- AIRS NRT
In order to generate data products within 3 hours of observation, a number of changes have been made to the standard processing approach to expedite the availability of input datasets.
The AIRS NRT data products are identical to the AIRS standard products except for the following:
- The NRT granules are produced without previous or subsequent granules if those granules are not available within 5 minutes (previous and subsequent granules are used for calibration and are generally present in 5 minutes)
- Predictive ephemeris/attitude data are used (in contrast, definitive ephemeris/attitude data are used for processing standard products)
- Nominally, a forecast surface pressure is used; if this is unavailable, a surface climatology is used
- No ice cloud properties retrievals are performed (ice cloud properties are a new retrieval in the Version 6 support product AIRX2SUP)
These differences have the following implications for the AIRS NRT Level 1B and Level 2 products.
Level 1B - The atmospheric surface pressure is not used in Level 1B processing; the AIRS Level 1B radiance data produced in the NRT system are nearly identical to the routinely processed AIRS data. There may be small differences in the geolocation (i.e., longitude and latitude). Also, since the NRT system does not wait as long for the previous or subsequent granule data to be present before processing there may be small differences in the radiances of NRT data that were processed without the previous or subsequent granules. This can happen when AIRS is entering or leaving the range of a downlink station (which usually occurs 17 times/day). The reason for the differences is because the standard product uses the space view offset from the previous and subsequent granules for calibration purposes.
Level 2 - The AIRS Level 2 NRT retrieval products can differ from the routine products because of small differences in assumed surface pressure and/or differences in the radiances. Although it is rare, if the forecast surface pressure is not available when the NRT data are produced the NRT algorithm assumes a surface pressure based on a Digital Elevation Map (DEM) rather than the more accurate forecast surface pressure. The differences in the assumed surface pressure tend to be small (~ 10 mbar) near the equator, but can become larger (~ 70 mbar) near synoptic weather events. The pressure differences can also lead to differences in the retrieved temperature and water vapor.
For more information, please see AIRS NRT data products: May 10, 2010, describing the difference between AIRS NRT and standard products.
- Accordion title
Accordion content.
- Accordion title
Accordion content.
- ICESat-2 NRT
ATLAS/ICESat-2 quick look datasets have a latency of 3 days from acquisition. The quick look data files will be archived and distributed from NSIDC DAAC until the standard data files are ingested. ATLAS/ICESat-2 standard datasets have an average latency of 45 days from acquisition
- MLS NRT
In order to generate data products within 3 hours of observation time, a number of changes have been made to standard processing approach to expedite the availability of input datasets.
For detailed information on MLS NRT data product quality, please see the user guide.
The MLS NRT algorithm uses a simplified fast forward model to meet NRT data latency requirements and are therefore not as accurate as the retrievals that constitute the standard MLS products. Nevertheless the results are scientifically useful in selected regions of Earth's atmosphere provided that the data are screened according to the recommendations in the MLS NRT user guide.
The NRT data are typically available within 3 hours of observation and are broken into files containing about 15 minutes of data. The seven most recent days of MLS NRT data are kept online for users to download. After this time users should download the standard, full production Level 2 data products.
- MODIS NRT
In order to generate data products within 3 hours of observation time, a number of changes have been made to standard processing approach to expedite the availability of input datasets.
Near Real-Time Approach
- The delivery of the Level 0 (L0) is expedited from the EOS Data and Operations System (EDOS). NRT processing acquires L0 files at the end of each downlink session (session-based files) within 10-30 minutes of real-time, whereas standard forward processing acquires 2-hour L0 files for Aqua and Terra within 7-8 hours of real-time
- For geolocation, MODIS Terra and MODIS Aqua differ in their approach. For MODIS Terra, the attitude and ephemeris data entrained in the L0 data are used rather than the flight data (FD) product for standard processing. The Aqua instrument (unlike Terra) lacks the ability to generate definitive ephemeris and attitude data, and therefore, NASA's Flight Dynamics Facility (FDF) calculates predicted platform ephemeris data and uploads it to the platform once per day. These data are later returned to the ground stations as Ground-Based Attitude Determination (GBAD) data, which are also time-coordinated with the other instruments. This GBAD ephemeris and attitude data are used to process Aqua MODIS standard products
- Most NRT Product Generation Executive (PGE) codes use the same science algorithm used in the standard processing. However, certain Level 2 (L2) codes have modified production rules to relax the requirements for ancillary data products.
