Applications and Data Assimilation 1

Over the past few years a growing number of space observations focusing on atmospheric composition have become available and this trend will continue with the launch of new satellites (ESA-Sentinels, NASA-TEMPO, KARI-GEMS and JAXA air quality and climate mission) in the near future. To justify the production and launch of these expensive instruments, there is a need for determining the added value of future satellite instruments and their optimal design in an objective way. One methodology that can do so is the OSSE (Observing System Simulation Experiment). Although extensively used in the meteorological community, its use in the field of air quality and climate is still limited and a common approach is desirable.

A full and realistic OSSE consists of the following elements: (1) a realistic nature run to simulate the true state of the atmosphere; (2) an observation simulator including full instrument description, full radiative transfer models or scene-dependent averaging kernels, cloud information, product retrieval scheme, and realistic error and error correlation estimates; (3) a well-established data assimilation model different from the nature run model; (4) model independent results; (5) a calibration run; and (6) a dedicated quantitative evaluation focusing on the driving science questions, and including statistical significance of the results. When containing this set of elements the OSSEs can be used to provide realistic answers on the added value and optimal design of future instruments for a specific application.

In this contribution we will present the methodology and the potential of OSSEs for  satellite observations of atmospheric composition. Illustrative examples will be given from existing air quality OSSEs, amongst others an OSSE for the potential value of small satellite instruments observing NO₂ columns and the ESA-ISOTROP OSSE investigating the benefits of sentinel 4 (GEO) and 5 (LEO) measurements of ozone, CO, NO₂ and HCHO to better constrain pollutant concentrations and precursor emissions that influence air quality.

Current satellite observations of tropospheric composition made from low Earth orbit provide at best one or two measurements each day at any given location. Coverage is global but sparse, often with large uncertainties in individual measurements that limit examination of local and regional atmospheric composition over short time periods. This has hindered the operational uptake of these data for monitoring air quality and population exposure, and for initializing and evaluating chemical weather forecasts. By the end of the current decade there are planned geostationary Earth orbit (GEO) satellite missions for atmospheric composition over North America, East Asia and Europe, with additional missions proposed. Together, these present the possibility of a constellation of GEO platforms to achieve continuous time-resolved high-density observations of continental domains for mapping pollutant sources and variability on diurnal and local scales. We describe Observing System Simulation Experiments (OSSEs) to evaluate the contributions of these GEO missions to improve knowledge of near-surface air pollution due to intercontinental long-range transport and quantify chemical precursor emissions. We discuss the requirements on measurement simulation, chemical transport modeling, and data assimilation for a successful OSSE infrastructure. Our approach uses an efficient computational method to sample a high-resolution global GEOS-5 chemistry Nature Run over each geographical region of the GEO constellation. The demonstration carbon monoxide (CO) observation simulator, which is being expanded to other chemical pollutants, currently produces multispectral retrievals and captures realistic scene-dependent variation in measurement vertical sensitivity and cloud cover. We use the DART Ensemble Adjustment Kalman Filter to assimilate the simulated observations in a CAM-Chem global chemistry-climate model Control Run. The impact of observing over each region is evaluated using data denial experiments. Finally, we report on international collaborations using the OSSE approach to determine expected performance of planned satellite systems and set requirements for future missions.

A changing environment in the 21^st Century, in particular because of constituents associated with air quality and climate change, and with consequences for societal well-being, illustrates the need to monitor the environment, including compliance with environmental regulations from societal actors such as public authorities, governments and industry. For air quality, this monitoring involves, inter alia, measurements of key pollutants (e.g., ozone and carbon monoxide) in the lowermost troposphere even in the atmospheric boundary layer at spatio-temporal scales relevant to policy makers (temporal frequencies of order less than 1 hour; spatial scales of order less than 10 km). Monitoring criteria also apply to greenhouse gases affecting climate change.

In this presentation, we identify the role of data assimilation observing system simulation experiments (OSSEs) in determining the future global observing system (GOS) to monitor atmospheric constituents in a changing environment. We describe requirements for constructing such OSSEs, and discuss caveats associated with setting up and interpreting the OSSEs. To illustrate the concept, we present OSSEs performed within the MUSICQA project to assess the added-value of planned and proposed geostationary satellite platforms to measure atmospheric constituents affecting air quality.

