OPINION

delivered the opinion of the Court,

A police officer stopped appellee, Vanessa Mendoza, and arrested her for driving while intoxicated. The trial judge granted appellee’s motion to suppress, concluding that the officer lacked reasonable suspicion to make his traffic stop. The State appealed that ruling and requested written findings of fact and conclusions of law. Based upon the judge’s written factual findings, the court of appeals inferred that the judge must have believed the officer’s testimony — testimony that was in some conflict with the dash-cam video recording. The court of appeals reasoned that, if the trial judge fully credited the officer’s testimony, then her legal conclusion that the officer did not have reasonable suspicion to stop appellee’s car was incorrect. ^[1] We granted appellee’s petition to review whether the court of appeals deferred sufficiently to the trial court’s factual findings. ^[2] We agree with appellee that reviewing courts should not make implied findings of fact and credibility determinations that are contrary to the trial judge’s ultimate ruling. But because the written findings in this case are ambiguous and there is no credibility determination, we remand this case to the court of appeals with instructions to abate the case to the trial judge for supplemental findings.

I.

The trial judge held an evidentiary hearing on appellee’s motion to suppress, in which Officer Davila was the only witness. He testified that he saw Ms. Mendoza’s car “ahead of [him] at a distance [and she] appeared to be traveling at a high rate of speed.” He started to pace her car and turned on his dash-cam video recorder, but by then she had slowed down and maintained her speed below the posted speed limit. Officer Davila followed Ms. Mendoza’s car — driving downhill and around curves — from about the 5500 block of[*668] Mesa Street to the 4100 block of Mesa Street. He testified that Ms. Mendoza weaved within her lane, braked as she went downhill around the curves, and, at one point, veered a “little bit” towards the right, where another car was traveling several car lengths ahead. On cross-examination, the officer agreed that Ms. Mendoza had not committed any traffic offense and that his purpose in stopping her was to do “a safety check on her to make sure she was okay.” The trial judge watched the dash-cam video of the incident as Officer Davila testified that nothing on that video showed Ms. Mendoza commit any traffic offense. ^[3] After taking the matter under advisement, the trial judge granted Ms. Mendoza’s motion and filed the following findings of fact and conclusions of law:

FINDINGS OF FACT

1. On January 29th, 2008, Officer Enrique Davila observed the defendant, Vanessa M. Mendoza, driving southbound around the 5500 or 5100 block of Mesa Street at about 2:28 a.m.

2. Officer Davila believed the defendant was going at a high rate of speed but did not ascertain the vehicle’s speed by radar or by pacing the vehicle.

3. Officer Davila noticed the defendant’s vehicle weaving within its lane a few times, the defendant was traveling on the far left lane of Mesa Street.

4. Officer Davila stated that the defendant’s car would continuously slow down and brake. The video depicts Mesa Street as a roadway going downhill with several curves.

5. When approaching the intersection of Executive Center Blvd. and Mesa Street, Officer Davila felt that the defendant’s car was going to hit another vehicle because it veered to the right. There was no testimony indicating if the defendant’s car weaved out of her lane. There was no testimony about the proximity of the other car. The video shows a vehicle about two or three car lengths in front of defendant’s vehicle traveling on the far right lane.

6. Officer Davila testified that he stopped the defendant because she weaved within a lane a few times, veered to the right and braked erratically.

CONCLUSIONS OF LAW

The Court finds that the facts of this case, (as seen on the video and the testimony of Officer Davila) did not justify a stop of the defendant’s vehicle and do not provide the police officer with sufficient reasonable suspicion to believe that the defendant was intoxicated.

The court of appeals held that the trial judge’s legal conclusion did not follow from her factual findings, because it inferred that the trial judge’s explicit findings created an implied finding that she believed the officer’s version of events. ^[4] After assuming that the trial judge must have believed the officer, the appeals court found that the State’s version of events constituted reasonable suspicion. ^[5] If the trial judge fully credited all of Officer Da-vila’s testimony and the logical inferences[*669] that flow from his testimony, then the officer did have reasonable suspicion to stop Ms. Mendoza’s car. However, in both her factual findings and legal conclusion, the trial judge also referred to the video recording, and that video is not entirely consistent with Officer Davila’s testimony. The problem with these factual findings is that they are ambiguous and do not contain any credibility determinations.

II.

