The SASC Report on Torture strongly suggests that CIA was following one set of guidelines on waterboarding, but had gotten approval from DOJ for another set of guidelines. 

The SASC reveals that on July 26, 2002, JPRA sent a report on SERE techniques. That report is almost certainly one of the resources Jay Bybee consulted when writing his memo, which was published on August 1, 2002. The SASC report says,

[DOD General Counsel] Haynes also recalled that he may have been "asked that information be given to the Justice Department for something they were working on," which he said related to a program he was not free to discuss with the Committee, even in a classified setting

See Valtin’s story showing that the data came from the same place.

Now, as SASC describes it, the JPRA document didn’t describe waterboarding as it used to be done in Navy training. 

JPRA’s description of the waterboarding technique provided in that first attachment was inconsistent in key respects from the U.S. Navy SERE school’s description of waterboarding. According to the Navy SERE school’s operating instructions, for example, while administering the technique, the Navy limited the amount of water poured on a student’s face to two pints. However, the JPRA attachment said that "up to 1.5 gallons of water" may be poured onto a "subject’s face." While the Navy’s operating instructions dictated that "[n]o effort will be made to direct the stream of water into the student’s nostrils or mouth," the description provided by JPRA contained no such limitation for subjects ofthe technique. While the Navy limited the use ofthe cloth on a student’s face to twenty seconds, the JPRA’s description said only that the cloth should remain in place for a "short period of time." And while the Navy restricted anyone from placing pressure on the chest or stomach during the administration of this technique, JPRA’s description included no such limitation for subjects of the technique. [my emphasis]

In other words, JPRA was advising waterboarding to be used in torture to use six times the amount of water as that used in training, and JPRA eliminated the 20 second limit on waterboarding.

Now, Bybee’s memo is closer to what it appears Navy did in SERE, with limits on timing (though twice as long as SERE allowed), and description of  water being poured from a "canteen cup."

Finally, you would like to use a technique called the "waterboard" in this procedure, the individual is bound securely to an inclined bench, which is approximately four feet by seven feet. The individual’s feet are generally elevated. A cloth is placed over the forehead and eyes. Water is then applied to the cloth in a controlled manner. As this is done, the cloth is lowered until it covers both the nose and mouth. Once the cloth is saturated and completely covers tbe mouth and nose, air flow is slightly restricted for 20 to 40 seconds due to the presence of the cloth. Tbis causes an increase in carbon dioxide level in the individual’s blood. This increase in the carbon dioxide level stimulates increased effort to breathe. This effort plus the cloth produces the perception of"suffocalion and incipient panic," i.e., the perception of drowning. The individual does not breathe any water imo his lungs. During those 20 to 40 seconds, water is continuously applied from a height of twelve to twenty-four inches. Afler this period, the cloth is lifted, and the individual is allowed to breathe unimpeded for three or four full breaths. The sensation of drowning is immediately relieved by the removal of the cloth. The procedure may then be repeated. The water is usually applied from a canteen cup or small watering can with a spout.

But, as we know, that’s not how waterboarding was done in practice.

The IG Report noted that in some cases the waterboard was used with far greater frequency than initially indicated, see IG Report at 5, 44, 46, 103-04, and also that it was used in a different manner. See id. at 37 ("[T]he waterboard technique  … was different from the technique described in the DoJ opinion and used in the SERE training. The difference was the manner in which the detainee’s breathing was obstructed. At the SERE school and in the DoJ opinion, the subject’s airflow is disrupted by the firm application of a damp cloth over the air passages; the interrogator applies a small amount of water to the cloth in a controlled manner. By contrast, the Agency Interrogator …  applied large volumes of water to a cloth that covered the detainee’s mouth and nose. One of the psychologists/interrogators acknowledged that the Agency’s use of the technique is different from that used in SERE training because it is "for real–and is more poignant and convincing.") [my emphasis] 

They got approval for SERE techniques. But they had already formally decided to far exceed the guidelines used in SERE.