NDACC Theory and Analysis Protocol

The primary goals of the Network for the Detection of Atmospheric Compostion Change, NDACC, have been established as:

1) To make the observations through which changes in the physical and chemical state of the stratosphere can be determined and understood. In particular, to make the earliest possible identification of changes in the Earth's ozone layer and to discern the causes of such changes.

2) To provide an independent calibration of satellite sensors of the atmosphere.

3) To obtain the data that can be used to test and improve multi-dimensional stratospheric chemical and dynamical models.

The achievement of such goals requires high precision state-of-the-art measurements together with an analysis and interpretation procedure that both ensure excellent data quality and provide ready data access. The NDACC Data Protocol was thus structured to yield a verifiable product, referred to as "NDACC data", that would form the basis for trends detection and understanding.

It is expected that the analysis and verification process through which atmospheric observations are transformed into NDACC data will necessarily involve the collaboration between the experimental measurement teams and theorists and statistical scientists. Such collaboration should be valuable in areas such as: (i) data analysis (both the forward model and the retrieval of geophysical parameters), (ii) statistical studies (ascertaining the significance of the results), (iii) application of atmospheric modelling for the validation of the data, and (iv) scientific analysis of the data both before and after the data are in the public domain.

Areas (i) and (ii) are cornerstones of the NDACC; hence, the formal organization of the appropriate theoretical and statistical help is essential. The targeting of areas (iii) and (iv) as important NDACC goals also requires action at the early stages of NDACC implementation since few experimentalists are accomplished modelers, and vice versa. In addition, the early addition of theorists and statistical scientists to the NDACC Science Team will speed the process of data planning, analysis, modeling, and interpretation.

The NDACC has been endorsed by the United Nations Environment Programme (UNEP) and the International Ozone Commission (IOC) of the International Association of Meteorology and Atmospheric Physics and by the World Meteorological Organization (WMO) as a major contributor of WMO's Global Ozone Observing System (GO3OS) within the framework of its Global Atmosphere Watch (GAW).

Current General Needs of the Instrument Groups:

Sampling Plans - Each Principal Investigator (PI) needs a plan for sampling that will accomplish early detection of trends for each species or parameter while not overburdening resources by collecting more data than is necessary for trend determination of the required sensitivity. These aspects of experimental design, including measurement frequency, need to be constructed from the statistical point of view, with attention given to parameter or species variation levels (e.g., sampling and testing standard deviations), autocorrelation in successive values, measurement locations, missing values, averaging periods, coincidental measures on other species, etc.

An overall plan for sampling must be designed to account for the general needs of the NDACC programs and the specific requirements for detecting changes in a given species measured by an instrument at a chosen site.

Quality Assurance and Control Plan - A general plan and set of guidelines for quality assurance and control will be needed by the PIs to maintain accuracy and precision of their instruments to the highest levels possible. This should have a strong statistical quality control basis that includes regular calibration checks and the detection of special cause variation (e.g., instrument drifts, outliers, inadequate viewing conditions, etc.). Experimental design procedures will need to be applied to identify controllable operating conditions that keep the instruments on target with minimal measurement (i.e., analytical) variation. Control charts on the instrument performance against a set of standards will be needed.

Intercomparison - NDACC measurements from different instruments will need to be compared for consistency and evidence of statistically significant differences. Consistency checks will be made against satellite instruments and with complementary measurements chosen for standard level comparisons. Causes of significant measurement differences between instruments measuring the same parameter will require interpretation from instrumental, theoretical, retrieval, and/or statistical bases.

Error Analysis - Error analysis and retrieval characterization need to be made in a unified and comparable manner, so that the nature of the remote measurements can be properly understood and intercomparisons can be carried out correctly. The Theory and Data Analysis Team will establish the guidelines for these procedures.

Statistical Analysis of Long-Term Trends - Time series trend models will need to be developed. These may be multivariate in form if certain species need to be viewed jointly. These models will be needed to evaluate seasonality in the means, trends, and standard deviations. Also, the autocorrelation structure in the data and the effects of QBO, ENSO, solar, and other factors need to be assessed for the inclusion in the trend models to improve precision and accuracy of the trend estimates.

Chemistry and Dynamics - Understanding of the chemistry and dynamics will be needed to assess the diurnal, seasonal, interannual, and spatial patterns in the data. Theoretical modeling will be an important component in evaluating trends.

A person (or persons) who could bring experience and expertise in the following areas could make a very valuable contribution.

Lidar Retrieval Techniques - In all cases lidar signals have to be processed to obtain range resolved values of the geophysical parameters of interest: ozone, temperature, and aerosol scattering ratios. This processing generally implies spatial filtering, temporal averaging, and range derivative of the raw signals. Various methods have been developed using different filtering techniques (polynomial, derivative filters, windowing, etc.) to adapt the range resolution of the measurements to the expected signal to noise ratio. In particular, the range resolution has to vary with altitude.

