Near Real-Time Products
Near Real Time (NRT) processing system, an element of the MODIS Adaptive Processing System (MODAPS) generates near science quality products meeting the timely needs of real time users such as numerical weather and climate prediction, forecasting and monitoring natural hazards, agriculture, air quality and disaster relief. Most L2 products are generated within 2-3 hours of satellite observations, and gridded daily products within few hours after the end of the day. NRT system uses the science process used by MODAPS to generate the standard products with minor changes to enable processing of data using ancillaries or attitude information different from the standard processing to meet the NRT latency requirements. LDOPE assesses the quality of the product prior to start of NRT processing of the product and also monitors the quality of the product from routine operational processing.
Differences between the NRT and Standard "n-day" products
The NASA MODIS and VIIRS "n-day", L3 land products produced by MODAPS and Land SIPS follow a slightly different processing paradigm in the NRT and Standard (STD) processing streams.
NRT -Following the requirements of quick turn around and minimum latency in the NRT streams, most n-day NRT products like MOD13A1N/Q1N, MxD09A1N, MCD43A1N/A2N/A2N are produced every day based on a rolling prior "(n-1)-day" input requirements, with "nth" day being the current day being processed. Hence, the product for day "t" will use inputs from day t-(n-1) to t. Thus the NRT stream follows an essentially "look back" approach and data day given in the product file name represents the current day or the last day (nthday) of processing. As an example, the 8-day MOD09A1N product from day 8, will have inputs from day 8 and 7 days prior and the filename will be something like: MOD09A1N.2018008.006*.hdf
STD -Contrary to this, in the STD processing streams, the products are generated based on fixed and pre-defined "n-day" intervals and follow a "look-ahead" approach, where a n-day product is produced, with inputs from day t to t+(n-1),with "t" being the start of the n-day period and "t+(n-1)" being the end of the n-day period. As an example, the 8-day MODIS Terra LSR product, MOD09A1, from the 8-day period starting on day 001, will have inputs from day 001 –008 and will have the starting day "001" indicated as part of the filename, like: MOD09A1.A2018001.006.*.hdf.
This general difference between the NRT and STD n-day products, becomes slightly different in case of the MODIS BRDF Albedo product suite (MCD43*).
- NRT BRDF process estimates BRDF for the current data day at the leading edge of the 16-day period.
- STD BRDF process estimates the BRDF for the 9th day in the 16-day period. Hence, operational BRDF processing will be lagging NRT BRDF by at least 8-day.
- The weighting scheme for the different days are also slightly different between NRT and STD. In NRT, greater weightage is given to immediately prior days with the weights dropping linearly with more distant days. In STD, equal weights are given to days on both sides of the data days with the weights dropping in proportion to the distance of the days on either side from the data day.
