identifying historical and recent land-cover changes in

Identifying historical and recent land-cover changes in Kansasusing post-classification change detection techniques

DANA L. PETERSON,1 STEPHEN L. EGBERT,1,2 KEVIN P. PRICE1,2 AND EDWARD A.MARTINKO1,3

1. Kansas Applied Remote Sensing Program, University of Kansas, Lawrence, KS66047 ([email protected])

2. Geography Department, University of Kansas, Lawrence, Kansas 660453. Department of Ecology and Evolutionary Biology, University of Kansas, Lawrence,

Kansas 66045

Statewide land-cover change detection analysis provides a useful tool for conservationplanning and environmental monitoring and addresses issues of habitat fragmentationand urban sprawl. Furthermore, land-cover data offer a historical and recentperspective on landscape dynamics. To this end, the first alliance level land-cover mapof Kansas (Kansas Vegetation Map) recently completed by the KARS Program wascompared to Küchler’s Potential Natural Vegetation map and the 1993 Kansas LandCover Patterns map. The post-classification change detection technique was used alongwith co-occurrence matrices to identify areas and directions of land-cover change.

Comparisons showed that the land cover of Kansas has changed drastically sinceEuropean settlement. Over 48% of the land is now cultivated and native vegetationtypes such as tallgrass and shortgrass prairie have been reduced dramatically in area.There are, however, millions of ha of these vegetation types remaining in Kansas.Comparisons between the two recent land-cover maps reveal that over 80% of the landin Kansas has remained unchanged in the five years between map development. Recentland-cover changes include conversion of grassland to cropland, cropland to grassland,and grassland to woodland. Many areas changing from cropland to grassland havebeen identified as land being enrolled in the Conservation Reserve Program (CRP).Post-classification change detection analysis also shows that forest and woodland typeshave increased over the five-year period and over 1 million ha of grassland have beenconverted to cropland. The magnitude of increases in woodland and forest isquestionable, however, and may be due to registration errors and classificationmethodologies used to generate the land-cover maps.

Keywords: remote sensing, change detection, land use, land cover.

TRANSACTIONS OF THE KANSAS

ACADEMY OF SCIENCE

Vol. 107, no. 3/4p. 105-118 (2004)

INTRODUCTION

Land-cover mapping using remotely senseddata not only provides a current inventory ofresources and land use, but also provides anopportunity to identify and monitor changingpatterns in the landscape. Traditionalmethods for monitoring land-cover changerely on field data and aerial photography;however, these traditional methods can be

costly and time consuming for relatively largeareas. For statewide or regional studies, theuse of remote sensing and geographicinformation systems (GIS) provides cost-effective tools and repeatable methods formonitoring land-cover change. Changedetection techniques based on remotely senseddata have been used to identify changes invariety of aquatic and terrestrial environmentsincluding coastal, agricultural, forested, and

urban areas (Berlanga-Robles and Ruiz-Luna2002; Howarth and Wickware 1981; Mas1999; Palandro et al. 2003). Loveland et al.(1999) are working currently on the firstnational-level change detection project. Theland Cover Trends Project uses five dates ofsatellite data to assess land-use and land-cover changes occurring during the past 30years. The Environmental Protection Agencyand the United States Geological Survey(USGS) are collaborating in this researcheffort to identify rates, causes, andconsequences of land-use and land-coverchange for the coterminous United States.

The post-classification change detectiontechnique is one of many change detectionmethods used by remote sensing scientists.The post-classification comparison techniquecompares, on a pixel-by-pixel basis, multiplemaps created from remotely sensed datacollected at different times. This changedetection technique not only identifies areasof change, but also provides directionalinformation of the observed change (Jensen1996).

The Kansas Applied Remote Sensing (KARS)program recently completed the KansasVegetation Map (KV) (Egbert et al. 2001;Stewart et al. 2000), an alliance level land-cover map as part of the Kansas Gap AnalysisProject (GAP). To provide a historicalperspective of land-cover change, Küchler’sPotential Natural Vegetation (KPNV) map(Küchler 1974) was compared with theKansas Vegetation Map. Küchler’s PotentialNatural Vegetation map provides estimatedland-cover information prior to Europeansettlement. The comparison provides insightto anthropogenic and naturally driven land-cover changes in Kansas over approximatelythe past 150 years.