This year, the Copernicus Atmosphere Monitoring Service (CAMS) is entering its operational phase. After many years of pioneering work through a series of EU and ESA funded projects (GEMS, Promote, MACC), a mature data assimilation and forecasting system for global atmospheric composition has been established. This system, which is based on the ECMWF numerical weather prediction system, forms one of the important core elements of the Service. It uses a wide array of satellite and in-situ data observing both meteorological and atmospheric composition variables to provide a best estimate of the current state of the atmosphere on a daily basis. These analyses are then used as initial conditions for 5-day global forecasts of atmospheric composition, covering aerosols, chemical species and greenhouse gases. The same system is also used to produce reanalyses allowing a wider range of data to be used, because there is no strict near real-time requirement.

CAMS forms an important component of the Copernicus programme adding significant value to the information coming from the current and future observing system. This presentation will provide an overview of the CAMS use of current satellite data, such as OMI, GOME-2, IASI, MOPITT, MLS, MODIS, and GOSAT. With the expected launch of Sentinel-3 and Sentinel-5p in the coming year crucial new information on atmospheric composition will become available, both for operational and research purposes. We will show how CAMS, in close collaboration with ESA, is preparing for these new observations to enable maximum benefit for the users of the atmosphere services as early as possible. We will also show how CAMS can provide important feedback about the data quality through careful data monitoring in the comprehensive data assimilation system. The latter information will be very beneficial for the scientific community to make more optimal use of the Sentinel data.

Emission inventories of air pollutants are crucial information for policy makers and form important input data for air quality models. Unfortunately, bottom-up emission inventories, compiled from large quantities of statistical data, are easily outdated for regions such as China, the Middle East, India, and South Africa, where rapid economic growth changes emissions accordingly. Alternatively, top-down emission estimates from satellite observations of air constituents have important advantages of being spatial consistent, having high temporal resolution, and enabling emission updates shortly after the satellite data become available.

Constraining emissions from concentration measurements is, however, computationally challenging. Within the GlobEmission project of the European Space Agency (ESA) a new algorithm has been developed, specifically designed for fast daily emission estimates of short-lived atmospheric species on a mesoscopic scale (0.25 × 0.25 degree) from satellite observations of column concentrations. The algorithm needs only one forward model run of a chemical transport model to calculate the sensitivities of concentrations to emissions, using trajectory analysis to account for transport. By using a Kalman filter in the inverse step, optimal use of the a priori knowledge and the newly observed data is made.

We apply the algorithm for NO_x emission estimates to East China, the Middle East, India, and South Africa, using the CHIMERE model together with tropospheric NO₂ column retrievals of the OMI and GOME-2 satellite instruments. The observations are used to construct time series of monthly emissions, which reveal important local emission trends such as the effect of emission reduction measures during the 2014 Youth Olympic Games in Nanjing, and the impact and recovery from the global economic crisis. The algorithm is also able to detect emerging sources (e.g. new power plants) and improve emission information for areas where proxy data are not or badly known (e.g. shipping emissions). The new emission estimates result in a better agreement between observations and simulations of air pollutant concentrations, facilitating improved air quality forecasts.

Since 2006 the Sodankylä satellite data centre has utilized the direct broadcast data downlink of the NASA’s Aura satellite to process Dutch-Finnish Ozone Monitoring Instrument (OMI) data and to distribute O3, SO2, UV-radiation and aerosol index observations within 20 min after the satellite overpass. These so called Very Fast Delivery OMI products have been used in monitoring and forecasting the transportation of volcanic plumes, observing ash clouds from forest fires and in following ozone abundance and UV-radiation. Since 2014 NASA’s Suomi NPP/OMPS data has also been received and processed in addition to OMI. The upgraded system, named SAMPO, was operational just in time to monitor the Bárðarbunga/Holuhraun fissure eruption in Iceland, which started in September 2014.

In this presentation we demonstrate the direct broadcast products of the SAMPO service and applications of the data. In particular, the recent volcanic eruptions in Iceland are discussed. As the volcanic SO2 plume in 2014 also reached Finland, the data have been compared to ground-based observations. Furthermore, future plans of the Sodankylä satellite data centre as the Copernicus Sentinel Collaborative Ground segment are discussed. The most recent plans include processing global TROPOMI/Sentinel 5 Precursor UV-radiation products.