Greene, J.:
Helmerich & Payne, Inc. (Taxpayer) appeals the district court’s affirmance of a decision of the State Board of Tax Appeals (BOTA) establishing the value for ad valorem tax purposes of an oil and gas leasehold interest in Seward County for tax year 2001. Taxpayer argues that BOTA’s approach was based on unsupported facts, relied on a misapplication of law, and failed to follow prescribed procedures for valuation of oil and gas interests. We reverse and remand with directions.
Factual and Procedural Background
Sometime in early 2000, Taxpayer recompleted a gas well for oil production on its “Home Royalty A-2” lease in Seward County. First oil production was achieved in late June 2000, and actual monthly oil production thereafter was as follows:
July 4,065 barrels
August 3,917 barrels
September 2,677 barrels
October 2,540 barrels
November none (down for repairs)
December 2,356 barrels
January 2001 2,256 barrels
February 2001 1,553 barrels
March 2001 1,721 barrels
Taxpayer requested use of a 50% decline rate based on actual production for calendar year 2000 and confirmed this rate with production data from the first quarter of 2001. In contrast, the appraiser defaulted to an assumed 30% decline rate, testifying that “30 percent is not always used, but in your first year there’s — unless it produced a whole year, you really can’t see a decline rate. And that’s why — that’s where that 30 percent comes in handy for the [55]*55Taxpayer and the County.” Use of these respective decline rates in the prescribed formula resulted in a significant variation in taxable value; the County valued the working interest at $936,805, and the Taxpayer valued the same interest at $345,062.
After a hearing, BOTA affirmed the County’s valuation, reasoning as follows:
“6. The Taxpayer requests that the subject property be appraised using a maximum decline rate of 50%. This decline rate is calculated mainly using production figures taken in 2001.
“7. The facts show that this is a new well, and should be appraised as such. Page five, of die 2001 Oil and Gas Guide, gives guidance as to how such wells are to be appraised. It suggests that an assumed 30% decline rate be used when the actual decline rate cannot be established. Further on it states that requests for a steeper decline curve should be documented by production decline.
“8. In this case, the production decline data supplied by the Taxpayer comes from a period well after the appraisal date, thereby making it irrelevant for appraisal as of Januaiy 1, 2001. The County did however, consider utilizing back to back quarters comparing the last quarter of 2000 to the first quarter of 2001. This analysis is commonly used where production figures for the whole year are hard to come by. In this analysis, the decline rate is slightly over 30%. Based on the best evidence available to it when the subject property was appraised, the Board finds that the County correctly appraised the subject property using a 30% decline rate. Therefore, the Board finds that the County’s valuation of the subjectproperty should be upheld.”
Taxpayer appealed BOTA’s decision to the district court, which affirmed BOTA after oral argument based upon the record at BOTA. This appeal followed.
Statutory Guidelines and Theory of Oil and Gas Leasehold Valuation for Ad Valorem Taxation in Kansas
For purposes of valuation and taxation in Kansas, all oil and gas leases and wells are considered personal property. K.S.A. 79-329. Persons who own such personal property are required to file a statement of assessment on standard rendition forms on or before April 1 of each tax year. K.S.A. 79-332a. In practice, the county appraiser then reviews the taxpayer’s rendition and determines whether changes to the valuation are required and thereafter notifies the taxpayer of the appraised value. See K.S.A. 2004 Supp. 79-1460. The county appraiser is obligated to follow the Oil and [56]*56Gas Appraisal Guide (Guide) prescribed by the Director of Property Valuation but may deviate from the Guide on an individual piece of property “for just cause shown and in a manner consistent with achieving fair market value.” K.S.A. 79-1456. In determining the value of such property, the appraiser must also consider statutory factors of value:
“Except as otherwise provided in subsection (b) of this section, in determining the value of oil and gas leases or properties the appraiser shall take into consideration tire age of the wells, the quality of oil or gas being produced therefrom, the nearness of the wells to market, the cost of operation, the character, extent and permanency of the market, the probable life of die wells, the quantity of oil or gas produced from the lease or property, the number of wells being operated, and such other facts as may be known by the appraiser to affect the value of the lease or property.” K.S.A. 2004 Supp. 79-331(a).
