Accuracy of the Michigan Rural Property Inventory (MRPI) for Historic Land Use/Land Cover Determinations.
This paper examines the accuracy of a secondary resource known as the Michigan Rural Property Inventory (MRPI). Housed at the State Archives of Michigan, Lansing, Michigan, the inventory embodied an effort to equalize property valuation across the state during the late 1930s. The collection is composed of thousands of two-sided, 8" x 10" cards that outline the land use, land cover, and vernacular architecture of rural Michigan. Those cards, and the survey effort that they represent, is the basis for property tax equalization in Michigan today.
Because virtually no research had been conducted with the Michigan Rural Property Inventory, the accuracy of the data on the cards was not known. In this paper, the authors undertook a study to examine the accuracy of the land use/land cover data represented in this Inventory. An earlier study conducted by Bauman and Westphal (1990) involved a comparison of the MRPI with two secondary historic resources (USDA Soil Conservation Service aerial photographs, 1938, and Hixson plat map, 1930). In this study, it was found that the data from the MRPI and SCS aerial photographs were far superior in detail and accuracy to standard plat maps. However, the absence of an index map for the MRPI sheets made it extremely difficult to assess the completeness, and thus the accuracy, of its data. In 1996, the State of Michigan Archive obtained the essential MRPI index maps, and the research went forward. In this study, the authors dropped the use of plat maps as a validity check on the MRPI and instead focused their attention on a comparison between the SCS 1938 aerial photographs and the MRPI.
HISTORY OF THE RURAL PROPERTY INVENTORY
Between 1935 and 1941, the State of Michigan embarked on a statewide, property tax equalization effort. As a part of a federally-funded Works Progress Administration program (Michigan WPA Project S-154) and using the skills of unemployed surveyors and civil engineers, the Michigan State Tax Commission surveyed the entire State of Michigan, excluding incorporated cities and Wayne County (which already had an unified property tax system in place). The product became known as the Michigan Rural Property Inventory (MRPI). It included "acreage property in the respective townships as well as corrected maps of recorded village plats and other subdivisions" (Michigan State Tax Commission 1936). Because of the effort, Michigan had the unique distinction of having the first statewide, uniform property assessment (Michigan State Tax Commission 1938). Three distinct products resulted from the effort. They were as follows:
The Michigan Rural Property Inventory (MRPI) Cards. The most significant product was the two-sided MRPI cards. These cards were organized in the same general fashion as the equalized property evaluation cards of today (Figures 1 and 2). The front of the card includes general information such as the year, school district, section, township and range, political township and county in which the parcel of land is located. The name of the property owner and a legal description of the property, including the total number of acres being described, are recorded on the card. Under certain circumstances, one card may describe only a portion of a parcel of property. For example, if a property is located in more than one section, it will be split up and described according to the portions located in their respective sections. If a property is extensive, several cards may have been produced to describe the property. (In that situation, the front of the card will be labeled "card ___ of ___ number of cards" above the prope rty owners name.) The front of the card also included a drawing of the floor plan of the main farmhouse, a detailed description of the home (e.g., presence or absence of utilities; structural information; additions or structures complimenting the main farmhouse), and information on road access and distance to schools and market centers.
The back of the card contained a sketch of field layout and detailed information on the types of crops in the respective fields. Information on the land use/land cover practices of the day was outlined by the types and number of acres in each crop. For example, Class A agricultural land included cropland and farmstead, wild hay, untillable pasture; Class B included special agriculture; Class C was swampland; Class D was commercial orchard, vineyards, or berries; Class E was nonagricultural (e.g., farm woodlot, cutover areas, sugarbush, road, marsh, lake, waste, or other special land types like gravel pit, parkland, etc.). The presence of roads across, or adjacent to, the property was recorded in the sketches. Buildings pertinent to the occupancy of the land were presented in great detail. Data on barns, barn additions, sheds, corn cribs, silos, pump houses, apiaries, milking parlors, and other structures identified the year constructed, materials used, alterations made, and actual foundation size. However, no ne of the buildings described are included on the sketches of the property; as a result, field verification or the use of aerial or oblique photos are necessary to establish patterns of farmstead organization and building orientation.
