If you are unfamiliar with the subtleties of the IJIS (now ADS) data, there is one point you should be aware of. The values appearing on the ADS site (and on their downloadable .csv files) are actually 2-day rolling averages. Hence, the nominal (i.e displayed) value for 12th June is, in fact, the arithmetic average of the measured values taken for the 11th and 12th.
I don't think that's how it works.
The AMSR-E orbital period is 99 minutes, and only part of the region is imaged on each pass. When they say they calculate extent based on 2 days' data, it means they calculate it by aggregating data from all the satellite passes in the preceding 48 hours. Within this time frame, some pixels will have been imaged many times, others only a few times.
It's not as simple as imaging everything once a day and then averaging two days together. It also means you CAN'T work backwards and try and reverse-derive daily values from the provided data set.
Hi Peter,
I was responding to a query by arctic-watcher at #4349 as to the provenance of the ADS (aka IJIS) data. As that was only the 5th posting by arctic-watcher, and following the old adage about
"don't try to run before you can walk", I tried to answer at what I considered to be the most appropriate level.
I fully appreciate that orbital periods generally don't line up with any 24 hour period, and that the coverage swathe paths most certainly do not tessellate, but rather overlap in a seriously convoluted fashion.
However, please try the following thought experiment...
The data from (say) Hour 0 to Hour 48 gets aggregated and presented as the extent for (say) Day X. The following day, the data from Hour 24 to Hour 72 gets aggregated and is presented as the extent for Day (X+1). The data garnered during Hour 24 to Hour 48 is common to both days, and therefore has no impact on the delta between Day X and Day (X+1)
The only thing that differs between the two days is that the data from Hour 0 to Hour 24 has been replaced by the corresponding data from Hour 48 to Hour 72.
Surely this is analogous to what happens on the NSIDC rolling 5-day
Charctic? The extent there on the 11th June was 11.297 million sq kms, and this dropped the following day to 11.219 million sq kms. Consequently, we have a delta of 78 thousand sq kms between the two days. The June 11
Charctic value is the arithmetic average of the single-day values obtained between June 7 and June 11, whereas the June 12
Charctic is the average of the June 8 to June 12 dailies. The daily values from June 8 to June 11 are common to both, and the difference is purely due to the replacement of the June 7 daily with the June 12 daily.
Moving to the NSIDC single day values, June 7th came in at 11.455 million sq kms, with the June 12th equivalent having dropped to 11.069 million sq kms. The delta is now 386 thousand sq kms, which, allowing for rounding error, is 5 times the difference between the June 11 and June 12 rolling 5-day averages.
Working backwards from the
Charctic numbers, although it is not feasible to reverse-engineer the precise one day values
(unless given a seed value to start with), it is a trivial exercise to calculate what the delta between (say) Day Z and Day (Z+5) must have been.
If the ADS data was presented in two formats - as a single day value and as an aggregated 2-day value - then the difference between Day X and Day (X+1) on the aggregated 2-day format would be one half of the difference between Day (X-1) and Day (X+1) on the single day format.