Consideration of these statutory factors is mandatory; failure to taire into consideration any of these statutory factors will invalidate the assessment. See Garvey Grain, Inc. v. MacDonald, 203 Kan. 1, 14-15, 453 P.2d 59 (1969). In the context of oil and gas valuation, failure to give consideration to known production decline in making an assessment may be considered inadequate consideration of the “probable life of the wells,” thus rendering the assessment arbitrary, capricious, and void as a matter of law. Angle v. Board of County Commissioners, 214 Kan. 708, 713, 522 P.2d 347 (1974).
Our Supreme Court recently addressed the theory of oil and gas valuation in Kansas in Board of Ness County Comm’rs v. Bankoff Oil Co., 265 Kan. 525, 960 P.2d 1279 (1998), where the following explanation was quoted with approval:
“ ‘[T]he theory of the [Oil and Gas Appraisal Guide issued by the Division of Property Valuation] is that we are appraising the reserves that are in the ground. And so the guide, the basic mechanics of the guide is to discount income over a period of time to reflect the production capabilities of that reserve. And then it combines with that a rate of decline which is indicating that that reserve is depleting. That is combined with a discount, discounting the money — for the money that is not received until a later time. You get a present worth factor, which is a multiple of money, and that’s multiplied times the value, or price of the oil, times the production; and that’s to indicate a probable reserve value, from which is deducted the expenses for lifting the oil, to get a net working interest. Then they add the equipment, production equipment, to that, for the working interest. So [57]*57it’s based on the probable life of the reserve, and the probable dollars per barrel that will be received by the operator, or the working interest.’ ” 265 Kan. at 529.
As then noted by the court, this theory provides the basis for a mathematical formula to compute gross reserves:
_ X _ = _ X _ = _
1. Total 2. Net Price 3. Estimated 4. Present 5. Estimated
Amount as of Jan. 1 Gross Income Worth Factor Gross Reserve
(barrels) Value
Production
265 Kan. at 529. Finally, the court recognized that “[i]t is apparent that the determination of the decline to be applied to the lease becomes the most critical factor in establishing its valuation.” 265 Kan. at 530.
Standard of Review
Judicial review of orders of BOTA is governed by K.S.A. 77-621. For purposes of this appeal, application of this statute requires the appellate court to grant relief if: (i) the agency has erroneously interpreted or applied the law, K.S.A. 77-621(c)(4); (ii) the agency has engaged in an unlawful procedure or has failed to follow prescribed procedure, K.S.A. 77-621(c)(5); (iii) the agency action is based on a determination of fact, made or implied by the agency, that is not supported by evidence that is substantial when viewed in light of the record as a whole, K.S.A. 77-621(c)(7); or (iv) the agency action is otherwise unreasonable, arbitrary, or capricious, K.S.A. 77-621(c)(8).
On appeal of BOTA’s decision, the party complaining bears the burden of demonstrating that the agency erred. K.S.A. 77-621(a)(1). When the district court has reviewed an agency decision prior to this court’s review, we focus on the agency action and apply the same standards of judicial review. Connelly v. Kansas Highway Patrol, 271 Kan. 944, 964, 26 P.3d 1246 (2001), cert. denied 534 U.S. 1081 (2002).
When construing tax statutes, imposition provisions are considered penal in nature and must be construed strictly in favor of the taxpayer. In re Tax Appeal of Harbour Brothers Constr. Co., 256 Kan. 216, 223, 883 P.2d 1194 (1994). Interpretation of any statute [58]*58is a question of law over which this court has unlimited review. Matjasich v. Kansas Dept. of Human Resources, 271 Kan. 246, 250-51, 21 P.3d 985 (2001). Although BOTA’s decisions in its area of expertise are given deference, this court will take corrective steps if BOTA’s actions are erroneous as a matter of law. In re Tax Appeal of Intercards, Inc., 270 Kan. 346, 349, 14 P.3d 1111 (2000).
Is BOTA’s Finding that Taxpayer’s Requested Decline Rate Was “Calculated Mainly Using Production Figures Taken in 2001” Supported by Evidence That Is Substantial When Viewed in Light of the Record as a Whole?
Taxpayer first argues that BOTA’s finding that Taxpayer’s requested decline rate of 50% was “calculated mainly using production figures taken in 2001” was not adequately supported by the record. We agree. This finding is a mischaracterization of Taxpayer’s approach to value and is not supported by the record.
Taxpayer’s requested decline rate was calculated using production numbers for July 2000 through December 2000. Taxpayer’s production manager testified at the BOTA hearing that die requested decline rate “was reached originally with just the production we had through the calendar year 2000. It was confirmed with production through April 2001 and a prorated number for the first 14 days of May in 2001. Then taking all current data up through April 2002, it was once again confirmed.”