Presently, the State of Michigan Archive has MRPI cards for fifty-eight counties in Michigan. While it is believed that nearly all counties in the Lower Peninsula of Michigan were completed, archivist Leroy Barnett believes that several counties in the Northern Peninsula were never completed due to World War II. Presently, there is no record of the MRPI for the other twenty-five counties in the State Archives. Field personnel for the State Archive are uncertain if any of these counties retained the original cards as they updated and computerized their tax records. The earliest addition to the State Archive collection occurred in 1954, with the bulk of the cards coming into the Archives during the 1980-1990s (Barnett 1997).
Ledger Books. The contents of the MRPI cards were copied into ledger books at the Michigan Department of Treasury for State Tax Commission use. It is unclear how these books actually were used, but it is known that state property tax equalization became a reality in the early 1940s.
The whereabouts of the ledger books today is unknown. Sometime in the 1940s, the original MRPI cards were distributed to various state, county, and township offices throughout Michigan; however, no record exists on the storage and/or distribution of the ledger books.
MRPI Index Map. The third product was the index maps for the MRPI cards; these maps were organized on a political township basis for each county. Each section within a Congressional township had individual properties coded with a letter (Figure 3) that matched the identification code on the card. The organization of the coding occurred in the following fashion: the property located in the northeastern most part of a section received the letter "A"; then the assignment of letters occurred in a counterclockwise direction until the lettering system returned to the first property.
The index maps were developed by the State Tax Commission to facilitate the use of the MRPI cards. Based on cursory information contained on the index maps, it appears that the maps were produced two-to-three years after the actual survey took place (Michigan State Tax Commission 1942). It is not known if copies of these maps were distributed to local government entities, but use of the MRPI is greatly facilitated by the presence of the index map. Today, all of the MRPI index maps received from the Department of Treasury are catalogued on a county and political township basis in the State Archive, Lansing, MI.
Research in ascertaining the accuracy of the land use/land cover data on the MRPI involved a comparison with data found on 1938 SCS aerial photographs (USDA 1938) from the Center for Remote Sensing, Aerial Photograph Archive, Michigan State University, East Lansing, Michigan, using planimeter measurements. (A planimeter is an instrument for measuring the area of a plane figure by tracing its perimeter.)
A geographically stratified cross-section of twelve sections within the four geographic townships represented on the Old Mission Peninsula was conducted; for each section in the study, the first ten properties listed in the MRPI were selected for analysis. Because we were using a planimeter to establish area values for various land use/land cover from the SCS photos, any property less than two acres in size was eliminated from the analysis due to the limits of the planimeter's accuracy and the scale of the aerial photographs. We also attempted to select SCS photographs in a flight line that aligned a MRPI property near the center of the aerial photograph; this was to minimize the distortion error that increases as one moves towards the perimeter of an aerial photograph. In total, 73 properties distributed within twelve sections were analyzed. Sketch data and numerical profiles of land use/land cover from the back of the MRPI were matched against planimeter data from the 1938 aerial photographs of the same pr operty. Acreage from the 1938 aerial photographs (USDA 1938) was determined by averaging 2-3 planimeter assessments for each land use/land cover area, and then comparing them with the recorded data on the MRPI for the same areas (Figure 4; Table 1).
The resulting data was further segmented on the basis of parcel size and land use/land cover type to determine whether those factors affected the accuracy of the MRPI (Table 2).
Seventy-three properties in twelve sections of land within Peninsula Township were assessed. A comparison between planimeter readings from the aerial photographs and the recorded acreage data from the MRPI sheets found reasonably good correspondence between the two sources. On average, the data from the SCS photographs matched the data from the MRPI sheets 88% of the time; that is to say, that planimeter determinations of acreage were on average +/-12% of recorded acreage values on the MRPI sheets (Table 3). The range of corresponding values on a land use/land cover basis, parcel by parcel, was very broad--i.e., 11-99% correspondence. This finding was attributed largely to surveyor error in either recording information on the MRPI or in determining the actual ownership boundaries of a parcel. For example, in the parcel where a particular land use/land cover corresponded with only 11% of the planimeter readings, a large forested area was omitted on the MRPI for that parcel, despite the correct legal boundary d escription for the property on the front side of the inventory card.
When attempting to determine where other inaccuracies arose, no correspondence was found: (1) in terms of particular vegetative land covers and amount of error found; or (2) in parcel size for the various land use/covers (i.e., 2-20, 21-40, 41-60, [greater than]60 acres) and error size.
