¶1 Thomas Morawek appeals the superior court order upholding a hearing examiner’s decision that his dog is a “dangerous dog” under the Bonney Lake Municipal Code (BLMC) because the dog killed a neighbor’s cat without provocation. Because substantial evidence does not support the lack of provocation finding, we reverse the dangerous dog designation. But, because the city of Bonney Lake (City) did not act arbitrarily or capriciously in pursuing the dangerous dog designation, we deny Morawek’s request for an award of attorney fees and other expenses under RCW 4.84.350. We grant his alternative request for statutory attorney fees and costs under RAP 14.2 and RAP 14.3, however, pending his compliance with RAP 18.1.
Melnick, J.FACTS
¶2 Morawek mowed his front yard, accompanied by Scout, his Labrador mix. Scout had been trained to stay on the Morawek property. Neighbors had never reported any problems with Scout.
¶3 As Morawek finished mowing, he noticed that Scout was no longer in the front yard. He went into the backyard, where he anticipated finding Scout, but the dog was not there. Morawek returned to his front yard and saw Scout *490sitting on the lawn. When Morawek called Scout into the house, he noticed a scratch on the dog’s nose.
¶4 Morawek’s neighbor Lynn Strong met him in his front yard and informed him that Scout had killed her cat, Oriel. Morawek looked around his house and shrubs but could not find the cat.
¶5 Lynn1 complained to Metro Animal Services, the animal control authority for Bonney Lake, that Scout had killed Oriel. She filed a statement explaining that she and her son Luke were sitting outside their residence when they heard a sound under the porch that sounded like animals fighting. Luke looked under the porch and saw Scout holding Oriel by the neck. Luke scared Scout away, and the dog ran off holding the cat. The Strongs never found Oriel.
¶6 Animal Control Officer Nicole Smith served Morawek with paper work stating that Scout satisfied the definition of a dangerous dog under the BLMC because he had killed a domestic animal without provocation while off his owner’s property. Morawek appealed the dangerous dog designation to the Bonney Lake chief of police and argued that it was more likely that the cat provoked the attack since the Strongs’ cats passed through his front yard regularly. The police chief rejected his appeal, and Morawek appealed that decision to the city hearing examiner.
¶7 Morawek, the Strongs, and Smith testified at the resulting hearing. Lynn testified that she did not see the animals under the porch. She added that her yard was not fenced and that Oriel had been an outside cat who came and went as she pleased. Luke testified that he heard a shriek and some scuffling before he looked under the porch and saw Scout holding Oriel.
¶8 Morawek argued that no proof of an unprovoked attack existed. He hypothesized that “[t]he cat could have been in my yard under a bush or shrub, a dog stuck his nose *491under there and could have been hit by the cat and that could have precipitated this whole thing.” Report of Proceedings (RP) (May 8, 2012) at 30. The hearing examiner disagreed and found no evidence that Oriel provoked the attack. He concluded as follows:
I just think the evidence is clear that the dog attacked this cat and killed the cat on private property, not his own. How he got there, I don’t know; but he was there and he killed the cat.
So I will find that Scout is a dangerous dog.
RP (May 8, 2012) at 31. The hearing examiner incorporated his oral findings into a written order upholding the dangerous dog designation.
¶9 Morawek then filed a writ of review in superior court. The superior court, acting in an appellate capacity, upheld the dangerous dog designation after rejecting the lack of provocation argument:
And then the argument is, is it with provocation or not with provocation? Dog versus cat? I think there is sufficient circumstantial evidence given the relative size of these animals that it would be without provocation.
RP (Jan. 18, 2013) at 23. Morawek petitioned this court for discretionary review, and we granted review solely on the lack of provocation issue.
ANALYSIS
I. Standard of Review
¶10 When examining a writ of review, we review the challenged administrative decision on the record of the administrative tribunal, not on the decision of the superior court acting in its appellate capacity. Nichols v. Seattle Hous. Auth., 171 Wn. App. 897, 904, 288 P.3d 403 (2012). We treat any findings of fact or conclusions of law the superior court made as surplusage. Grader v. City of Lynnwood, 45 Wn. App. 876, 879, 728 P.2d 1057 (1986). Issues of law are *492reviewed de novo, and issues of fact are reviewed for substantial evidence. City of University Place v. McGuire, 144 Wn.2d 640, 647, 30 P.3d 453 (2001); Nichols, 171 Wn. App. at 904. “Substantial evidence is evidence in sufficient quantum to persuade a fair-minded person of the truth of the declared premise.”Holland v. Boeing Co., 90 Wn.2d 384, 390-91, 583 P.2d 621 (1978).
¶11 We apply the rules of statutory construction to local ordinances. Sleasman v. City of Lacey, 159 Wn.2d 639, 643, 151 P.3d 990 (2007). Where an ordinance is unambiguous, construction is not necessary because the plain meaning controls. McTavish v. City of Bellevue, 89 Wn. App. 561, 565, 949 P.2d 837 (1998). When words are not defined, we may refer to dictionary definitions and to common usage. Armstrong v. State, 91 Wn. App. 530, 538, 958 P.2d 1010 (1998). In addition, when construing an ordinance, we must give meaning to all words and clauses. Gilbert H. Moen Co. v. Island Steel Erectors, Inc., 128 Wn.2d 745, 762, 912 P.2d 472 (1996).
