In the early morning hours of November 2, 2011, debris from a two-car accident on 1-35 in Lenexa caused authorities to shut the interstate to traffic. While Kansas Highway Patrol Trooper Nicholas Wright was working the accident, a car driven by Heather Miller came toward hirn—it was the only car on a roadway that Wright thought was still closed to vehicle traffic. So he stopped Miller for the apparent violation of a state law requiring drivers to comply with traffic-control directions.
It turned out that the officers who Wright thought were still blocking the roadway had been diverted briefly to another spot, and Miller simply drove onto the interstate without disregarding any officers, emergency flares, or traffic cones. Unfortunately for her, once the trooper stopped her car, he quickly suspected she was under the influence of alcohol—a suspicion that was confirmed through field-sobriety testing and an evidentiary breath test.
Miller argued in the district court that the evidence against her should be suppressed because she had done nothing wrong by driving onto the scene, thus she shouldn’t have been stopped in the first place. The district court agreed. Though the court said the officer was “honestly mistaken” and showed no bad faith in stopping Miller, the court concluded that the officer nonetheless was mistaken and shouldn’t have made the stop. The court suppressed the evidence found after the stop, including the evidence that Miller was driving while intoxicated.
The State has appealed, contending that an officer’s action isn’t invalidated for a mistake of fact so long as the officer had a reasonable belief that facts existed that would have justified the traffic stop. We agree.
On appeal from a district court’s order on a motion to suppress evidence, we accept the district court’s factual findings if they are supported by substantial evidence. But we independently review *493its legal conclusions, without any required deference. State v. Sanchez-Loredo, 294 Kan. 50, 54, 272 P.3d 34 (2012). Here, other than saying that the officer was “honestly mistaken,” the district court really made no specific factual findings. We will discuss the basic facts of the case, which do not appear to be in dispute, and then consider independently whether the evidence should have been suppressed.
Trooper Wright was the officer in charge at the accident scene, and he had directed closure of the interstate while the accident was investigated since there was debris scattered over the roadway. He had personnel from the Kansas Department of Transportation block traffic by forcing southbound vehicles to exit at 95th Street, and he had Lenexa police officers block the 95th Street entrance ramp going onto the interstate southbound. Wright observed those roadblocks in place. While working at the scene for 1 hour and 20 minutes, Wright saw no unauthorized vehicles come through.
He then saw Miller s car coming toward him. His understanding of police procedure was that the interstate would remain blocked until he—as the officer in charge of the investigation—ordered it to be reopened. Accordingly, he initially thought the vehicle was a police or Department of Transportation vehicle, but he testified that he was “shocked” to see that it was a private vehicle. Wright said he believed the driver either drove around traffic cones that were blocking the highway or drove around the Lenexa police officers who were blocking the entrance ramp. So he initiated a traffic stop.
Miller seemed confused when Wright asked her what she was doing. That reaction was understandable since Miller said there hadn’t been any police cars blocking the entrance ramp. And that proved to be true—Lenexa officers later came to the scene and told Wright that they’d left their post on the entrance ramp briefly because they had to deal with a disturbance involving a trucker.
By that time, however, Wright had begun to suspect that Miller was driving while intoxicated. He arrested her after she performed poorly on some field-sobriety tests, and her evidentiary breath test resulted in a .122 reading, well above the legal limit of .08.
*494So where does this leave us legally? Based upon what Trooper Wright knew at the time, he believed that Miller had committed a traffic infraction. It’s a violation of K.S.A. 8-1503 to refuse to comply with an officer’s traffic-control directives, and an officer may conduct a traffic stop where an officer knows of specific and artic-ulable facts—facts that can be clearly expressed—that create a reasonable suspicion that a person has violated a traffic law. See State v. Coleman, 292 Kan. 813, 817-18, 257 P.3d 320 (2011); State v. Marx, 289 Kan. 657, 661-62, 215 P.3d 601 (2009).
The State argues that Wright gave specific facts supporting reasonable suspicion—a particularized and objective basis for suspecting that the person stopped is involved in criminal activity. See Coleman, 292 Kan. at 817-18. Indeed, Wright identified several facts supporting his belief that Miller had disobeyed an officer’s traffic instructions:
• Wright had directed that the interstate highway be closed to traffic while he carried out the accident investigation and cleanup.
• He understood that die standard policy was to keep the road closed until the officer in charge of the investigation—in this case, himself—had directed the road to be reopened.
• He had seen that police officers had taken positions blocking the entrance ramp and drat otiier personnel had placed cones directing traffic already on the interstate to exit before the accident site.
• Other than emergency and law-enforcement vehicles, Miller’s vehicle was the first—and only—car he had seen driving on tire closed section of tíre interstate in more than an hour, and he had not reopened the roadway to traffic.
Those facts, specific and articulated by Wright, provided reasonable suspicion that Miller had violated K.S.A. 8-1503, thus justifying a traffic stop.
Of course, we now know that Wright’s overall conclusion—that Miller had disobeyed an officer’s instructions—was wrong. But drat doesn’t negate reasonable suspicion. So long as the officer operates in good faith, a reasonable mistake of fact can still provide the *495reasonable suspicion required to make a traffic stop. See Houston v. Clark County Sheriff Deputy John Does, 174 F.3d 809, 813 (6th Cir. 1999); United States v. Ornelas-Ledesma, 16 F.3d 714, 718 (7th Cir. 1994), vacated on other grounds 517 U.S. 690, 116 S. Ct. 1657, 134 L. Ed. 2d 911 [1996]); United States v. Bailey, 417 F.3d 873, 877 (8th Cir. 2005); United States v. Shareef, 100 F.3d 1491, 1505-06 (10th Cir. 1996); State v. Trammell, No. 1 CA-CR 10-0489, 2011 WL 2306697, at *3 (Ariz. App. 2011) (unpublished opinion).