In 1993, the KARS program completed theKansas Land Cover Patterns (KLCP) map, amodified Anderson Level I classificationderived from Landsat Thematic Mapper

106 Peterson, Egbert, Price and Martinko

imagery (Whistler et al. 1995). Comparingthe two land-cover maps generated by KARSprovides an opportunity to monitor morerecent land-cover changes in Kansas. Post-classification change detection analysis wasperformed using the georegistered land-covermaps of Kansas.

METHODOLOGY

Mapping historical land-cover change

A comparison of the KPNV and KV mapsprovides a general overview of historical land-cover change in Kansas. A 1:800,000 KPNVprinted map was manually digitized, resultingin a digital vector file. The vector file wasthen converted to a raster data file with a 30 mcell size for map comparison purposes. Giventhe small scale of the KPNV map, digitizingefforts likely resulted in linear inaccuraciesof the defined vegetation boundaries.Nonetheless, the vast spatial extent of thevegetation regions defined somewhatsubjectively by Küchler, still provides anaccurate, regional account of historical land-cover change when compared to the KV map.

For comparison purposes, the KPNV map wasrecoded to eight classes (shortgrass prairie,mixed prairie, tallgrass prairie, oak-hickoryforest, floodplain vegetation, transition ofmixed and tallgrass prairie, mosaic oftallgrass prairie and forest, and lakes andreservoirs). Next, the KV map was recoded tobest match the corresponding KPNV classes.A co-occurrence matrix was generated toquantify agreement (no change) anddisagreement (change) between the two land-cover maps.

Mapping recent land-cover change

Comparing the KPNV data with the KV dataillustrates how the Kansas landscape haschanged since European settlement, but failsto identify the most recent land-coverchanges. Therefore, the KLCP map wascompared with the KV land-cover map to

Transactions of the Kansas Academy of Science 107(3/4), 2004 107

identify recent trends in land-cover change.Since the KLCP classes were more generalthan the KV map, the KV land-cover classeswere recoded to correspond to the KLCPland-cover classes.

A co-occurrence matrix quantified agreementand disagreement between the two land-covermaps. Disagreement between the two land-cover maps indicates potential recent changesin land-cover. While a co-occurrence matrixquantifies change, it does not provide a visualrepresentation of land-use/land-cover change.Therefore, a change detection matrix functionwas used to generate an image with pixelscoded to reflect directional change. Eachcombination of change or no change wascoded to a unique value. For example, pixelsmapped grasslands in the KLCP map andcropland in the KV map were coded as class 2on the change map where as pixels mapped ascropland in both land-cover maps were codedas class 7.

Kansas vegetation maps

Küchler’s Potential Natural Vegetation(KPNV) map. Küchler’s potential vegetationclassification system divides Kansas’ land-cover into four major classes: prairies; forests;mosaics, transitions, and boundaries; andlakes and reservoirs. Within each vegetationclass, Küchler divides the vegetation into twolevels of subclasses (Fig. 1). In the firstsubclass level, the prairie class is divided intothree zones from east to west: tallgrassprairies, mixed prairies and shortgrassprairies. The forest class consists of oak-hickory forests and floodplain vegetation. Themosaics, transitions, and boundaries classconsists of transitional zones of vegetation ormosaics of multiple vegetation types. Küchlermapped the potential vegetation of Kansasusing library sources along with field andlaboratory data (Küchler 1974). Geology,soils, topography, and climate data combinedwith field observation were key to thedevelopment of this vegetation map.

On Küchler’s map, oak-hickory forests areconcentrated in the eastern third of the state,and floodplain vegetation is concentratedalong waterways throughout the state (Fig. 1).The prairies divide Kansas into three east-west zones. Shortgrass prairie dominates thewestern third of Kansas, mixed prairiedominates the central third, and tallgrassprairie the eastern third of Kansas. Mixedprairie intermixes with shortgrass prairie inthe west and tallgrass prairie in the east. Sandprairies are found in the south-central andsouthwestern regions of Kansas. As its namesuggests, sand prairie typically occurs onsandy soils and often intermixes with theother prairie types in the area. Of the fourprairie types, sand prairie is the smallestmapped prairie type.

Kansas Land Cover Patterns (KLCP) map.The Kansas Land Cover Patterns map wascreated using an unsupervised classificationapproach on single-date (16 images) LandsatThematic Mapper (TM) imagery dating from1988 to 1991 (Whistler et al. 1995). The TMdata were subset to the county level creating105 subsets. Unsupervised classification basedon ISODATA clustering was used to map tenland-use/land-cover classes: five urbanclasses (residential, commercial/industrial,open land, woodland, and water) and five ruralclasses (cropland, rangeland, woodland, water,and other) (Fig. 2). The thematic image foreach county was then generalized to a 2hectare (ha) minimum mapping unit (MMU).The county land-cover maps were thengeometrically rectified to geographiccoordinates and edge matched to create astatewide land-cover map.