In a motion to suppress hearing, the trial judge is the sole trier of fact and judge of the weight and credibility of the evidence. ^[6] In reviewing the ruling on a motion to suppress, appellate courts must give almost total deference to a trial judge’s findings of historical fact and credibility determinations. ^[7] First, “the trial judge is ‘Johnny-on-the-Spot,’ personally able to see and hear the witnesses testify.” ^[8] Second, “[t]he trial judge’s major role is the determination of fact, and with experience in fulfilling that role comes expertise.” ^[9] Third, appellate courts that duplicate the trial judge’s factfinding role likely contribute only marginally to accuracy while diverting large amounts of scarce judicial resources. ^[10] And, finally,

the parties to a case on appeal have already been forced to concentrate their energies and resources on persuading the trial judge that their account of the facts is the correct one; requiring them to persuade three more judges at the appellate level is requiring too much.... [T]he trial on the merits should be the “main event” ... rather than a “tryout on the road.” ^[11]

A suppression ruling includes two types of trial-judge rulings; historical factual findings, often based on credibility determinations ^[12] (subject to abuse-of-discretion review), and ultimate legal rulings that determine whether reasonable suspicion or probable cause existed (subject to de novo review). The Supreme Court has explained that

[t]he principal components of a determination of reasonable suspicion or proba[*670] ble cause will be the events which occurred leading up to the stop or search, and then the decision whether these historical facts, viewed from the standpoint of an objectively reasonable police officer, amount to reasonable suspicion or to probable cause. The first part of the analysis involves only a determination of historical facts, but the second is a mixed question of law and fact ... [that] should be reviewed de novo on appeal. Having said this, we hasten to point out that a reviewing court should take care both to review findings of historical fact only for clear error and to give due weight to inferences drawn from those facts by [trial] judgesf.] ^[13]

But sometimes it is not clear precisely what historical facts a trial judge actually found or what her credibility determinations were. Because “there is no justification for us to ‘assume that the trial court made implicit findings of fact that support its ruling,’ ” we held, in State v. Cullen, ^[14] that a trial judge, “upon the request of the losing party ... must make [on-the-record] findings of fact and conclusions of law adequate to provide an appellate court” a basis for reviewing the ruling’s correctness. ^[15] Thus, the losing party in a motion to suppress hearing may require the trial judge to make explicit historical findings of fact and credibility determinations. These may be either written out and signed by the judge or dictated on the record. ^[16] This rule has the salutary effect of ensuring that the appellate courts will grant almost total deference to the trial judge’s findings of historical fact while reviewing de novo the application of the law concerning probable cause or reasonable suspicion to those historical facts. ^[17]

Occasionally, the trial judge may make explicit findings that she considers sufficient and dispositive of the historical facts, but the appellate court determines that those findings are either ambiguous or insufficient to resolve the legal issue. For example, in State v. Elias, ^[18] the trial judge made written findings of fact and conclusions of law concerning a motion to suppress, but those findings did not directly address the potentially “dispositive” historical fact of whether the defendant “activated his turn signal within a hundred feet of the intersection.” ^[19] One way of dealing with insufficient factual findings might be to revert to our former Ross 20 presumption that the trial judge made all possible implied findings of fact that are supported by the record and consistent with her ultimate legal ruling. ^[21] But, in Elias, this Court decided that the more prudent course would be to remand the case to the trial judge to make findings of fact with greater specificity if the original findings are insufficient to resolve the legal question. ^[22] In Elias, this Court continued the[*671] route it charted in Cullen to ensure that reviewing courts need not presume, assume, or guess at what historical facts a trial judge actually found when making a ruling in a motion to suppress hearing. In both Cullen and Elias, this Court followed Judge Womack’s admonition in Ross to take steps “toward the just resolution of these appeals — a resolution that is based on the reality of what happened rather than on assumptions that may be entirely fictitious.” ^[23]

III.

In this case, the court of appeals began with the premise that the trial judge “implicitly accepted] the State’s version of events and [found] the detaining officer’s testimony credible,” thus “the only question remaining for this Court is to determine whether the trial court properly applied the law to those facts.” ^[24] If the trial judge fully credited Officer Davila’s testimony that appellee was “weaving within the lane, braking erratically, and drifting toward another vehicle,” ^[25] then the court of appeals properly concluded that the combination of these facts, “considered in the context of the time of night, the location, and Officer Davila’s training and experience investigating DWI offenses, ... provided a reasonable basis for him to conclude Appellee was engaged in criminal activity.” ^[26]

But the trial judge’s written findings contain both “weasel words” and factual juxtapositions that cast doubt on an implied finding that she fully credited Officer Davila’s version of events.

First, we look at the “weasel words.” The trial judge began her factual findings with a strong verb: Officer Davila “observed” appellee driving. That is an historical fact; he saw her drive. But then the judge stated that the officer “believed” appellee was speeding; he “noticed” her weaving a few times; he “stated” that she would continuously slow down; he “felt” that appellee’s car was going to hit another car; he “testified” that he stopped appellee “because she weaved within a lane a few times, veered to the right and braked erratically.” Clearly, the officer “testified” to those facts; any reviewing court can read the record and see Officer Davila’s testimony, but did the trial judge believe that testimony?