Other errors could have been methodological on the part of the research team. For example, failure to use a stereoscope while examining the photographs could have improved delineation of property boundaries. Likewise, failure to digitize the data instead of using a planimeter also could account for some of the error. The fact that early SCS photographs were taken during "leaf on" periods due to the need to record crop coverage, may have provided additional error. Finally, not all parcels fell in the center of an aerial photograph, so some error due to plane view distortion as one focuses on objects toward the perimeter of an aerial photograph (usually in the +/-3% range) could be expected (Hudson 2000).
The use of the 1938 SCS aerial photographs indicates that there is reasonably high accuracy (average: 88%) in the data present on the MRPI cards in terms of land use/land cover. Correspondence of the recorded field data on the cards, with the aerial photographs (range: 11%-99%), suggests that several factors could affect accuracy. For example, surveyor error in determining actual property boundaries within a parcel lead to some gross underestimations of particular land uses/land covers. In other cases, planimeter readings and/or aerial photograph quality may be accounting for the average 12% error due to areas being obscured on the photographs as a result of leaf cover, cloud cover, shadow effect, or location in the photograph (distortion effect).
Earlier research efforts (Bauman and Westphal 1990; 1997) found that plat records do little more than show the outlines of the properties in a section and ownership information; the MRPI cards, on the other hand, graphically represent the entire property, while SCS aerial photographs capture organization pattern and site layout. Furthermore, the MRPI cards provide descriptions of land uses, roads, and buildings, which can be verified with the aerial photographs. In some instances, the MRPI data aided researchers in interpreting the land cover present on the aerial photographs to a certain degree (scale of the photographs being the limiting factor). A shortcoming of both the MRPI data sets and the SCS aerial photographs is that neither provides definitive locations for the structures on the property. This was because the aerial photographs are too small in scale to identify respective buildings with the naked eye, and with leaf cover, it was difficult to locate structures (a stereoscope with magnification wou ld have improved this situation). Meanwhile, the MRPI simply failed to map the locations of the various structures listed. Because the MRPI cards recorded amenities like electricity and plumbing and building conditions, general economic and living conditions are suggested for various residents--and the region, in general. Plat maps are devoid of all this information.
Among the limitations in the use of MRPI cards is: (1) the failure to identify the proper parcel identification code on the map index; (2) the lack of information on structure locations on the site; (3) incomplete coverage of the state (only 58 of 83 counties in the state have cards on file at the State Archive, and the completeness of each counties' set of cards is unknown); and (4) no information on the inter-reliability of surveyor assessments. Whether the cards can be considered a reliable set of information on the historic nature of landscapes in particular areas of Michigan will be determined in general by the availability of MRPI cards and index maps at the State Archive.
With these thoughts in mind, the authors feel that the Michigan Rural Property Inventory cards are a valuable research tool for providing a description of land resources in the 1930s. Containing far greater data on individual properties than plat maps, the data contained therein appears to be a reasonably accurate and timely picture of property that was a part of the Michigan landscape during the Great Depression. It is hoped by the findings presented in this paper will encourage other researchers to use this important historic data set.
Sincere thanks to the State Archive for alerting us to this resource. The research team would also like to thank Professor John B. Anderton, Department of Geography, Northern Michigan University, for his thorough review and excellent suggestions for improving the content of this paper.
BARNETT, LEROY. 1997. Historical background on the acquisition of the Michigan Rural Property Inventory by the State Archive of Michigan. Conversation with the author, Lansing, MI, 14 October.
BAUMAN, GRANT, AND JOANNE WESTPHAL. 1990. Report on the Validity and Reliability of the Rural Property Inventory. Unpublished report, Landscape Architecture Program, Michigan State University: East Lansing, MI.
-----. 1997. An examination of the Validity and Reliability of the Michigan Rural Property Inventory. Paper presented at the East Lakes Geography Meeting, at Michigan State University, East Lansing, MI. October.
BUREAU OF HISTORY. 1995. Registered Centennial Farms, Peninsula Township, Grand Traverse County, MI. List of sites and property owners. Secretary of State: Lansing, MI.
HUDSON, W. 2000. Sources of aerial photography error. Lecture given at Michigan State University, E. Lansing, MI, November 3. E. Lansing, MI: Remote Sensing and Aerial Photograph Archive.
