It looks like sloppy graph drawing to me. I think the data is based on monthly increments, and the graph drawer made a notional effort to do some curve fitting, but didn't work very hard at it.
Also don't forget that the lines plotted over these 'average bands' are themselves averages. The weekly EIA reports use 4 week averages to report all numbers.

Averages can hide things... when traders react to X being over or under the 5 year average, its often without benefit of realizing that the "average band" would be higher if a smaller averaging period were used. 2001 was an anomoly year... and demand has been very strong over the past 2 years:

1.2% 2000 (for comparison)
0.7% 2001
1.0% 2002
1.6% 2003
3.1% 2004
3.8% (2005 est, it will be lower, but still high)

Stocks must be near the high end of the 5 year range or we are in trouble!

MMS Status - oil shut out edged up a little higher from 1,527,630 BOPD. NG shut in down ever so slightly. Now 5 days in, almost virtual GOM total shut down.

---
Today's shut-in oil production is 1,512,937 BOPD.  This shut-in oil production is equivalent to 100% of the daily oil production in the GOM, which is currently approximately 1.5 million BOPD.

Today's shut-in gas production is 7.856 BCFPD.  This shut-in gas production is equivalent to 78.56% of the daily gas production in the GOM, which is currently approximately 10 BCFPD.

The cumulative shut-in oil production for the period 8/26/05-9/27/05 is 36,361,383 bbls, which is equivalent to 6.641 % of the yearly production of oil in the GOM (approximately 547.5 million barrels).

The cumulative shut-in gas production 8/26/05-9/27/05 is 172.506 BCF, which is equivalent to 4.726% of the yearly production of gas in the GOM (approximately 3.65 TCF).

http://www.mms.gov/ooc/press/2005/press0927.htm