OPINION

ANDERSON, Justice.

This case comes to us on certiorari to the Minnesota Tax Court. Relator, Harold Chevrolet, Inc., challenged the 1990 assessed value of three tracts of property it was leasing and using as a car dealership (subject property). The Hennepin County assessor estimated that the fair market value of the subject property was $3,738,600. The tax court lowered the market value to $3,680,000. Harold moved for amended findings or a new trial. The court denied Harold's motions. On appeal, Harold argues: (1) the court erroneously double-counted the value of two of the subject property's four parcels, and (2) the court relied too heavily on the cost approach to market value. We affirm.

I.

On the date of assessment, Harold operated a car dealership on the subject property at the intersection of Highways 494 and 35W in Bloomington, Minnesota. Harold was leasing the subject property, which consisted of four separate parcels of land contained in three noncontiguous tracts. The tracts included: the main Harold Chevrolet dealership site, which consisted of two contiguous parcels of land (main tract); a used car lot (used car lot); and an employee parking lot (employee parking lot). The Hennepin County assessor determined that the market value of the subject property was $3,738,600 as of January 2, 1990, for taxes payable January 2, 1991.

The land area of the subject property was 326,639 square feet. The combined gross building area of the subject property was 61,040 square feet. The land area of the main tract was approximately 270,300 square feet and the tract was improved with a main office, a showroom service building, and a small used car office. The gross building area of the main tract improvements was 59,720 square feet. The used car lot, located across the street from the main tract, had a land area of 35,410 square feet and the tract was improved with a single story concrete building of approximately 1,320 square feet. Finally, the land area of the employee parking lot, which was located two blocks away from the main tract and across the street from the used car lot, was approximately 21,929 square feet. The employee parking lot contained no improvements.

II.

The three traditional approaches for determining market value are: market data or sales comparison, replacement cost, and capitalization of income. Lewis Harris v. County of Hennepin, 516 N.W.2d 177, 178 (Minn. 1994); Fed. Reserve Bank of Minneapolis v. County of Hennepin, 372 N.W.2d 699, 700 (Minn. 1985); Northerly Centre Corp. v. County of Ramsey, 311 Minn. 335, 337, 248 N.W.2d 923, 925 (1976). Under the market approach, an appraiser surveys the market and compares the subject property with other comparable properties, adjusting for differences such as time of sale, location, size, and topography. Northerly Centre Corp., 311 Minn. at 337 n. 2, 248 N.W.2d at 925 n. 2. Under the cost approach, the appraiser determines the current cost of constructing the existing improvements on the property, subtracts depreciation to determine the current value of the improvements, and then adds the value of the land. Id. Finally, under the income approach, the appraiser determines the rental income the property should be generating, subtracts expenses, and then capitalizes the net income at a rate of capitalization that investors would expect to obtain. Id.

In the present case, each party presented the testimony of an expert appraiser who analyzed the market value of the subject property using the three traditional approaches to market value. Harold's appraisal expert, Jeffrey Johnson, a licensed real estate appraiser, opined that the market value of the subject property was $2,800,000. Under his market approach, Johnson opined that the value for all three tracts was $2,990,000. Under his cost approach, Johnson separately calculated the land values of each of the three tracts and concluded the total land value for the subject property was $2,340,000. Based on a depreciation rate of 81.6%, he opined that the value of the depreciated improvements was $667,920. Johnson's total value for all three tracts under the cost approach was $3,010,000. Under his income approach, Johnson opined that the value for all three tracts was $2,560,000.

In establishing a market value of $2,800,000 for the subject property, Johnson testified that he relied most heavily on the market and income approaches, with minor consideration for the cost approach. He concluded the cost approach was unreliable given the significant amount of depreciation the improvements had experienced.