MICHIGAN STATE TAX COMMISSION. 1936. Report of the State Tax Commission and State Board of Assessors. 19th Report; Years 1935-36. Lansing, MI: Franklin DeKline Co. (state printers).
-----. 1938. Report of the State Tax Commission and State Board of Assessors. 22th Report; Years 1937-38. Lansing, MI: Franklin DeKline Co. (state printers).
-----. 1942. Report of the State Tax Commission and State Board of Assessors. 22nd Report; Years 1935-36. Lansing, MI: Franklin DeKline Co. (state printers).
-----. 1939. Michigan Rural Property Inventory Cards. Grand Traverse County, MI. Peninsula Township. Works Progress Administration project for state tax equalization. Michigan Department of Treasury: Lansing, MI. Cards found in State Archive, Lansing MI.
-----. 1942. Michigan Rural Property Inventory Cards Index Map. Grand Traverse County, MI. Peninsula Township. Works Progress Administration project for state tax equalization. Michigan Department of Treasury: Lansing, MI. Map found in State Archive: Lansing, MI.
U.S. DEPARTMENT OF AGRICULTURE. 1938. Aerial photographs of Peninsula Township, Grand Traverse County, MI. E. Lansing, MI: Aerial Photograph Archive, Remote Sensing Center.
W. W. HIXSON AND CO. 1930. Grand Traverse County, MI, Plat Book. Rockford, IL: W. W. Hixson and Co.
Rural Property Inventory vs. Aerial Photography Accuracy: Sample Planimeter recordings for parcels of land in T28N R10W Section 30. RPI Data Photograph Data Parcel Land Cover/Acreage Total Acreage Land Cover/Acreage Cropland: 26.5 Ac. Cropland: 36.4 Ac. D Orchard: 31 Ac. 63.5 Acres Orchard: 24.7 Ac. Forest: 6 Ac. Forest: 6.5 Ac. Cropland: 6.5 Ac. Cropland: 8 Ac. E Orchard: 30.5 Ac. 37 Acres Orchard: 30.6 Ac. Forest: - Ac. Forest: - Ac. Cropland: 0.5 Ac. Cropland: 1 Ac. F Orchard: 4 Ac. 4.5 Acres Orchard: 5 Ac. Forest: - Ac. Forest: 33.5 Ac. Cropland: 1 Ac. Cropland: 0.8 Ac. G Orchard: 14.5 Ac. 16.5 Acres Orchard: 20 Ac. Forest: 1 Ac. Forest: 0.8 Ac. Cropland: 11 Ac. Cropland: 15 Ac. I Orchard: 47 Ac. 62 Acres Orchard: 38 Ac. Forest: 4 Ac. Forest: 4.3 Ac. Cropland: 15 Ac. Cropland: 14.5 Ac. J Orchard: 13 Ac. 38 Acres Orchard: 15.3 Ac. Forest: 10 Ac. Forest: 7 Ac. Cropland: 22 Ac. Cropland: 25.5 Ac. L Orchard: 7.5 Ac. 32.5 Acres Orchard: 8 Ac. Forest: 3 Ac. Forest: 3 Ac. Cropland: 1 Ac. Cropland: 1 Ac. N Orchard: 9 Ac. 20.5 Acres Orchard: 14.6 Ac. Forest: 10.5 Ac. Forest: 14.6 Ac. Cropland: 4.5 Ac. Cropland: 2 Ac. P Orchard: 3 Ac. 13.5 Acres Orchard: 7.2 Ac. Forest: 6 Ac. Forest: 4 Ac. Cropland: Ac. Cropland: - Ac. Q Orchard: 8 Ac. 40 Acres Orchard: 3.6 Ac. Forest: 32 Ac. Forest: 26 Ac. Parcel Total Acreage Level of Error 95% Range D 67.6 Acres 94% Suspect E 38.6 Acres 96% Acceptable 39.5 Acres F Missing [*] Forest 11% Suspect on RPI G 21.6 Acres 76% Suspect I 57.4 Acres 93% Suspect J 36.8 Acres 97% Acceptable L 36.5 Acres 89% Suspect N 30.2 Acres 68% Suspect P 13.2 Acres 