Based on a uniquely dense network of surface towers measuring continuously the atmospheric concentrations of greenhouse gases (GHGs), we developed the first comprehensive monitoring systems of CO 2 emissions at high resolution over the city of Indianapolis. The urban inversion evaluated over the 2012-2013 dormant season showed a statistically significant increase of about 20% (from 4.5 to 5.7 MtC ± 0.23 MtC) compared to the Hestia CO 2 emission estimate, a state-of-the-art building-level emission product. Spatial structures in prior emission errors, mostly undetermined, appeared to affect the spatial pattern in the inverse solution and the total carbon budget over the entire area by up to 15%, while the inverse solution remains fairly insensitive to the CO 2 boundary inflow and to the different prior emissions (i.e., ODIAC). Preceding the surface emission optimization, we improved the atmospheric simulations using a meteorological data assimilation system also informing our Bayesian inversion system through updated observations error variances. Finally, we estimated the uncertainties associated with undetermined parameters using an ensemble of inversions. The total CO 2 emissions based on the ensemble mean and quartiles (5.26-5.91 MtC) were statistically different compared to the prior total emissions (4.1 to 4.5 MtC). Considering the relatively small sensitivity to the different parameters, we conclude that atmospheric inversions are potentially able to constrain the carbon budget of the city, assuming sufficient data to measure the inflow of GHG over the city, but additional information on prior emission error structures are required to determine the spatial structures of urban emissions at high resolution.

We present a high-resolution atmospheric inversion system combining a Lagrangian Particle Dispersion Model (LPDM) and the Weather Research and Forecasting model (WRF), and test the impact of assimilating meteorological observation on transport accuracy. A Four Dimensional Data Assimilation (FDDA) technique continuously assimilates meteorological observations from various observing systems into the transport modeling system, and is coupled to the high resolution CO 2 emission product Hestia to simulate the atmospheric mole fractions of CO 2 . For the Indianapolis Flux Experiment (INFLUX) project, we evaluated the impact of assimilating different meteorological observation systems on the linearized adjoint solutions and the CO 2 inverse fluxes estimated using observed CO 2 mole fractions from 11 out of 12 communications towers over Indianapolis for the Sep. ). Wind speed MAE and ME are larger in daytime than in nighttime. PBL depth MAE is reduced by ~1 0%, with little bias reduction. The inverse results indicate that the spatial distribution of CO 2 inverse fluxes were affected by the model performance while the overall flux estimates changed little across WRF simulations when aggregated over the entire domain. Our results show that PBL wind observations are a potent tool for increasing the precision of urban meteorological reanalyses, but that the impact on inverse flux estimates is dependent on the specific urban environment.

The objective of the Indianapolis Flux Experiment (INFLUX) is to develop, evaluate and improve methods for measuring greenhouse gas (GHG) emissions from cities. INFLUX's scientific objectives are to quantify CO 2 and CH 4 emission rates at 1 km 2 resolution with a 10% or better accuracy and precision, to determine whole-city emissions with similar skill, and to achieve high (weekly or finer) temporal resolution at both spatial resolutions. The experiment employs atmospheric GHG measurements from both towers and aircraft, atmospheric transport observations and models, and activity-based inventory products to quantify urban GHG emissions. Multiple, independent methods for estimating urban emissions are a central facet of our experimental design. INFLUX was initiated in 2010 and measurements and analyses are ongoing. To date we have quantified urban atmospheric GHG enhancements using aircraft and towers with measurements collected over multiple years, and have estimated whole-city CO 2 and CH 4 emissions using aircraft and tower GHG measurements, and inventory methods. Significant differences exist across methods; these differences have not yet been resolved; research to reduce uncertainties and reconcile these differences is underway. Sectorally-and spatially-resolved flux estimates, and detection of changes of fluxes over time, are also active research topics. Major challenges include developing methods for distinguishing anthropogenic from biogenic CO 2 fluxes, improving our ability to interpret atmospheric GHG measurements close to urban GHG sources and across a broader range of atmospheric stability conditions, and quantifying uncertainties in inventory data products. INFLUX data and tools are intended to serve as an open resource and test bed for future investigations. Well-documented, public archival of data and methods is under development in support of this objective.

As part of the Indianapolis Flux (INFLUX) experiment, the accuracy and biases of simulated meteorological fields were assessed for the city of Indianapolis, IN. The INFLUX project allows for a unique opportunity to conduct an extensive observation-to-model comparison in order to assess model errors for the following meteorological variables: latent heat and sensible heat fluxes, air temperature near the surface and in the planetary boundary layer (PBL), wind speed and direction, and PBL height. In order to test the sensitivity of meteorological simulations to different model packages, a set of simulations was performed by implementing different PBL schemes, urban canopy models (UCMs), and a model subroutine that was created in order to reduce an inherent model overestimation of urban land cover. It was found that accurately representing the amount of urban cover in the simulations reduced the biases in most cases during the summertime (SUMMER) simulations. The simulations that used the BEP urban canopy model and the Bougeault & Lacarrere (BouLac) PBL scheme had the smallest biases in the wintertime (WINTER) simulations for most meteorological variables, with the exception being wind direction. The model configuration chosen had a larger impact on model errors during the WINTER simulations, whereas the differences between most of the model configurations during the SUMMER simulations were not statistically significant. By learning the behaviors of different PBL schemes and urban canopy models, researchers can start to understand the expected biases in certain model configurations for their own simulations and have a hypothesis as to the potential errors and biases that might occur when using a multi-physics ensemble based modeling approach.

Many precipitation-driven data products from land data assimilation systems support assessments of droughts, floods, and other societally-relevant land-surface processes. The accumulated precipitation used as input to these products has a significant impact on water budgets; however, the effects of daily distribution of precipitation on these products are not well known. A comparison of the Integrated Multi-satellite Retrievals for GPM (IMERG) and Climate Hazards Group InfraRed Precipitation with Stations version 2 (CHIRPS2) rainfall products over the continental United States (CONUS) was performed to quantify the impacts of the daily distribution of precipitation on biases and errors in soil moisture, runoff, and evapotranspiration (ET). Since the total accumulated precipitation between the IMERG and CHIRPS product differed, a third precipitation product, CHIRPS-to-IMERG (CHtoIM), was produced that used CHIRPS2 accumulated precipitation totals and the daily precipitation frequency distribution of IMERG. This new product supported a controlled analysis of the impact of precipitation frequency distribution on simulated hydrological fields. The CHtoIM had higher occurrences of precipitation in the 0–5 mm day−1 range, with a lower occurrence of dry days, which decreased soil moisture and surface runoff in the land-surface model. The surface soil layer had a tendency to reach saturation more often in the CHIRPS2 simulations, where the number of moderate to heavy precipitation days (>5 mm day−1) was increased. Using the blended CHtoIM product as input reduced errors in surface soil moisture by 5–15% when compared to Soil Moisture Active/Passive (SMAP) data. Similarly, ET errors were also slightly decreased (~2%) when compared to SSEBop data. Moderate changes in daily precipitation distributions had a quantifiable impact on soil moisture, runoff, and ET. These changes usually improved the model when compared to other modeled and observational datasets, but the magnitude of the improvements varied by region and time of year.