- For Level 3, rolling products such as MOD141N/Q1N, MxD09A1N, MCD43A1N/A2/A2N download this PDF to explain the differences between the NRT and standard "n-day" products
To meet the latency requirement, most L2 NRT Atmosphere PGEs (Cloud Mask/Profile, Cloud, and Aerosol) are allowed to use ancillary input products (e.g. GDAS, NISE, OZONE, TOAST) from an earlier time period, and sometime even several days prior to the current data day while the operational processing could be delayed until the best ancillary data are available. Use of ancillaries from different data days can result in significant difference in the daily cloud and aerosol product generated at NRT compared to the operational products from MODAPS.
The L2 Fire, Snow, and Sea Ice from NRT are nearly identical to the operational product as these PGEs do not make use of ancillary data products. A special version of the L2 Land Surface Reflectance for near real-time use was developed by the Principal Investigator (PI); this code uses the NOAA Global Forecast System (GFS) ancillary product as opposed to Global Data Assimilation System (GDAS) used in the standard processing version. Difference in L2 surface reflectance product from this change to processing code in NRT processing isn’t significant.
The near real-time PGE codes were extensively science tested and all products were compared to the baseline products generated by the standard processing. This comparison was made by the atmosphere and land Quality Assessment (QA) (Land Data Operational Products Evaluation [LDOPE]) staff and by the PIs. All PGEs have been validated for use in LANCE-MODIS by the PIs. This process will be repeated as new versions of the PGEs become available and are considered for inclusion in LANCE-MODIS.
- OMI NRT
Accordion content.In order to generate data products within 3 hours of observation time, a number of changes have been made to standard processing approach to expedite the availability of input datasets.
The quality of OMI NRT Level 2 products compares favorably to the standard products. There appear to be some variances at high latitudes with high solar zenith angles. In the following table we display the maximum differences for a single day and averaged over a week.
| Product | Variable | Daily Maximum Percentage Difference | Weekly Average Maximum Percentage Difference |
|---|
| OMTO3 | Total Ozone Column | 2.64% | 1.40% |
| OMCLDRR | Cloud Fraction | 6.02% | 1.42% |
| | Cloud Pressure | 2.82% | 0.67% |
| OMAERUV | AOD (388 nm) | 5.95% | 2.31% |
- OMPS NRT
In order to generate data products within 3 hours of observation time, a number of changes have been made to standard processing approach to expedite the availability of input datasets.
The quality of OMPS NRT Level 2 products compares favorably to the standard products. There appear to be some variances at high latitudes with high solar zenith angles. In the following table we display the maximum differences for a single day and averaged over a week.
NMTO3-L2 and NPBUVO3-L2
The science team has not found any large difference in the quality of the products; a more detailed analysis will be provided in the near future.
A comparison between Near Real-Time SNPP/OMPS PCA SO2 product and the Standard SNPP/OMPS PCA SO2 product
In order to generate data products within 3 hours of observation time, a number of changes have been made to standard processing approach to expedite the availability of input datasets for the Suomi NPP/OMPS principal component analysis (PCA) SO2 product (NMSO2-PCA-L2).
In general, the NRT Suomi NPP/OMPS PCA SO2 product is very similar to the standard processing product. On average, the differences between the NRT and standard SO2 products are 0.03 Dobson Units (DU) or less. Note that under most circumstances, the actual SO2 loading in the atmosphere is very small, and the percentage difference would be significant even for a very small absolute difference. There are stripes (or rows) with relative large differences between the NRT and the standard products, especially for high latitudes or the edge of the swath. This is likely due to the setup of the PCA retrieval algorithm, in which principal components (PCs) are extracted from the input radiance data for each cross-track position (row) of the each OMPS orbit. The PCs are then used in spectral fitting for all pixels in the same row. Differences in the input radiance and other input data may lead to different PCs and also different number of PCs that are actually used in the spectral fitting, resulting in relatively large differences for some rows.
The table below summarizes the differences between NRT and standard SNPP/OMPS PCA SO2 products.
| Parameter | Max Difference (DU)* | Mean Difference (DU)** |
|---|
| SO2_PBL (total column amount assuming all SO2 near the surface) | 1.3 | 0.027 |
| SO2_TRL (total column amount assuming 3-km SO2 plume height) | 1.2 | 0.022 |
| SO2_TRM (total column amount assuming 8-km SO2 plume height) | 0.6 | 0.013 |
| SO2_TRU (total column amount assuming 13-km SO2 plume height) | 0.4 | <0.01 |
| SO2_STL (total column amount assuming 18-km SO2 plume height) | 0.3 | <0.01 |
*The maximum of the absolute difference between NRT and standard products for one day (10/09/2017).
*The mean of the absolute difference between NRT and standard products for one day (10/09/2017).
