivity responses that we observe will reflect treatment conditions throughout the annual period of activity, which is particularly important for the grasses because their activity occurs primarily during July-September.

Data collection and research findings
Data collection on the water-addition plots has begun as we have completed the process of establishing our experimental plots. Initial vegetation transect and shrub growth measurements are complete, soil moisture sensors are installed, connected to dataloggers and are being measured, and the treatment will begin in October, 2004. Below we summarize our findings to date across the control and rainout plots that are already established. These responses will be a key point of comparison to the water-addition data that we are presently collecting.

Productivity
Aboveground net primary productivity (ANPP) is measured annually for grasses and 3 – 5 times per year for shrubs (March, late June, and October). Shrub ANPP is measured using allometric methods based on the change in size of first order twigs. This method is appropriate because it has greater sensitivity for measuring growth over shorter periods than measurements of whole canopy dimensions. Twigs to be measured are marked with paint and the basal diameter and length are measured. Then regressions are used to convert twig dimensions into biomass. A second set of regressions relating the number of first order twigs per plant to plant volume are used to obtain whole plant biomass. Incremental changes in whole plant biomass can then be estimated by repeated sampling. Grass ANPP is measured with destructive sub-sampling of 5, 10 x 10 cm areas per plot. Samples are sorted to separate green biomass before drying and weighing. These samples also allow us to calculate LAI for grasses by separating leaf tissue and using a regression between leaf area and mass.

Over the first two years of treatment, shrubs have not exhibited significant differences in twig growth between control and rainout plots. In contrast, we have observed a significant decrease in NPP between control and rainout plots at the grassland site. This trend has also been observed at the ecotone plots but has not been significant.

Vegetation transects
For comparison with productivity data and overhead photos (see below), the field crew of the Sevilleta LTER measures belt transects (1 m width, 3 per plot) across the narrow dimension of each plot to measure the abundance and cover of all plant species. Transects are measured at peak activity during the spring and fall of each year. Each transect is measured by placing a 1 m square frame, sub-divided into 100 cm2 cells to aid in measurement of percent cover. All species in the frame are measured, providing the data required to assess changes in composition over the course of the study. Plant cover, litter and bare soil are be measured at ground level across the transect while shrub canopy cover is taken as a separate measurement layer. This permits ground-level cover measurements to sum to 100% while shrub canopy can be treated separately in analyses of cover change.

Species cover and abundance
For comparison with species productivity data, plot scale plant cover and bare soil are assessed with overhead photos collected using a digital camera and analyzed using ArcGIS software. Three cover types have been identified: (1) shrub canopy; (2) grass canopy and surrounding litter; and (3) interspace. The amount and spatial distribution of different cover types are measured using digital photos (3.34 megapixel) of each plot, collected at peak biomass to capture species dynamics and changes in the drainage network. A Nikon CoolPix 990 camera mounted on a specially designed boom achieves sufficient elevation (7 m) to cover an entire plot with a set of 6 photos. After rectification and assembly, ArcGIS software is used to measure total grass and shrub cover. These mosaics allow detailed, field checked, measurements of changes in grass and shrub size, stone cover, and interspace connectivity. Mosaics are assembled for each plot on a yearly basis.

Comparisons of overhead photos from Fall of 2002 and 2003 in the control and drought plots indicates that one year of drought treatment has led to a significant decrease in grass cover in the drought treatment of grassland plots. These cover changes are consistent with the decreased productivity observed in the grass drought plots in both 2002 and 2003. Although the decrease in cover may have been exacerbated by low precipitation and late and below-average summer monsoon, no such differences were detected in the control plots over the same period. Overhead photos and productivity measurements from 2004, with a strong ongoing summer monsoon, will help reveal whether this result is indicative of a long-term trend in the drought treatment. If this pattern of decreasing cover continues, we expect to have an opportunity to observe how the drainage network and redistribution of precipitation are affected by long-term drought.

Plant and Soil Water potential
Soil water potentials, measured in vertical profiles under grass and shrub canopies and under bare soil, have continued to exhibit the pattern that we have observed in this system. For a given storm > 2 mm, infiltration is greater under plant canopies than under bare soil. The spatial pattern of infiltration is thus linked to the pattern of vegetation. As our treatments begin (water-addition) and continue (rainout), we will be able to determine how changes in plant cover influence these patterns of redistribution and the final soil moisture pattern that drives ecosystem response to a pulse of precipitation.

Belowground responses
Belowground responses are difficult to measure in long term studies such as ours because of the need to maintain the integrity of the soil system to prevent artifacts related to destructive soil sampling etc. In our ecotone plots, we installed minirhizotron tubes in all treatments in Fall 2003 and began collecting data from these tubes in Spring of 2004. The tapes collected at several time points during the year will be analyzed during winter 2004/2005 to track differences in the amount and timing of root growth and mycorrhizal associations. In addition, we are discussing the use of resin bags to assay nutrient availability under different canopy types in each treatment. If we implement this approach, bags would be placed in the soil through permanently installed access tubes to limit plot disturbance. This would allow us to assess the consequences of the spatial pattern of infiltration for integrated nutrient availability under our treatment regimes.