BOTA misstated the evidence in concluding Taxpayer’s requested rate was “calculated mainly using production figures taken in 2001,” when the record shows it was calculated using 2000 figures and merely confirmed with figures from 2001 and 2002. BOTA tiien based its decision to adopt the County’s assumed 30% decline rate at least partially on the fact that BOTA erroneously perceived Taxpayer’s approach as based on periods beyond the calendar year preceding assessment. Because this factual finding was not supported by evidence that is substantial when viewed in light of the record as a whole, we are obligated to set aside BOTA’s finding in addressing the ultimate issue of leasehold valuation. See K.S.A. 77-621(c)(8).
Did BOTA Err in Declaring the Production Data for 2001 “irrelevant for appraisal as of January 1, 2001”?
[59]*59Taxpayer next argues BOTA misapplied applicable law by failing to consider the subject well’s post-valuation production data and in finding the well’s production data for 2001 was “irrelevant for appraisal as of January 1,2001,” contrary to Bankoff, 265 Kan. 525. Taxpayer further argues that BOTA made inconsistent findings when it found that 2001 post-valuation date production data was irrelevant but simultaneously approved the County’s use of the same data.
We conclude that BOTA’s statement regarding the irrelevance of the 2001 data is an erroneous interpretation or application of the law because it is contrary to the Guide and the holding in Bankoff. According to our Supreme Court,
“[a]llowmg consideration of production data collected after January 1 when alease suffers a decline late in the year conforms to generally accepted appraisal procedures in the oil and gas industry and is consistent with the definition of fair market value of K.S.A. 79-503a, despite the apparent incongruity with the statement that property is to be assessed as if a sale had been consummated on January 1.
“Refusing to allow consideration of post-January 1 production data when assessing an oil or gas lease that has suffered a decline in production would be to ignore relevant and available factual information pertaining to the lease’s future productivity and income. This result is not required by K.S.A. 79-503a, is contrary to assessment principles embodied in the statutory scheme, and denies the expertise wielded by the Director of Property Valuation in adopting the Guide.” 265 Kan. at 543-44.
The County is incorrect in stating that Bankoff does not apply. Bankoff essentially recognized the importance of considering all available data, whether or not available on the valuation date.
“Under our ad valorem taxation scheme, the valuations are being reported, altered, tested, appealed from, and determined during essentially the first half of the tax year. An overview of the entire process and consideration of the numerous statutes makes it clear that all available appraisal information is to be utilized regardless of whether the data originated before, on, or after January 1 of the tax year.” (Emphasis added.) 265 Kan. at 543.
In this case, Bankoff allows consideration of post-valuation production data to confirm prevaluation market trends in determining future productivity and earning potential. BOTA erred as a matter of law in declaring the 2001 production data “irrelevant” to deter[60]*60mination of the fair market value of the subject interest. See K.S.A. 77-621(c)(4); Bankoff, 265 Kan. at 544.
Did BOTA Err in Upholding the County’s Valuation?
Finally, Taxpayer argues generally that BOTA’s errors are “outcome determinative” and warrant reversal of the valuation determined, arguing that the County’s approach departs from prescribed procedure set forth in the Oil and Gas Appraisal Guide and that the Taxpayer’s approach conforms to the Guide. Both parties are in agreement that the subject property should be treated as a “new lease” for purposes of identifying the applicable provisions of the Guide. Those provisions, in their entirety, state as follows:
“Use the first few months of production to establish the decline rate.
“If the first few months of production and all data available indicate no decline, it is suggested the Appraiser consider the use of an assumed 30% annual decline rate and evaluate tifie property on this basis. This, however, is not automatic and is to be used only when the actual decline rate cannot be established. Use of a proven neighborhood decline rate may be considered appropriate after proper consideration for flush production, but only when the new well or wells are completed in the same reservoir. Requests for consideration of percentage decline above 30% or adjustment by the Appraiser below 30% should be documented by production decline, water cut and/or gas oil ratio curves.
“Abnormal sharp decline is usually found with initial production from newly completed wells on new leases, added wells on existing leases, and recompletions or workovers on existing leases. The Appraiser should consider application of historic declines when actual declines are uncertain or are obscured from lease development or workovers. A lease with initial ‘flush’ production will show an abnormal sharp decline followed by change in the decline rate to normal rate of decline. If the property shows a constant rate of decline after the ‘flush’ production, the appropriate present worth factor for that rate should be used with production annualized for tire period reflecting the stabilized production period.