For example, in her second factual finding, the trial judge stated, “Officer Davila believed the defendant was going at a high rate of speed[.]” If the trial judge fully credited Officer Davila’s testimony that he believed appellee was speeding, why not state the straightforward historical fact— “Officer Davila saw the defendant speeding.” Even if the trial judge concluded that Officer Davila was factually incorrect about the historical fact — that appellee was speeding — his testimony might still be credited if the judge concluded that Officer Davila “reasonably believed” that she was speeding. It is the officer’s “reasonable belief’ — one that would be held by an objectively reasonable and prudent police officer that governs the question of reasonable suspicion or probable cause. But here, the trial judge did not add the word “reasonably” to Officer Davila’s belief.

In her final factual finding, the trial judge stated that Officer Davila “felt” that appellee was going to hit another car because her car veered to the right. That is not an historical fact, it is a subjective[*672] thought. If the trial judge concluded that Officer Davila “reasonably felt” that appel-lee was going to hit another car, then that finding would provide support for a legal conclusion of reasonable suspicion, but if the trial judge concluded that Officer Davi-la did not have an objectively reasonable basis for his “feeling,” that finding does not support a reasonable suspicion.

In sum, these factual findings do not find many historical facts, nor do they find that Officer Davila’s beliefs and feelings were objectively reasonable ones. They also contain no credibility assessment of Officer Davila’s testimony.

In addition, the trial judge also used a juxtaposition of phrases and facts that creates more ambiguity in her findings. For example, in her second factual finding, the trial judge states that “Officer Davila believed the defendant was going at a high rate of speed,” but she then contrasts that phrase with the additional fact that he “did not ascertain the vehicle’s speed by radar or by pacing the vehicle.” In her fourth finding, the trial judge notes, “Officer Da-vila stated that the defendant’s car would continuously slow down and brake.” But she immediately contrasts that statement with the further fact that “the video depicts Mesa Street as a roadway going downhill with several curves.” And the video shows appellee braking as she approaches each downhill curve and then taking her foot off the brake as she moves into the curve. It is impossible to tell whether the trial judge considered appel-lee’s braking “erratic” or appropriate or whether she concluded that Officer Davi-la’s statement was an objectively reasonable one based on his perception or an objectively unreasonable one based upon her viewing of the video. And, in the fifth finding, the trial judge noted that Officer Davila “felt” that appellee’s car was going to hit another car “because it veered to the right.” But that statement is immediately followed by a contrasting statement: “There was no testimony indicating if defendant’s car weaved out of her lane. There was no testimony about the proximity of the other car. The video shows a vehicle about two to three car lengths in front of the defendant’s vehicle traveling in the far right lane.” Indeed, looking at the video, it appears that appellee’s car was two lanes away from and several car lengths behind the other car, but if the trial judge concluded that Officer Davila reasonably felt that appellee’s car did pose a danger to the other car, then that fact would support a finding of reasonable suspicion.

Finally, in her conclusion of law, the trial judge stated that “the facts of this case, (as seen on the video and the testimony of Officer Davila) did not justify” the traffic stop. But what does that mean? What actual historical facts did the trial judge find as opposed to what Officer Da-vila testified to, believed, felt, and stated? And, if she based her findings more on the video rather than Officer Davila’s testimony, it cannot necessarily be inferred that she believed Officer Davila’s testimony to be entirely accurate and credible. ^[27]

The written findings in this case could be construed to imply that the trial judge believed Officer Davila was entirely rea[*673] sonable and that she simply misapplied the law of reasonable suspicion. If that were clearly the case, the court of appeals’s application of de novo review was proper, and its ruling should stand. ^[28] However, it is at least equally possible that the trial judge made an implied credibility finding that supported her ruling, namely that she disbelieved the officer or the reasonableness of his beliefs, feelings, and statements.

However, we should not make either assumption, because either assumption could be incorrect. Just as the trial judge was best positioned to evaluate the officer’s credibility at the hearing, so she is in the best position to clarify her ambiguous factual findings and make an explicit credibility determination. Those new findings should dispel any confusion about precisely what historical facts the trial judge found and whether she found Officer Davila’s beliefs and conclusions objectively reasonable, based on the historical facts. Presumably the trial judge will then deny the motion to suppress if she accepts Officer Davila’s version of the facts because the court of appeals correctly set out the applicable legal principles. And presumably the trial judge will again grant the motion to suppress if she disbelieved Officer Davi-la’s version or finds that he was objectively unreasonable in his factual beliefs and conclusions. The losing party could once again appeal, but with more explicit factual findings and credibility assessments, we anticipate that the trial judge will apply the pertinent Fourth Amendment law as set out by the court of appeals to those facts.

We remand this case to the court of appeals with instructions to abate it to the trial court for further proceedings consistent with this opinion.