Analysis of these various processing techniques can be conducted following the general methodology adopted for data processing evaluation in the Report of the International Ozone Trends Panel 1988. This will allow the determination of the real altitude resolution and precision of the measurements. In addition, the magnitude of various systematic errors related to the use of ancillary data (ozone cross-section dependence on temperature, influence of pressure boundary conditions on temperature retrievals, calibration of aerosol measurements, etc.) should be studied further.

Microwave Spectroscopy - Microwave retrievals can only be as good as the forward radiative transfer models used in the inversion process. The accuracy of the forward models depends mainly on the line list and the spectral line parameters used in these models. The theory and analysis team should include an individual who is an expert in microwave spectroscopy. The person would be responsible for interfacing with the PIs to ensure that their forward model line lists are kept current. In addition, the team could help ensure that the radiative transfer models themselves are up to standard by coordinating forward model comparison studies.

Inversion Techniques - It is important that the microwave retrievals are performed in the 'optimum' way and that such fundamental properties of the retrievals as vertical resolution and errors are properly characterized. Expertise in these areas is certainly required in the theory and analysis team. The team should provide support to the PIs in developing and maintaining their operational retrieval algorithms. In addition, the team should provide oversight to ensure the quality of these retrieval algorithms. A coordinated "blind" inversion simulation comparison study, conducted by the team, could be helpful in this regard.

UV/Visible observations of chemical species are carried out largely during twilight conditions and, therefore, require a good understanding of twilight chemistry. In addition, understanding the air mass factors characterizing the vertical columns of species measured requires a detailed analysis of scattering processes. Spectroscopy in the 300-700 nm region is a key aspect of the data analysis including consideration of the Ring effect, polarization, and the effects of clouds. Finally, as with other measurement techniques, a thorough understanding of error analysis is needed to aid in the NDACC data analysis and verification process.

The infrared investigators need one or more theorists who will specialize in infrared data and help primarily in the data analysis. They will help in: (i) the selection of small regions (microwindows) for the retrieval of individual compounds, (ii) the determination of the molecular parameters and their uncertainties for use in the retrievals, design, and evaluation of retrieval algorithms, and (iii) studies of the information content of the line profile for determination of the vertical profiles of constituents. In these last two areas, the infrared work has a lot in common with microwave work, and the "technique theorists" for the two groups will want to work closely together.

Planning:

The specific needs of each instrument group for assistance from Theoretical Investigators (TIs) will continue to be determined by the group concerned and communicated to the NDACC Steering Committee. Common or general needs of the NDACC for TIs will be determined by the Steering Committee with advice from the Science Team. A preliminary outline of such general and specific needs has been presented above.

On the basis of these needs, proposals from scientists in the appropriate fields will be solicited by the Steering Committee. Such proposals are expected to identify specific tasks that are designed to address these needs.

Proposals for "open interaction" with the NDACC from theorists, statisticians, and data analysts will also be considered.

Scientists wishing to participate in the NDACC as theoretical investigators should submit descriptions of their proposed theory or analysis activity, identifying either the specific tasks that will be impacted or the value of the "open interaction" that they seek. Such proposals should be brief and be submitted to the NDACC Steering Committee for assigned evaluation. The NDACC evaluation will be conducted as follows:

1) Proposals in which the proposed tasks are specific to one of the NDACC primary instrument types (microwave, infrared, UV/Visible, lidar) shall be evaluated by that NDACC Principal Investigator (PI) instrument subgroup and approved by the Steering Committee.

2) Proposals whose tasks are general to the whole Network or to more than one instrument type will be evaluated by appropriate subgroups of the NDACC Science Team and approved by the Steering Committee. Upon selection, each theorist, statistical scientist, or data analyst whose tasks are general may be assigned by the Steering Committee to a particular NDACC subgroup, where the person's skills can be most readily and appropriately used.

3) For cases not directly covered under items 1 or 2 above, the NDACC Steering Committee shall decide on an appropriate evaluation procedure. External peer review may also be sought where it is deemed appropriate.

If the proposing investigator requires specific resources in support of the theory and analysis activity, the proposal should also be submitted by the investigator to the appropriate funding agency. In such cases, a "point of contact" within the agency should be identified in the proposal submitted to the NDACC Steering Committee so that the results of the NDACC evaluation can be communicated to the funding agency.

The investigators selected will form a Theory and Analysis Team so that the common experimental design, analysis, dynamical, and chemical approaches are shared amongst all instrument subgroups. Such an overall Team formation does not preclude the additional assignment to a specific measurement subgroup. It is expected that the Theory and Analysis Team will schedule regular technical exchange meetings.

As with all investigators associated with the NDACC, members of the Theory and Analysis Team must comply with the NDACC publication and data access policies set forth in the NDACC Data Protocol. The proposed interaction between a TI and the NDACC may be for a limited or open-ended period of time.