“A decline curve, with downtime noted, should be submitted with the filing when an adjustment for abnormal decline is requested. No production period less than four to six months should be used to establish an abnormal decline. In addition to decline curves, water cut, and/or gas oil ratio curves may be filed with the filing to document changes in reservoir behavior.” 2001 Oil and Gas Appraisal Guide, pp. 5-6.
The Taxpayer relies on that provision requiring use of “the first few months of production to establish the decline rate” whereas [61]*61the County argues that there is obvious “flush” production in the first 2 months after recompletion that should not be considered in calculating decline rate, requiring use of the “constant rate of decline after the ‘flush’ production.” (Emphasis added.) The County relies on that provision authorizing “use of an assumed 30% annual decline rate,” whereas the Taxpayer argues that “this is not automatic and is to be used only when the actual decline rate cannot be established.” The Taxpayer argues that it relies on “all data available,” whereas the County argues that “no production period less than four to six months should be used to establish an abnormal decline.”
It is clear to this court that the Guide alone may not provide sufficient precision to resolve the issues framed in this appeal. Accordingly, in resolving apparent ambiguities in the Guide, we rely on principles of law set forth in our statutory scheme as interpreted and applied by our appellate courts. To the extent that the Guide is not consistent with these principles of law, it is erroneous as a matter of law. See Garvey Grain, 203 Kan. at 12.
Flush Production Data
First, we must determine what should be and what should not be considered eligible data from which to determine annual production and decline rate for the prescribed formula. The County argues that July and August 2000 production should not be considered because the recompleted well was experiencing flush production. The Taxpayer argues and the district court held that the County’s failure to raise this issue at BOTA bars consideration of the argument on appeal. We hold that, whether or not argued at BOTA, flush production is so common and so prominently addressed in the Guide and our case law that it cannot be ignored.
The Guide advised that the “prior year production may not represent tire production capability of the lease for several reasons,” including when “[l]ease production began during die year with ‘gusher’ characteristics (flush production) followed by rapid decline to a stabilized level.” Moreover, for new leases, the Guide is explicit in stating that where a more constant rate of decline is apparent “after the ‘flush production,’ ” this more constant rate of decline [62]*62should be utilized. Additionally, we know from Bankoff that “flush production” is the reason for the discount provided in K.S.A. 79-331(b) and that this subsection “compels appraisers to anticipate a decline in production, even though no decline is apparent from a lease’s prior production history.” 265 Kan. at 540. Bankoff also recognizes that the failure to account for flush production would “result in an inaccurate valuation and assessment.” 265 Kan. at 541-42; see State ex rel. Stephan v. Martin, 230 Kan. 747, 641 P.2d 1011 (1982).
For all of these reasons, we hold that BOTA erred as a matter of law in ignoring the impact of flush production apparent in the months of July and August 2000 for the subject property. We decline to remand for a finding as to flush production because the phenomena can and should be identified exclusively from an analysis of raw production data. 2001 Oil & Gas Appraisal Guide, pp. 2-3 (requiring adjustment of production for “gusher” characteristics based solely on rendition data). Here, the production from both July and August 2000 exceeded any subsequent monthly production by more than 50%. Production data from these months was clearly abnormal and should not be considered for purposes of determining decline or for purposes of determining annual production as factors to be employed in the formula to calculate gross reserve value.
First Quarter 2001 Production Data
As discussed and concluded above, we follow Bankoff in holding that first quarter production data from the subject property was relevant to “the lease’s future productivity and income”; clearly the information is relevant to determination of decline rate here just as it was in Bankoff.
Proper Use of Data in Determining Decline Rate and Annual Production
Under the Guide, proper valuation of any oil well begins with a determination of the annualized production for that well. BOTA did not address the discrepancy between the parties’ annualized production figures. The County used the proper calculation but [63]*63improperly included the months of flush production. Taxpayer merely multiplied the December 2000 production by 12 — a method not recommended by the Guide.
For leases which have produced less than 12 months during the prior year but commenced production prior to July 1, page 3 of the Guide recommends calculating the annualized production “by dividing the production for the period of the prior year that the lease did produce by the number of days it produced to ascertain the bbls/day and then multiply the B/D by 365 to calculate the annualized production.” After disregarding flush production, the subject well produced for 92 days and produced 7,573 barrels, or 82.3 barrels per day, yielding an annualized production of 30,045 barrels.