PRODUCTS

• Completed installation of rainfall addition plots
• Designed and tested irrigation system to apply rainfall additions.

Tingjun Zhang

Title: Investigation of the Spatial and Temporal Variations of the Seasonally Frozen Ground
in the Contiguous United States

Results to Date:

Based on the validated frozen soil algorithm, we conducted investigation on the near-surface soil freeze/thaw cycle in the contiguous United States from 1978 through 2003 using SMMR and SSM/I data. As proposed, we investigated the timing (onset of soil freeze in autumn and last day of soil freeze in spring), frequency of freeze/thaw cycle, number of days of soil freeze, and daily area extent. The thickness of daily soil freeze will be investigated through numerical modeling. When we say the near-surface soil freeze/thaw cycle, we mean it is approximately 10 cm or less from the ground surface.

• Onset of soil freeze in autumn: Based on data from 1978-2003, the onset date of soil freeze in autumn starts in October along the Rocky Mountains and North Dakoda (Figure 1), then gradually expand into west U.S. and south regions. The maximum extent of soil freeze is reached in January. Approximately, 80% of the land in the contiguous United States experiences freezes in winter. Overall, the trend of the onset date of soil freeze in autumn becomes late, especially in the east past of the contiguous United States (Figure 2). The delay can be up to three weeks in some parts of the study area, which is very significant change. Further validation of these results is needed, such as using ground-based measurements.

• Last day of soil freeze in spring: The climatology of the last date of near-surface soil freeze starts from the south in January and becomes late in the north (Figure 3). Over the majority of western and northern States, the last date of soil freeze is in April or May in the Rocky Mountains area. The last date of soil freeze becomes later in the central plain and eastern States, while in the western States, especially along the Rocky Mountains, the last date of surface soil freeze becomes earlier than the long-term average (Figure 4). There are two completely different regimes of changes in the surface soil freeze in spring over the study area. This change potentially has great impact on surface hydrology and carbon exchange.

• Duration of the surface soil freeze: The duration of the surface soil freeze is defined as the time period from the onset date of soil freeze in autumn to the late date of soil freeze in the spring (Zhang et al., 2003). Overall, the average duration varies from les than a month in the south to more than eight months in the Rocky Mountains region (Figure 5). Along the Rocky Mountains, duration lasts about five to six months. Over the central plain and eastern part of the States, the duration usually last three to four months, while in the northern and western States, the duration can last about four to five months. Except the Rocky Mountains region, the duration of surface soil freeze becomes longer in much for the country (Figure 6). This is mainly due to the delay of spring thaw of the frozen soil.

• Frequency of the surface soil freeze: The frequency of surface soil freeze is defined as the number of days soil freeze divided by the total number of days of a year. Along the Rocky Mountain, more than 50% of the time within a year, near-surface soil is in frozen state (Figure 7). Note, even the duration of surface soil freeze can last as long as eight months, the frequency is just above 50%. This means there are days surface soil may not in frozen state between freeze/thaw cycles. Over the majority of the country, surface soil is in frozen state for about 20% to 40% of the time within a year. In southern States, the frequency decreases accordingly. Interestingly, the frequency along the Rocky Mountains is decreasing while the rest of the country is increasing (Figure 8). Further study is needed to validate the discoveries in this study. Potential implication changes in frequency of soil freeze can be extremely significant in surface hydrology and carbon exchange studies.

• Modeling study of soil freeze/thaw over the Niwot Ridge: We have conducted detailed numerical simulation on soil seasonally freezing and thawing over several sites near the Niwot Ridge Mountain Research Station. Ground-based measurements can go back as far as early 1950s. We used daily air temperature and snow cover depth to drive our model. The output was used against the ground-based measurements. The comparison between modeled results and in-situ data agree well. The modeled data set will be used to evaluate carbon fluxes over this area. The modeled outputs include daily soil temperatures up to 3.2 m and daily soil freeze/thaw depths. We expect at least one paper will come out this work. This work will be continued in year 3.

So far, we have not completed the countrywide soil freezing depth simulation, mainly due to the model we used is not applicable. Now, the new model is in place, we expect soil freeze/thaw depth can be modeled in next few months. Further analysis will be conducted when we have soil freeze/thaw depth data, such as climatology of soil freezing depth and changes in soil freezing depth over the past decades in the contiguous United States.

PRODUCTS

• Five-day composite of the near-surface soil freeze-thaw status of soils over the contiguous United States from 1978 to 2003

• Gridded daily air temperature and snow thickness with resolution of 2.5 km X 2.5 km over the contiguous United States from 1978 to 2000