The county's appraisal expert was John G. Pasternacki, a certified assessor for the City of Bloomington. Pasternacki, unlike Harold's expert, did not value the three tracts as a single unit. For his market approach, Pasternacki calculated only the value of the main tract and opined that its value was $3,437,000. For his cost approach, Pasternacki separately calculated the value of each of the three tracts, using an improvement depreciation rate of 85%. His land and depreciated improvement value totals were: $3,561,200 for the main tract; $417,200 for the used car lot; and $230,000 for the employee parking lot. For his income approach, Pasternacki again analyzed only the main tract and opined that its value was $3,470,200.

In establishing a final market value for the subject property, Pasternacki provided three separate market value figures. He opined that the market value of the main tract was $3,450,000; the market value of the used car lot was $417,200; and the market value of the employee parking lot was $230,300. His total market value for the subject property was $4,097,500. Pasternacki testified that he relied most heavily on the market and income approaches for the main tract because he concluded these methods offered the most reliable information. Pasternacki relied exclusively on the cost approach for the used car lot and the employee parking lot.

After a two-day trial, a viewing of the subject property, and the submission of post-trial briefs, the tax court lowered the market value on the subject property from $3,738,000 to $3,680,000. The court analyzed the subject property under the three traditional approaches to value. For its cost approach, the court adopted Harold's estimated depreciated improvement value of $667,920. The court used land values of $2,662,455 for the main tract; $300,985 for the used car lot; and $186,397 for the employee parking lot. The court's total value for all three tracts under the cost approach was $3,817,757.

For its income approach, the tax court discussed the figures submitted by the parties' experts and, without further explanation, held that the main tract alone had a value in the range of $2,800,000 to $3,000,000. The court did not explicitly state a market approach value for the subject property. It did, however, discuss the comparable properties used by the two expert appraisers and the adjustments made by each of them. It concluded the adjustments made by the county's appraisal expert were more reasonable, and the court placed more weight on his market approach value. For its final conclusion as to market value, the court placed equal weight on the cost and income approaches, and less weight on the market approach.

III.

This court will not disturb the tax court's valuation of property for tax purposes unless the tax court's decision is clearly erroneous, which means the decision is not reasonably supported by the evidence as a whole. Westling v. County of Mille Lacs, 512 N.W.2d 863, 866 (Minn. 1994). The tax court's decision should be considered clearly erroneous only when this court is left with a " 'definite and firm conviction that a mistake has been committed * * * '." Id. (quoting In re Assessments of Silver Lake Apartments v. County of Olmsted, 295 Minn. 548, 549-50, 204 N.W.2d 415, 416 (1973)).The taxpayer has the burden of showing that the valuation reached by the assessor is excessive. Id. The inexact nature of property assessment necessitates that this court defer to the decision of the tax court unless the tax court has either clearly overvalued or undervalued the property, or has completely failed to explain its reasoning.

Harold alleges the tax court committed two reversible errors. First, Harold argues the court "double-counted" the value of the two smaller tracts, artificially inflating the market value of the subject property. According to Harold, the court double-counted the value of the two smaller tracts when the court allegedly added value for these two smaller tracts to its determination of market value for the main tract. Harold argues the double-counting occurred because the court adopted an erroneous land-to-building ratio, which the county's appraisal expert included in his analysis. Harold contends the court relied on the county's land-to-building ratio to conclude that the two smaller tracts were "excess" land, which meant they were not needed to support the car dealership on the main tract. Harold claims the car dealership cannot function on the main tract without the two smaller tracts; therefore, the value of the main tract must include the value of the two smaller tracts.