98% Good Q 29.6 Acres 74% Suspect Summary of percent correspondence of MRPI data (in acres) with SCS planimeter readings based on size categories of parcels. Parcel Size No. of Parcels Range of Correspondence [less than or equal to]20 acres 22 11%-99% 21-40 25 68%-99% 41-60 12 79%-99% [greater than or equal to] 61 11 85%-99% Parcel Size Mean Value of Median Value Correspondence Correspondence [less than or equal to]20 acres 83.1% 87% 21-40 88.0% 89% 41-60 87.0% 87% [greater than or equal to]61 91.4% 91% Summary of data from seventy-three properties in twelve Sections located on the Old Mission Peninsula, Peninsula Township, Grand Taverse County, Michigan. Location MRPI MRPI versus SCS % Correspondence Parcel ID Predominate acreage (Acres) T28N R10W S6 A Orchard 3.5/3.1 89 T28N R10W S5 A Crop 19.5/18.7 96 B Crop 14.0/12.3 88 C Crop 45.0/39.3 87 D Crop 68.5/58 85 E Crop 10.0/9.3 93 F Orchard 39.6/41.1 96 G Crop 60.0/64.8 93 H Crop 20.0/18.1 91 I Orchard 20.0/20.3 99 J Crop 14.4/12.2 85 T28N R10W S17 C Orchard 71.6/78.7 91 H Orchard 29.0/26.1 90 J Orchard 40.0/47.3 85 T28N R10W S18 A Orchard 19.5/16.1 83 E Forest 9.0/10.9 83 D Crop 79.0/70.3 89 F Orchard 37.0/27.9 75 I Forest 28.0/30.4 92 J Orchard 31.0/26.7 86 K Orchard 7.0/5.7 81 L Crop 10.0/9.4 94 T29N R10W S2 A Crop 2.8/3.9 72 B Crop 49.0/63.2 77 C Orchard 65.0/61.0 94 D Crop 61.0/49.8 82 E Forest 46.0/43.3 94 T29N R10W S16 A Crop 38.0/44.0 86 B Crop 20.0/23.1 87 F Crop 58.5/68.5 85 K Orchard 38.0/40.2 95 M1 Crop 40.0/34.8 87 M2 Crop 40.0/35.9 90 N1 Crop 40.0/36.5 91 P Forest 40.0/37.5 94 Q Crop 53.5/44.2 83 R Crop 35.0/37.3 94 T29N R10W S17 A Forest 4.0/4.2 95 B Crop 66.0/62.7 95 C Crop 114.4/107.5 94 E Crop 79.5/74.1 93 T30N R10W S27 A Crop 115.0/80.6 70 B Crop 34.5/39.4 88 H Forest 41.0/41.6 99 I Crop/Orchard 81.0/76.9 96 J Crop 9.5/9.0 95 K Orchard 31.0/31.4 99 L Crop 42.0/53.5 79 M Crop 39.5/34.6 88 N Crop 41.0/34.3 84 T30N R10W S28 A Forest 8.0/9.9 81 B Crop 14.5/9.6 66 C Crop 78.5/78.9 99 T28N R10W S30 D Orchard 63.5/67.6 94 E Orchard 37.0/38.6 96 F Crop [*] 4.5/39.5 11 G Orchard 16.5/21.6 76 I Orchard 62.0/57.4 93 J Crop 38.0/36.8 97 L Crop 32.5/36.5 89 N Forest 20.5/30.2 68 P Forest 13.5/13.2 98 Q Forest 40.0/29.6 74 T29N R10W S3 A Orchard 15.0/11.7 78 B Orchard 64.5/56.5 88 C Orchard 59.0/53.4 91 D Orchard 57.0/48.6 85 E Orchard 76.5/85.8 89 F Orchard 34.0/25.8 87 G Crop 27.0/31/3 86 I Orchard 79.0/71.4 90 J Orchard 13.5/14.8 91 K Orchard 49.0/42.4 87 (*.)missing forest acreage on MRPI
|Printer friendly Cite/link Email Feedback|
|Author:||WESTPHAL, JOANNE N.; ALBAN, GREGORY; RIES, DAVID|
|Date:||Aug 1, 2000|
|Previous Article:||An Evaluation of Techniques for Tallgrass Prairie Creation, Rose Lake Wildlife Research Area, Clinton County, Michigan.|
|Next Article:||Tax Incentives for the Environmental Cleanup of Urban Areas: Initial Considerations, Comparisons, and Challenges.|