- VIIRS Land NRT
In order to generate data products within 3 hours of observation time, a number of changes have been made to standard processing approach to expedite the availability of input datasets:
- NRT processing uses session-based Level 0 files acquired at the end of each downlink session within 10-30 minutes of real-time, delivered by the EOS Data and Operations System (EDOS), whereas standard forward processing acquires 2-hour Level 0 files (time based) within 7-8 hours of real-time
- For geolocation, NRT uses definitive orbit information similar to the operational processing. However, there could be still minor difference in the geolocation due to difference in certain other ancillary products used by the geolocation processing such as polar wander files. Error introduced by use of such difference ancillaries is minor, but in some situations this position error may be large (several kilometers). In particular, the larger errors may occur after platform maneuvers. Impacted time period may vary from ~30 minutes in case of a lunar calibration roll maneuver to ~ 2 hours in the case of an inclination maneuver. In these cases users are advised to use the standard product generated using the best available platform position data and after the quality assurance team has reviewed and removed data degraded by the platform maneuver
- Level 1B and most Level 2 products use the same algorithm and look up tables (LUTs) as used by the operational processing except for a difference in some ancillary. For example land cloud mask processing might use NISE data from a prior data day if the data are not available for the current day processing resulting in minor difference in land cloud mask product and downstreams. Similarly, there could be significant difference in retrieved LST (as large as 5 deg K) from the use of NCEP at NRT as opposed to the MERRA-2 used by the standard processing for water vapor input
- For Level 3, rolling products download this PDF to explain the differences between the NRT and standard "n-day" products
VIIRS C1 Land (Suomi)
VIIRS C1 LST&E (VNP21)
VIIRS C2 Land (JPSS1)
VIIRS C2 VI (EVI/NDVI)
- VIIRS Atmosphere NRT
In order to generate NRT VIIRS Atmosphere products within 3 hours of observation time, several changes have been made to the standard processing approach. These differences are outlined as follows.
NRT processing utilizes session-based Level 0 files ingested by the Atmosphere SIPS from the EOS Data and Operations System (EDOS) available within 1 to 2 hours of real-time. Standard forward stream processing uses time-based Level 0 files ingested within 5 to 7 hours of real-time.
- For geolocation, NRT uses definitive orbit information similar to the operational processing. However, there could be still minor difference in the geolocation due to differences in ancillary products used by the geolocation processing such as polar wander files or the geolocation LUT files. Error introduced by the use of such difference ancillaries is typically minor, but in some situations this position error may be large (several kilometers). Larger errors may occur after platform maneuvers. Impacted time period may vary from ~30 minutes in case of lunar calibration roll maneuver to ~ 2 hours in the case of an inclination maneuver. In these cases, users are advised to use the standard product generated using the best available platform position data and after the quality assurance team has reviewed and removed data degraded by a platform maneuver
- The Standard Forward Stream Processing Mode creates the best possible global VIIRS Level 1B and Atmosphere products using the highest fidelity input data sources, including (but not limited to) retransmitted Level 0 data from EDOS; calibration LUTs from offline analysis by the NASA VIIRS Characterization Support Team; and final versions of ancillary data from external providers (e.g. LAADS, UCAR, NCDC, etc.). The philosophy for these products is to wait until the best input data are available, and then create the best possible VIIRS Level 1B, VIIRS Atmosphere Level 2, and Level 3 products
- The NRT Processing Mode creates low latency global VIIRS Level 1B and Atmosphere Level 2 products, within the constraints imposed by the time delay in receiving the session-based VIIRS Level 0 files. The philosophy for the NRT products is to provide rapid turnaround to Science Team members and other investigators who require the VIIRS Atmosphere NRT products. The VIIRS Level 1B/Level 2 products are created using the same or similar algorithms used to create the Standard Forward Stream Mode products, however the calibration LUTs and ancillary data available at the time the Level 0 data are received are used. For NRT production, ancillary input products (e.g. NISE) from an earlier time period, and sometime even several days prior, may be used, while standard mode operational processing may be delayed until the best ancillary data are available
- For example, the VIIRS MVCM Cloud Mask NRT product uses the NOAA Global Forecast System (GFS) ancillary product as opposed to the Global Data Assimilation System (GDAS) product used in the standard processing version
- The VIIRS Deep Blue Aerosol NRT product utilizes the Goddard Earth Observing System (GEOS-5) model forecasts instead of the reanalysis product. In addition, the Deep Blue product implements a different QA flag (QA flag Ocean = 2) for the NRT product associated with the effects of GEOS-5 wind speed on AOD retrievals over oceans in regions near sunglint