A land-to-building ratio is sometimes used by appraisers to determine whether a tract has sufficient land available to serve and support the tract's existing improvements. It is normally calculated by dividing gross building area by total land area. The county's appraiser testified that he calculated the land-to-building ratio of the subject property and all comparable properties by dividing the land area by the "footprint" (i.e., first floor) area of the improvements, rather than by the gross building area. The improvements' footprint area shows how much of the land the improvements occupy. Once this value is subtracted from the total land area, a car dealership must have sufficient land remaining to display inventory. The county's appraisal expert admitted that using the footprint area is not the normal method for calculating a land-to-building ratio. He testified, however, that a ratio based on footprint area better represents the relationship between land and improvements when analyzing the needs of a car dealership. Harold's appraisal expert agreed that the area available to display inventory is critical and unique for valuing car dealerships. We conclude that the county's incorporation of this unique aspect of car dealership property into its land-to-building ratio calculation was reasonable.

Notwithstanding our conclusion that the county's land-to-building ratio was reasonable, we do not agree that the tax court adopted the county's land-to-building ratio methodology as an essential element in reaching its conclusion as to market value. We note that the county's appraisal expert testified that his land-to-building ratio did not affect his separate valuation of the three tracts. He valued the tracts separately because they were noncontiguous and could be sold separately in fee simple. The court heard this testimony as well as Harold's concerns regarding the county's land-to-building ratio calculation. Nothing in the court's opinion suggests it adopted the county's land-to-building ratio or even considered the land-to-building ratio of the property as an essential factor in its analysis.

For many of the same reasons, we are unable to accept Harold's argument that the tax court double-counted the value of the two smaller tracts. The court treated the three tracts separately only in its cost approach analysis. For its cost approach, the court, like both expert appraisers, separately calculated the land values of the three tracts and then determined the value of the depreciated improvements. Nothing in the court's opinion suggests it deviated from the method used by Harold's appraiser, or that it added extra value for the two smaller tracts to the value of the main tract. Harold has placed too much importance on the discrepancy between its land-to-building ratio analysis and the county's land-to-building ratio analysis. The county's appraiser testified that his decision to value each tract separately was independent of his land-to-building ratio calculation. The court was able to observe these differences and assess their significance as they related to the market value of the subject property. We conclude the court did not erroneously double-count the value of the subject property's two smaller tracts.

Harold also argues the tax court erred by relying too heavily on the cost approach and by placing less weight on the market approach. Harold argues the high level of depreciation suffered by the improvements makes the cost approach unreliable in this case. The court's analysis, however, reveals that its reliance on the cost approach, over the market approach, was not clearly erroneous. The court noted that most of the subject property's value was in the land. It used a sales comparison approach to determine the value of the land for purposes of its cost approach analysis. As a result, any reliance by the court on the cost approach also includes heavy reliance on a market or sales comparison approach. Moreover, the court specifically adopted Harold's depreciated improvements value, and this value accounts for less than one-fifth of the court's entire cost approach valuation. Under these circumstances, we cannot conclude that the court's decision to place more reliance on the cost approach than on the market approach was clearly erroneous.

We emphasize that real estate appraisal is an inexact value determination. See Lewis Harris, 516 N.W.2d at 180; Montgomery Ward Co. v. County of Hennepin, 482 N.W.2d 785, 791 (Minn. 1992). The respective weight placed upon each of the three traditional approaches to value depends on the reliability of the data and the nature of the property being valued. This court has held that "[w]hatever weight priority may usually attach to each approach, the priority and quantum of reliance depends on the facts of each case." Lewis Harris, 516 N.W.2d at 180. In Montgomery Ward Co., this court stated "[t]he Tax Court brings its own expertise and judgment to the hearing, and its valuation need not be the same as that of any particular expert as long as it is within permissible limits and has meaningful and adequate evidentiary support." 482 N.W.2d at 791 (citing Northerly Centre Corp., 311 Minn. at 341-42, 248 N.W.2d at 927).

In the present case, the tax court heard the testimony of the two expert appraisers, viewed the subject property, and reviewed detailed post-trial briefs. The court discussed all three of the traditional approaches to market value and the methodology used by each expert appraiser. It reached a reasoned final value, using the values provided by the two experts. Under these circumstances, we cannot say that its decision was clearly erroneous.

Affirmed.