Having established the annual production for the formula, we address the proper approach to determine decline rate and its corresponding present worth factor. In Bankoff, the Supreme Court affirmed BOTA’s reliance on two approaches to determine decline in similar circumstances: (i) comparison of “back to back quarters,” specifically the last quarter of the prior year compared to the first quarter of the tax year; and (ii) comparison of the previous year s production with an annualization of first quarter production from the tax year. In Bankoff, these approaches achieved the same rate of decline but this may have a been mathematical coincidence.
The back-to-back quarter analysis is acknowledged by the Guide, especially where “the production is manifesting an accelerated rate of depletion.” Guide, p. 6. Here, BOTA recognized that back-to-back quarter analysis “is commonly used where production figures for the whole year are hard to come by,” but neither BOTA nor the County performed the analysis correctly. In these circumstances, where the lease was down for repairs during 1 month of the last quarter 2000, production from the final 3-month period of production (September, October, and December) should be compared to production from the first quarter of 2001. Comparing the sum of (adjusted) the last quarter of 2000 [7,573 barrels] to the sum of first quarter of 2001 [5,530 barrels], the quarterly decline is 26.9%, which — based on the table provided in the Guide — corresponds to an annual decline in excess of 50%.
[64]*64The alternative analysis adopted by BOTA in Bankojf was not performed here. We note that the Seward County Appraiser performed an iteration of this analysis, however, and concluded that the lease was experiencing a 36% annual decline. The mathematical problem with the appraiser s iteration is that she annualized comparative data from 6 months in 2000 and 3 months in 2001. This annualization tends to flatten the decline curve that would be expected if one had a complete production history from 1 or both years; i.e., the hypothetical missing production from early 2000 would be greater than the final 6 months, and the projected missing production from later in 2001 would be less than the first 3 months. Where data is incomplete, the most accurate mathematical calculation of decline employs actual production for equal time periods on both sides of the fraction, as is tire case with the back-to-back quarter analysis approved above.
In order to confirm the accuracy of the result achieved by the singular use of back-to-back quarters, we examine the record to determine whether other evidence tends to support or refute this result. First, we note that the Taxpayer supported its decline rate request with engineering curves produced in the ordinary course of its business; all such curves demonstrate a projected decline rate in excess of 50% for the subject property. Second, we note that Taxpayer also submitted two engineering reserve analyses demonstrating that gross estimated oil recovery for the subject property was nearly 20% less than the gross recovery associated with the 50% decline rate; clearly, use of the 50% rate has not undervalued this property. Third, we note that monthly decline for the months in calendar year 2000following flush production averaged 6%, thus corresponding to an annual decline far in excess of 50%. Fourth, we note that first production in late June 2000 barely disqualified this property for special treatment under K.S.A. 2004 Supp. 79-331(b), applicable to new wells producing after July 1 of the calendar year preceding the tax year; if 79-331(b) had applied and its methodology was adopted with known data for the subject property, the resulting value would he within 2% of the working interest value when calculated with a decline rate in excess of 50%. Finally, even if we were to ignore flush production and tire 2001 data, actual [65]*65production achieved in calendar year 2000 by the new well clearly supports a decline rate in excess of 50%.
In summary, we conclude that BOTA: (i) based its action on a determination of fact, i.e., that Taxpayer calculated its decline rate “mainly using production figures taken in 2001,” that is not supported by evidence that is substantial when viewed in light of the record as a whole, requiring relief pursuant to K.S.A. 77-621(c)(8); (ii) erroneously applied the law in considering first quarter 2001 production “irrelevant” for valuation purposes, requiring relief pursuant to K.S.A. 77-621(c)(4); (iii) failed to follow prescribed procedure in adopting an assumed decline rate when there was adequate data to establish a rate, requiring relief pursuant to K.S.A. 77-621(c)(5); and (iv) failed to follow prescribed procedure in adopting tire County’s purported “back-to-back quarter” comparison that was mathematically flawed, requiring relief pursuant to K.S.A. 77-621(c)(5).
In granting relief pursuant to K.S.A. 77-622(b), we reverse the district court, vacate the order of BOTA, and we remand to the district court with directions to remand to BOTA for a determination of the value of the subject property in a manner not inconsistent with this opinion, together with a calculation and order of any refunds to the Taxpayer and royalty interest that may be required as a result.
Reversed and remanded with directions.