MARK T. PITTMAN, JUSTICE
Deputy Christopher Kristufek of the Parker County Sheriffs Office stopped Appellee Randall Lee Binkley in the Horseshoe Bend area solely because of an unconfirmed return from the state vehicle insurance database regarding whether Binkleys vehicle had liability insurance. Deputy Kristufek ultimately arrested Binkley for driving while intoxicated-felony repetition (DWI), a grand jury indicted him, and Binkley filed a motion to suppress all evidence gleaned from the stop. The trial court granted the motion. Neither party requested findings of fact and conclusions of law, and the trial court did not issue them.
In its interlocutory appeal, the State contends in its sole point that sufficient reasonable suspicion existed to stop [Binkleys] vehicle when the detaining deputy received information from the state vehicle insurance database that the insurance policy on said vehicle had been expired for over five months. Deferring to the trial courts resolution of conflicts in the testimony and the weight to be accorded the evidence, we affirm the trial courts order granting Binkleys motion to suppress.
I. STATEMENT OF FACTS
A. Evidence Regarding Binkleys Stop
Deputy Kristufek testified:
• He was employed as a patrol deputy on or about January 24, 2016;
• He was leaving the Horseshoe Bend area when he noticed a vehicle driving a little slow;
• He entered the license plate number into his in-car database and found out that the car was registered to Binkley;
• The database showed unconfirmed regarding the liability insurance on the vehicle Binkley was driving;
• Deputy Kristufek did not remember whether the database reported any details other than unconfirmed regarding Binkleys stop;
• Deputy Kristufek stopped Binkleys vehicle just to verify the actual insurance, if there was insurance on the vehicle;
• Deputy Kristufek noticed that the database appeared to be working correctly that day;
• The database did not give him any error messages that day;
• He had no reason to think that the database was any less reliable on Binkleys traffic stop than it was on other traffic stops; and
• After looking at the printout from Binkleys stop, Deputy Kristufek noticed that Binkleys insurance policy was effective July 1, 2015, and had expired August 1, 2015.
The printout related to Binkleys stop was admitted as States Exhibit No. 3. It indicates that the insurance status of the vehicle he was driving was Unconfirmed, that the Vehicle Coverage [had] Expired, and that the Vehicle [s] Last Match [was] Not within 45 days.
B. Evidence Regarding the Database and Its Reliability
Deputy Kristufek testified:
• In addition to ownership and registration information, the database would also provide a vehicles VIN number, the type of vehicle, the make, the model, if it has a color unknown or known-and it[ would] give the color on that-and then it[ would] give ... an insurance brief which is confirmed, unconfirmed or verify manually;
• Verify manually [means] that either the company hasnt added the insurance to the database yet or they havent uploaded or theyre just such a small company that they dont put it into the database;
• He would not stop vehicles when the return said verify manually because maybe they do have insurance, maybe they dont. But nine times out of ten if [he] ma[d]e that stop, they[ would] be like, well, heres my insurance and then the stop [would be] done;
• The database also indicated if a vehicle was current on insurance and whether any warrants were associated with the specific plate number;
• He performed a database search on vehicles every day that [he] work [ed] as a patrolman, 30 to 40 times [per] hour;
• Unconfirmed insurance is when a person pretty much doesnt pay or their insurance has fallen off within, like, 45 days or so. And itll say unconfirmed no later-or later than 45 days. I mean, just little specifics. And sometimes it wont. It just says unconfirmed;
• In his opinion, the in-car database is reliable based on
The number of vehicles that [he] look[s] through. ... [Y]oull look through 100, 200 vehicles with just license plates alone, and 99 of them will have current insurance or verify manually. There will be that one that doesnt. And at that point you-I mean, yeah, is it possible that theres two [vehicles] that could be side by side and you just pick an unconfirmed or whatever, yeah. ....;
• Sometimes youll get sporadic where youll have issues with two-one will be verify manually, [and] one would be unconfirmed [on the same vehicle];
• He speculated that maybe that error occurred because the person had just purchased the insurance policy but admitted that he did not know;
• He did not know if those who input the data had any incentive to be accurate;
• He did not know of any rates of error associated with the database;
• He did not know of any insurance company being more accurate than others;
• He did not know but assumed that people would have to go through their insurance companies to correct the information in the database;
• He had not learned anything about the database during training in his two years as a patrol officer before the stop;
• He did not know the name of the program that allowed him to have the information in his patrol car, or when it was established;
• He did not know who entered the information into the database or how often it was updated;
• He did not know of safeguards to ensure the databases accuracy;
• He did not know how the database handled self-insured people;
• DPS administered the program;
• [M]aybe three or four times a week, he would pull over a driver to investigate an unconfirmed return, and the driver would provide him with proof of insurance;
• Deputy Kristufek testified that he would stop about three or four vehicles per week;
• He testified that he would probably do about two or three ... investigations for the insurance. And a traffic stop is totally different than a traffic investigation;
• He testified that he would make three traffic investigations for unconfirmed insurance in a week;
• He testified that in one of the three cases, the person would actually have insurance;
• He conducted an estimated twenty-five traffic stops per week, but not for the unconfirmed insurance. He said that the deputies could not do a traffic stop for just unconfirmed insurance; and
• He answered affirmatively questions asking whether he had run thousands and thousands and thousands of plates while on patrol and whether he found the database reliable and accurate.
