Abstract. Isoprene fluxes were estimated using eight different measurement techniques at a forested site near Oak Ridge, Tennessee, during July and August 1992. Fluxes from individual leaves and entire branches were estimated with four enclosure systems, including one system that controls leaf temperature and light. Variations in isoprene emission with changes in light, temperature, and canopy depth were investigated with leaf enclosure measurements. Representative emission rates for the dominant vegetation in the region were determined with branch enclosure •neasurements. Species from six tree genera had negligible lsoprene emissions, while significant emissions were observed for Quercus, Liquidambar, and Nyssa species. Abovecanopy isoprene fluxes were estimated with surface layer gradients and relaxed eddy accumulation measurements from a 44-m tower. Midday net emission fluxes from the canopy were typically 3 to 5 mg C m-2 h-l, although net isoprene deposition fluxes of-0.2 to -2 mg C m-2 h-1 were occasionally observed in early morning and late afternoon. Above-canopy CO2 fluxes estimated by eddy correlation using either an open path sensor or a closed path sensor agreed within +5%. Relaxed eddy accumulation estimates of CO2 fluxes were within 15% of the eddy correlation estimates. Daytime isoprene mixing ratios in the mixed layer were investigated with a tethered balloon sampling system and ranged from 0.2 to 5 ppbv, averaging 0.8 ppbv. The isoprene mixing ratios in the mixed layer above the forested landscape were used to estimate isoprene fluxes of 2 to 8 mg C m-2 h-1 with mixed layer gradient and mixed layer mass balance techniques. Total foliar density and dominant tree species composition for an approximately 8100 km2 region were estimated using high-resolution (30 m) satellite data with classifications supervised by ground measurements. A biogenic isoprene emission model used to compare flux measurements, ranging from leaf scale (10 cm2) to landscape scale (102 km2), indicated agreement to within +_25%, the uncertainty associated with these measurement techniques. Existing biogenic emission models use isoprene emission rate capacities that range from 14.7 to 70 [tg C g-1 h-1 (leaf temperature of 30øC and photosynthetically active radiation of 1000 gmol m-2 for oak foliage. An isoprene emission rate capacity of 100 gg C g-1 h-1 for oaks in this region is more realistic and is recommended, based on these measurements.

We performed an atmospheric inversion of the CO₂ fluxes over Iowa and the surrounding states, from June to December 2007, at 20 km resolution and weekly timescale. Eight concentration towers were used to constrain the carbon balance in a 1000 × 1000 km² domain in this agricultural region of the US upper midwest. The CO₂ concentrations of the boundaries derived from CarbonTracker were adjusted to match direct observations from aircraft profiles around the domain. The regional carbon balance ends up with a sink of 178 TgC±35 TgC over the area for the period June–December, 2007. Potential bias from incorrect boundary conditions of about 0.4 ppm over the 7 months was corrected using mixing ratios from four different aircraft profile sites operated at a weekly time scale, acting as an additional source of uncertainty of 18 TgC. We used two different prior flux estimates, the SiBCrop model and the inverse flux product from the CarbonTracker system. We show that inverse flux estimates using both priors converge to similar posterior estimates (10 TgC difference), in our reference inversion, but some spatial structures from the prior fluxes remain in the posterior fluxes, revealing the importance of the prior flux resolution and distribution despite the large amount of atmospheric data available. The retrieved fluxes were compared to eddy flux towers in the corn and grassland areas, revealing an improvement in the seasonal cycles between the two compared to the prior fluxes, despite large absolute differences due to representation errors. The uncertainty of 35 TgC (about 35 gC m²) was derived from the posterior uncertainty obtained with our reference inversion of about 25 to 30 TgC and from sensitivity tests of the assumptions made in the inverse system, for a mean carbon balance over the region of −178 TgC, slightly weaker than the reference. Because of the potential large bias (~20 TgC in this case) due to choice of background conditions, proportional to the surface but not to the regional flux, this methodology seems limited to regions with a large signal (sink or source), unless additional observations can be used to constrain the boundary inflow

Ever-increasing demand for computing capability is driving the construction of ever-larger computer clusters, typically comprising commodity compute nodes, ranging in size up to thousands of processors, with each node hosting an instance of the operating system (OS). Recent studies El, 41 have shown that even minimal intrusion by the OS on user applications, e.g. a slowdown of user processes of less than 1.0% on each OS instance, can result in a dramatic performance degradation-50% or more-when the user applications are executed on thousands of processors.The contribution of this paper is the explication, and demonstration by way of a case study, of a methodology for analyzing and evaluating the impact of the system (all software and hardware other than user applications) activity on application performance. Our methodology has three major components: 1) a set of simple benchmarks to quickly measure and identdy the impact of intrusive system events; 2) a kernel-level profiling tool OprofiZe to characterize all relevant events and their sources; and, 3) a kernel module that provides timing information for in-depth modeling of the frequency and duration of each relevant event and determines which sources have the greatest impact on performance (and are therefore the most important to eliminate).The paper provides a collection of experimental results conducted on a state-of-the-art dual AMD Opteron duster running GNU/Linux 2.6.5. While our work has been performed on this specific OS, we argue that our contribution readily generalizes to other open source and commercial operating systems.

While the performance of compute-bound applications can be effectively guaranteed with techniques such as space sharing or QoS-aware process scheduling, it remains a challenge to meet QoS requirements for end users of I/O-intensive applications using shared storage systems because of the difficulty of differentiating I/O services for different applications with individual quality requirements. Furthermore, it is difficult for end users to accurately specify performance goals to the storage system using I/O-related metrics such as request latency or throughput. As access patterns, request rates, and the system workload change in time, a fixed I/O performance goal, such as bounds on throughput or latency, can be expensive to achieve and may not provide performance guarantees such as bounded program execution time.We propose a scheme supporting end-users' QoS goals, specified in terms of program execution time, in shared storage environments. We automatically translate the users' performance goals into instantaneous I/O throughput bounds using a machine learning technique, and use dynamically determined service time windows to efficiently meet the throughput bounds. We have implemented this scheme in the PVFS2 parallel file system and have conducted an extensive evaluation. Our results show that this scheme can satisfy realistic end-user QoS requirements by making highly efficient use of the I/O resources. The scheme seeks to balance programs' attainment of QoS requirements, and saves as much of the remaining I/O capacity as possible for best-effort programs.