II. Lack of Provocation
¶12 The BLMC defines a dangerous dog as any dog that, according to the city records,
1. [h]as inflicted severe injury on a human being without provocation while on public or private property;
2. [h]as killed a domestic animal without provocation while off the owner’s property; or
3. [h]as been previously found to be potentially dangerous, the owner having received notice of such, and the dog again aggressively bites, attacks, or endangers the safety of humans or domestic animals.
BLMC 6.04.010(G).2
*493¶13 The hearing examiner upheld Scout’s designation as a dangerous dog under subsection (2) of this definition. The only issue here is whether substantial evidence supports the hearing examiner’s finding that Scout killed Oriel without provocation.
¶14 The term “provocation” is undefined in the municipal code, so we look to its ordinary meaning, as supplied by the dictionary definition. State v. Silva, 106 Wn. App. 586, 591, 24 P.3d 477 (2001). “Provocation” is the act of inciting another to do something, or is something that affects a person’s reason and self-control. Black’s Law Dictionary 1346 (9th ed. 2009).
¶15 Morawek argues that this definition requires an observation of how an incident started. We disagree. Evidence can be either direct or circumstantial, and one type of evidence is not necessarily more or less valuable than the other. Rogers Potato Serv., LLC v. Countrywide Potato, LLC, 152 Wn.2d 387, 391, 97 P.3d 745 (2004). However, as Morawek correctly asserts, nobody saw how the fight between Scout and Oriel began. Compare Commonwealth v. Seyler, 929 A.2d 262, 264-66 (Pa. Commw. Ct. 2007) (when neighbor called out to see if dogs’ owner needed help, and dogs ran over to neighbor and bit her, the evidence showed lack of provocation), with McCoy v. Lucius, 36,894 (La. App. 2 Cir. 3/5/03); 839 So. 2d 1050, 1055 (plaintiff’s testimony that neighbor’s dog charged her and her dogs showed neighbor’s dog provoked the attack). Additionally, this is not a case where there is considerable circumstantial evidence to show that the incident started without provocation, as there was in Logan County Animal Control Warden v. Danley, 211 Ill. App. 3d 198, 569 N.E.2d 1226, 1229, 155 Ill. Dec. 615 (1991). In that case, despite no eyewitness testimony to the attack, the evidence showed that the defendant’s dogs ran loose and were seen around the plaintiff’s hog pen when dead and injured hogs were discovered. When *494combined with the fact that the hogs had been in the fenced pen for only two days, the evidence supported an inference that the hogs did not provoke the attack. Danley, 569 N.E.2d at 1229.
¶16 The City responds that Scout’s act of trespassing on the Strong property shows that he provoked the attack and nullifies any claim that the cat supplied the provocation. We agree that the killing occurred on the Strongs’ property, which satisfied the “location” element of the dangerous dog definition. We disagree, however, with the City’s argument that the location of the killing also proves it occurred without provocation. This argument conflates two elements of the dangerous dog definition into one and renders the “lack of provocation” element superfluous.
¶17 The City adds that it would be relatively impossible for a cat to scratch a dog’s nose without the dog poking into the “personal space” of the cat and that the act of the dog’s nose coming toward the cat would be provocation. Resp’t’s Br. at 11. This argument is speculative. There are no facts in the record to support this proposition. The evidence showed that Oriel roamed freely and passed through Morawek’s front yard regularly. Additional evidence showed that Scout was trained to stay on Morawek’s property. As Morawek argued below, it is possible that the cat provoked the attack by scratching Scout’s nose when the dog stuck it under a bush on the Morawek property. We reject the City’s argument that the scratch on Scout’s nose is proof of provocation.
¶18 The City also argues that because the Strongs did not hear anything before they heard the cat’s shriek under the porch, the attack began with the shriek and was unprovoked. Here again, this argument is rooted in speculation. The fact that the Strongs’ first indication of a fight was a shriek does not prove when the incident began or whether it was provoked or unprovoked.
*495 ¶19 The City also cites the superior court’s observations regarding the relative size of the animals as evidence of lack of provocation. As stated earlier, any findings by the superior court are surplusage. Moreover, the court’s observations were not evidence and were not based on the evidence. At the hearing, no evidence showed the size of either animal. Even if we may take judicial notice of the fact that dogs are generally larger than cats, this fact alone does not show that Scout attacked Oriel without provocation.
¶20 We agree with Morawek that the finding that Scout acted without provocation is not supported by substantial evidence, and we reverse the superior court order upholding the dangerous dog designation.
¶21 A majority of the panel having determined that only the foregoing portion of this opinion will be printed in the Washington Appellate Reports and that the remainder shall be filed for public record in accordance with RCW 2.06.040, it is so ordered.
Johanson, C.J., and Bjorgen, J., concur.
We refer to Lynn Strong and her son Luke by their first names for clarity.
The dangerous dog designation required Morawek to pay a $250 permit fee to the Animal Control Agency as well as an annual $100 to renew the permit; obtain a surety bond or liability insurance in the sum of $500,000; tattoo or microchip his dog; pass a sight inspection of his dog’s premises; provide photographic identifi*493cation of his dog; and provide proof that his dog had been neutered and vaccinated for rabies. See BLMC 6.04.183, .184, .186.