Melissa Burkhardt, coordinator of the TexasSure Vehicle Insurance Verification program for the Texas Department of Insurance (TDI), testified:
• The program matches registered vehicles to insurance policies and provides that information to ... end users with the Department of Motor Vehicles [DMV] and DPS;
• The program has been in existence since 2008;
• [T]he database is maintained by a third-party vendor thats managed by the State;
• The only third-party vendor to have maintained the database since 2008 is HDI Solutions;
• HDI Solutions also created the database;
• [A]ll insurance companies that write personal auto policies in the state of Texas are required to report to [the] vendor once a week all insurance policies that are in force;
• [T]he vendor also gets information on the registered vehicles from the ... [DMV]. They report that information once a week;
• More than 300 insurance companies report to the database;
• Its approximately 43 million records per week;
• All of that information is input into one database;
• New insurance typically takes three weeks to be reflected in the database;
• TDI has had no issues with the third-party vendor regarding information links or ongoing mistakes;
• TDI has quality control systems in place to maintain the third-party vendors quality and control;
• Data leaks would be addressed immediately;
• Four possible responses come out of the TexasSure database: [C]onfirmed, unconfirmed, not found, [and] multiple;
• But DPS is responsible for how that insurance displays to law enforcement throughout the state;
• It is her understanding that DPS uses verify manually for any returns that would otherwise be not found and multiple;
• Unconfirmed means that the vehicle was located, but either the policy hasnt been reported in more than 45 days or the policy is more than 21 days expired;
• The only reason other than lag time she could think of for the 33% error rate Deputy Kristufek experienced would be an error in the way their information was recorded by their [insurance] company;
• TDI does not keep records regarding rates of error;
• Self-insurance is maintained by DPS, and DPS reports to the vendor weekly;
• TexasSure is the only database used in Texas that matches the registration to the insurance of a private vehicle;
• The vendor sends written notice to owners of registered vehicles that go 90 days without an insurance match that the database does not show that they have insurance;
• In her opinion, the database is reliable in spite of the 33% error rate experienced by Deputy Kristufek;
• Since its inception, the database has been unavailable[,] ... down [,] or broken for five seconds here or there, but nothing significant;
• She had no independent recollection that the database was not accessible on January 24, 2016; and
• She knew nothing about Binkleys case.
II. MOTION TO SUPPRESS
A. Standard of Review
We review a trial courts ruling on a motion to suppress evidence under a bifurcated standard of review. Amador v. State , 221 S.W.3d 666, 673 (Tex. Crim. App. 2007) ; Guzman v. State , 955 S.W.2d 85, 89 (Tex. Crim. App. 1997). In reviewing the trial courts decision, we do not engage in our own factual review. Romero v. State , 800 S.W.2d 539, 543 (Tex. Crim. App. 1990) ; Best v. State , 118 S.W.3d 857, 861 (Tex. App.-Fort Worth 2003, no pet.). The trial judge is the sole trier of fact and judge of the witnesses credibility and the weight to be given their testimony. Wiede v. State , 214 S.W.3d 17, 24-25 (Tex. Crim. App. 2007) ; State v. Ross , 32 S.W.3d 853, 855 (Tex. Crim. App. 2000), modified on other grounds by State v. Cullen , 195 S.W.3d 696 (Tex. Crim. App. 2006). Therefore, we give almost total deference to the trial courts rulings on (1) questions of historical fact, even if the trial courts determination of those facts was not based on an evaluation of credibility and demeanor, and (2) application-of-law-to-fact questions that turn on an evaluation of credibility and demeanor. Amador , 221 S.W.3d at 673 ; Montanez v. State , 195 S.W.3d 101, 108-09 (Tex. Crim. App. 2006) ; Johnson v. State , 68 S.W.3d 644, 652-53 (Tex. Crim. App. 2002). But when application-of-law-to-fact questions do not turn on the credibility and demeanor of the witnesses, we review the trial courts rulings on those questions de novo. Amador , 221 S.W.3d at 673 ; Estrada v. State , 154 S.W.3d 604, 607 (Tex. Crim. App. 2005) ; Johnson , 68 S.W.3d at 652-53.