We should point out, however, that an officer can make one type of mistake that won’t provide reasonable suspicion, and that’s a mistake of law. Officers are charged with knowledge of the law, and they don’t have reasonable suspicion to stop someone when they mistakenly believe that there’s a law that a person’s actions would have violated when, in fact, such a law doesn’t exist. See United States v. Chanthasouxat, 342 F.3d 1271, 1279-80 (11th Cir. 2003); United States v. Ramstad, 219 F.3d 1263, 1267 (10th Cir. 2000).
As would be expected, since the principles at issue come from cases interpreting the United States Constitution, Kansas caselaw is in in line with these cases from other jurisdictions.
Our Supreme Court made the point in Martin v. Kansas Dept. of Revenue, 285 Kan. 625, 638-39, 176 P.3d 938 (2008), that a mistake of law can negate reasonable suspicion, noting that a police officer must be held to a more demanding standard of legal knowledge than are tire citizens who are subject to the officer’s authority. In Martin, an officer stopped a vehicle because one of its three rear-brake lights wasn’t working. But it turned out that the law only required two working rear-brake lights, so the officer made a legal error in thinking that the vehicle didn’t comply with legal requirements. Accordingly, the court found that the officer had no authority to stop the vehicle; there was no traffic infraction, and the officer’s conclusion that there was one was based on a mistake of law.
A similar situation arose in State v. Knight, 33 Kan. App. 2d 325, 104 P.3d 403 (2004). There, a police officer stopped a vehicle that turned from a grocery-store parking lot onto a public street without *496using a turn signal. The officer mistakenly believed that city traffic ordinances required the use of a turn signal, but our court determined that the ordinance had no requirements for signals when turning from a private driveway or parking lot onto a public street. Since diere was no traffic violation—and the officer s suspicion of a traffic violation was based on a mistake about what the law was, not what the officer had factually observed—our court ruled tíiat the stop was unlawful.
Though we are not aware of a specific Kansas case noting that an officer can be factually in error and still have reasonable suspicion, that interpretation is consistent with another tenet of reasonable-suspicion analysis: In determining whether reasonable suspicion existed, a court must consider all the circumstances that existed when the officer made the stop and may not eliminate factors that might have an innocent explanation. See Coleman, 292 Kan. at 817-18. That’s true even when, as here, an officer later learns that diere was an innocent explanation.
Here, Trooper Wright made a mistake of fact, not one of law. Wright reasonably believed—factually—that officers were still in place on die entrance ramp, blocking cars from entering and directing them elsewhere. Wright wasn’t mistaken that it would have been illegal for Miller to go around those officers; he was mistaken factually that those officers were still in place. Accordingly, the traffic stop was based on reasonable suspicion of a traffic infraction, and the district court should not have suppressed the evidence based on Miller’s claim that the stop was improper.
Miller makes one final argument in an attempt to avoid this result. Citing State v. Niblock, 230 Kan. 156, 161, 631 P.2d 661 (1981), she notes that the knowledge of one officer may be imputed to others—meaning that if one officer knows something, odier officers should also be assumed to have knowledge of it. Thus, Miller argues, since the police officers who left their roadblock position knew they had done so, Trooper Wright should be deemed to have known tiiat too. With that imputed knowledge, Wright would have had no reason to stop Miller’s car.
The rule for imputing knowledge from one officer to another is known variously, as the collective-knowledge doctrine or the fellow-*497officer rule. Under the doctrine, one officer may stop, search, or arrest a suspect at the direction of another officer or police agency, even if the officer does not have firsthand knowledge of the facts that would provide the necessary level of suspicion to permit the given action. United States v. Williams, 627 F.3d 247, 252 (7th Cir. 2010). In Williams, the Seventh Circuit stated a three-part test for the application of this doctrine, starting with something that didn’t occur in Miller’s case—transmission of information from officers away from the scene to the officer who takes the action in question:
“In order for the collective knowledge doctrine to apply, (1) the officer taking the action must act in objective reliance on the information received, (2) the officer providing the information—or the agency for which he works—must have facts supporting the level of suspicion required, and (3) the stop must be no more intrusive that would have been permissible for the officer requesting it.” 627 F.3d at 252-53.
The collective-knowledge doctrine applies when one officer acts based on the instructions from another officer. Under the doctrine, the instructing officer need not share all of the underlying facts that form the basis for the instruction; it’s enough that some officer has that information. Thus, in the Niblock case that Miller cites, there was no problem where one police officer stopped a robbery suspect when the arresting officer had no personal knowledge of the facts and had merely received a description of the suspect vehicle and driver from a police dispatcher over the radio.
But in Niblock, there was communication from one officer to another, and under tire collective-knowledge doctrine, the officer taking action must have acted in objective reliance on some information received from another. Williams, 627 F.3d at 252-53. Here, the officers at the roadblock hadn’t communicated any information—directly or indirectly—to Wright about their having left their post before Wright stopped Miller’s car. There is no basis to apply the collective-knowledge doctrine here. See United States v. Lyons, 687 F.3d 754, 767 (6th Cir. 2012) (approving Seventh Circuit’s three-part test for application of collective-knowledge doctrine); United States v. Massenburg, 654 F.3d 480, 492-96 (4th Cir. 2011) (limiting collective-knowledge doctrine to situations in which one *498officer acts on information and instructions received from other officers).
The district court’s judgment is therefore reversed, and the case is remanded for further proceedings.