Contrary to the States assertions, material facts of this case are in dispute; we therefore view the evidence in the light most favorable to the trial courts ruling. See Wiede , 214 S.W.3d at 24 ; State v. Kelly , 204 S.W.3d 808, 818 (Tex. Crim. App. 2006). When the record is silent on the reasons for the trial courts ruling, or when there are no explicit fact findings and neither party timely requested findings and conclusions from the trial court, we imply the necessary fact findings that would support the trial courts ruling if the evidence, viewed in the light most favorable to the trial courts ruling, supports those findings. State v. Garcia-Cantu , 253 S.W.3d 236, 241 (Tex. Crim. App. 2008) ; see Wiede , 214 S.W.3d at 25. We then review the trial courts legal ruling de novo unless the implied fact findings supported by the record are also dispositive of the legal ruling. Kelly , 204 S.W.3d at 819.
We must uphold the trial courts ruling if it is supported by the record and correct under any theory of law applicable to the case even if the trial court gave the wrong reason for its ruling. State v. Stevens , 235 S.W.3d 736, 740 (Tex. Crim. App. 2007) ; Armendariz v. State , 123 S.W.3d 401, 404 (Tex. Crim. App. 2003), cert. denied , 541 U.S. 974, 124 S.Ct. 1883, 158 L.Ed.2d 469 (2004).
B. Reasonable Suspicion
A detention, as opposed to an arrest, may be justified on less than probable cause if a person is reasonably suspected of criminal activity based on specific, articulable facts. Terry v. Ohio , 392 U.S. 1, 21, 88 S.Ct. 1868, 1880, 20 L.Ed.2d 889 (1968) ; Carmouche v. State , 10 S.W.3d 323, 328 (Tex. Crim. App. 2000). An officer conducts a lawful temporary detention when he or she has reasonable suspicion to believe that an individual is violating the law. Crain v. State , 315 S.W.3d 43, 52 (Tex. Crim. App. 2010) ; Ford v. State , 158 S.W.3d 488, 492 (Tex. Crim. App. 2005). Reasonable suspicion exists when, based on the totality of the circumstances, the officer has specific, articulable facts that when combined with rational inferences from those facts, would lead him to reasonably conclude that a particular person is, has been, or soon will be engaged in criminal activity. Ford , 158 S.W.3d at 492. This is an objective standard that disregards any subjective intent of the officer making the stop and looks solely to whether an objective basis for the stop exists. Id.
Reasonable suspicion is a lower level of suspicion than probable cause, and probable cause falls far short of a preponderance of the evidence standard. Baldwin v. State , 278 S.W.3d 367, 371 (Tex. Crim. App. 2009). Nevertheless, reasonable suspicion is dependent on both the content of the information known to police and its level of reliability. Alabama v. White , 496 U.S. 325, 330, 110 S.Ct. 2412, 2416, 110 L.Ed.2d 301 (1990).
C. Stops Based on Unconfirmed Insurance Returns
As we stated in Ellis v. State , Texas drivers must maintain proof of responsibility for their vehicles. 535 S.W.3d 209, 213-14, 2017 WL 5618686, at *4 (Tex. App.-Fort Worth 2017, pet. filed) (Pittman, J.); see Tex. Transp. Code Ann. § 601.051 (West 2011). Operating a vehicle for which no financial responsibility has been established is a misdemeanor punishable by a fine. Ellis , 535 S.W.3d at 213-14, 2017 WL 5618686, at *4 ; see Tex. Transp. Code Ann. § 601.191 (West Supp. 2017).
Cases addressing the validity of stops based on an officers database-derived suspicion from ambiguous terms like unconfirmed that the driver may be committing this misdemeanor fall into two general groups: cases in which the evidence dispels the ambiguity and shows that the data is reliable and cases in which the evidence falls short of doing so. Ellis , 535 S.W.3d at 213-14, 2017 WL 5618686, at *4 ; see United States v. Broca-Martinez , 855 F.3d 675, 678-79 (5th Cir. 2017) (noting that federal appellate cases addressing the issue turn on evidence of reliability).
1. Cases in Which Reliability of the Database Return Was Established
Two of this courts previous cases are illustrative of that group of cases in which the evidence supports a reliability finding and an ultimate determination of reasonable suspicion. In Swadley v. State , the evidence showed:
• What unconfirmed meant to the arresting officer, based on his training and experience;
• That the database was accurate and reliable based on the officers extensive experience;
• That the database was updated weekly; and
• What the possible responses from the database were and what they indicated.
No. 02-15-00085-CR, 2016 WL 7241564, at *1, *3, *6 (Tex. App.-Fort Worth Dec. 15, 2016, pet. refd) (mem. op., not designated for publication). Our court therefore upheld the trial courts determination that the officer had reasonable suspicion, concluding (1) that the officer had sufficient, specific articulable facts from which he could rationally infer that Swadleys vehicle had an insurance issue justifying further investigation and (2) that the evidence supported the finding that the database was sufficiently reliable. Id. at *6.
In Ellis , Ellis argued that evidence did not establish the databases reliability, 535 S.W.3d at 215-16, 2017 WL 5618686, at *6, but the arresting officer testified:
• He had used the database [t]ens of thousands of times;
• Only a handful of the (h)undreds, if not thousands of returns of unconfirmed he received from the database were in error; and
• The database was very accurate based on his experience.
Id. We therefore upheld the trial courts implicit finding that [the officers] knowledge was sufficient to establish the databases reliability for the purposes of establishing reasonable suspicion. Id. at 216-17, 2017 WL 5618686, at *7.
2. Cases in Which Reliability of a Database Return Was Not Established
At the other end of the spectrum, two of our sister courts cases and a Tenth Circuit opinion authored by now Supreme Court Justice Neil Gorsuch present situations in which the evidence did not prove the databases reliability. In Gonzalez-Gilando v. State , the database returned unavailable, not available, or undocumented regarding liability insurance coverage when troopers and a deputy sheriff typed in the information for Gonzalez-Gilandos vehicle. 306 S.W.3d 893, 894-95 (Tex. App.-Amarillo 2010, pet. refd). The deputy testified that the return led him to believe that the vehicle did not have insurance coverage, but he did not explain why. Id. at 896-97. On the other hand, one of the troopers testified that the return could have meant either that the vehicle was insured or that it was not insured. Id. at 897 n.2. No other evidence:
• Examined where the information in the database came from;
• Described what unavailable, not available, or undocumented meant;
• Demonstrated the error rate of the database;
• Established the age of the information in the database; or
• Proved the database was even accessible to the deputy or troopers at the time of the arrest.
Id. at 897. Our sister court in Amarillo therefore held that the deputys inference was not reasonable and that he lacked reasonable suspicion for the stop. Id.
In State v. Daniel , the State stipulated that the sole reason for the stop was that the vehicles status was unconfirmed by the database, and no evidence concerning the meaning of that term was mentioned in the opinion. 446 S.W.3d 809, 811-16 (Tex. App.-San Antonio 2014, no pet.). Relying on Gonzalez-Gilando , our sister court in San Antonio upheld the trial courts determination that the arresting officers lacked reasonable suspicion. Id. at 815-16.
In United States v. Esquivel-Rios , a Kansas trooper asked his dispatcher to verify a minivans temporary Colorado tag number with a law enforcement database. 725 F.3d 1231, 1234 (10th Cir. 2013). The dispatcher replied, [T]hats a negatory on record, not returning, but also said, Colorado temp tags usually dont return. Id. at 1234, 1235. The trooper testified that he had experienced (Colorado temporary tags) returning not on file. Id. at 1238 (emphasis removed). Nevertheless, the district court concluded,
[T]he trooper had reasonable suspicion to believe the minivan was displaying a forged tag. In Kansas, as elsewhere, vehicles must be registered with some lawful authority. When a law enforcement database yields no information about a registration tag, the [district] court reasoned, that raises a non-trivial possibility the tag wasnt lawfully issued in the first place but falsified in some way.
Id. at 1235. The district court also found that the trooper had encountered Colorado temporary tags before, and had them come back on file when he ran them through dispatch. Id. at 1238. Justice Gorsuch explained why the appellate court rejected the district courts ruling:
The district courts reasoning was right as far as it went. This court and others have regularly upheld traffic stops based on information that the defendants vehicles registration failed to appear in a law enforcement database-at least when the record suggested no reason to worry about the databases reliability.
The difficulty we face in this case stems from that last and critical qualification. When Trooper Dean asked about the minivans temporary tag, the dispatcher replied not only that the tag yielded a no return response from the queried database. The dispatcher also added that Colorado temp tags usually dont return. And this is a piece of evidence our cases havent confronted before: evidence admitted by a district court suggesting that the database on which the officer relied to justify his stop might bear a real problem-a problem that might mean a no return doesnt suggest criminal conduct but only some bureaucratic snafu.
The Fourth Amendment reasonable suspicion analysis requires a careful consideration of the totality of the circumstances. Yet in a case that hinges entirely on the reliability of a computer database, the district court overlooked one critical circumstance-the dispatchers comment-casting doubt on its reliability. And the district court misstated another critical circumstance doing the same thing-Trooper Deans comment about his experience with the database. All of the relevant circumstances, thus, were not considered.
Id. at 1235, 1238 (citations omitted); see also Broca-Martinez , 855 F.3d at 678-79 ([A] state computer database indication of insurance status may establish reasonable suspicion ... as long as there is either some evidence suggesting the database is reliable or at least an absence of evidence that it is unreliable.).
Like the courts in Ellis and Gonzalez-Gilando , however, the Esquivel-Rio s court heard very little evidence about the insurance database itself. See Esquivel-Rios , 725 F.3d at 1236.
D. We Uphold the Trial Courts Determination of No Reasonable Suspicion After Considering the Totality of the Circumstances.
This case does not fall clearly into either group. Rather, the record in the case before us contains a plethora of information about the database and testimony supporting its reliability, but it also contains evidence casting doubt on the reliability of the database, such as:
• Deputy Kristufeks testimony that he would stop about three or four vehicles per week coupled with his testimony that in a week, maybe three or four times, he would pull someone over to investigate an unconfirmed return, and the driver would provide him with proof of insurance;
• Deputy Kristufeks testimony that he would make three traffic investigations for unconfirmed insurance per week and in one of the three cases, the person would actually have insurance; and
• Database coordinator Melissa Burkhardts testimony that she could not explain the error rate experienced by Deputy Kristufek and that TDI does not keep records on rates of error.
We are presented with a close case. Had the trial court denied Binkleys motion to suppress and he appealed, given the same record and the applicable standard of review, we likely would have affirmed the trial courts decision. See Garcia-Cantu , 253 S.W.3d at 241 ; Wiede , 214 S.W.3d at 25 ; Ellis , 535 S.W.3d at 215-16, 2017 WL 5618686, at *7 ; Swadley , 2016 WL 7241564, at *6. This cases procedural posture-a States appeal of the trial courts order granting a defendants motion to suppress-does not affect the standard of review: we must imply the necessary fact findings that would support the trial courts ruling if the evidence, viewed in the light most favorable to the trial courts ruling, supports those findings. Garcia-Cantu , 253 S.W.3d at 241 ; see Wiede , 214 S.W.3d at 25.
Viewing the evidence in the light most favorable to the trial courts ruling, the trial court attached greater significance and credibility to Deputy Kristufeks testimony indicating a weekly error rate of 33% and potentially up to 100% in his experience with the database and to database coordinator Burkhardts inability to explain the error rate experienced by Deputy Kristufek. That evidence supports the trial courts implied finding that the database was not reliable. The undisputed evidence shows that Deputy Kristufek had no basis for the stop other than the return from the database. As Justice Gorsuch wisely recognized, [W]orries about a computer databases reliability might diminish if officers also possess other facts independently suggestive of unlawful activity. Esquivel-Rios , 725 F.3d at 1236. Thus, considering the totality of the circumstances in the record before us, we hold that the trial court did not err by implicitly finding that Deputy Kristufek did not have reasonable suspicion to support the stop.
III. CONCLUSION
Consequently, we overrule the States sole point, and we affirm the trial courts order granting Binkleys motion to suppress.