Classification accuracy assessments wereconducted on a county-level basis. Aerialphotography was compared to the county-level maps to determine map accuracy levels.Sample sites were systematically selectedwith a minimum of 5% of the total countyarea sampled. The sample sites were manuallydigitized and interpreted from the aerial


photography. The statewide land-cover maphad an overall accuracy of 91% (Whistler etal. 1995).

The KCLP map shows that cropland andgrasslands are the two dominant land-covertypes in Kansas (Fig. 2). Woodland covers arelatively small area and is concentrated inthe eastern half of Kansas and along streamcorridors in central and western Kansas.Cropland dominates central and westernKansas, more so than in eastern Kansas.Large extents of grassland are scatteredthroughout the state with the largest extentlocated in the Flint Hills.

Kansas Vegetation (KV) map. The KansasVegetation Map was created using a hybridtwo-stage classification approach onmultitemporal Landsat Thematic Mapperdata. Forty-eight Landsat Thematic Mapperscenes dating from 1992 to 1996 wereacquired to provide multitemporal (spring,summer, and fall) coverage of the state. Forty-one land-cover classes were mapped including38 alliance-level vegetation classes, cropland,urban and water (Lauver et al. 1999; Egbert etal. 2001) (Fig. 3). An alliance is defined as anaggregation of community types where thesame dominant species and physiognomyexist (Lauver et al. 1999). In the vegetationclassification hierarchy, the alliance levelresides between the formation level andcommunity level with the community levelbeing more detailed.

Land-use/land-cover in each multitemporalimage was mapped using a two-stage hybridclassification approach. The first stage(unsupervised classification) separatedcropland from natural vegetation. The secondstage (supervised) mapped the naturalvegetation class into the 38 alliance-levelvegetation classes. Over 3500 field sitesvisited during the summers of 1996 to 1998were used to train the classification algorithm.Following the supervised classificationprocess, the land-cover maps were generalized

to a minimum mapping unit of 2 ha. Lastly,the sixteen image-based vegetation maps weremosaicked to create a statewide vegetationmap.

A number of post-hoc techniques were used torefine the statewide map as determined byecologists/biologists and by remote sensingscientists from the Kansas Biological Surveyand KARS Program, respectively.Refinements were conducted on a statewide,regional, county, or scene-by-scene basis,depending on the refinement performed.Ancillary data such as geology, soils, andhydrography were used to refine the land-cover map.

Map accuracy levels were assessed by land-cover type and were reported along withoverall accuracy, and the Kappa statistic.Accuracy levels were calculated by comparingthe classified data with 828 field sites thatwere collected throughout the state in 2000.Error matrices were generated at threeclassification levels: the alliance level,formation, and Anderson Level I. Overallaccuracy levels for the state at AndersonLevel I, formation, and alliance level were88%, 65%, and 52%, respectively (Egbert etal. 2001). An Anderson Level I classificationis a commonly used land-use and land-coverUSGS classification system consisting ofnine general classes (urban, agriculture,rangeland, forest, water, wetland, barren,tundra, and perennial snow or ice).Anderson’s classification system is ahierarchy of classification levels (e.g.Anderson Level II and Level III) with moredetailed land-use/land-cover classes with eachnumeric increase (Anderson et al. 1976).

The Kansas Vegetation Map is the firstalliance-level land-cover map of Kansas. Themap indicates cropland and grasslanddominate the Kansas landscape (Fig. 3). Ofthe grassland types remaining in Kansas,there is a mixture of non-native grasslands,former cropland enrolled in the Conservation


Reserve Program and native prairies includingtallgrass, mixed, and shortgrass prairie thatfollow an east-west gradient similar to theKPNV and KLCP maps. Forest and woodlandland-cover types are concentrated in easternKansas and along stream corridors in centraland western Kansas.

RESULTS AND DISCUSSION

Historical land-cover change

According to the KPNV map, tallgrass prairie(including mosaic and transition zones) oncecovered up to 36% of the state (Table 1). TheKV map shows that nearly half of thepotential tallgrass prairie region has changedto either cropland, non-native grassland, or toland enrolled in the CRP (Fig. 4c; Table 2).Of the three major prairie types shown,tallgrass prairie has the highest percentagechanging to non-native grasslands (~11%).Other dominant alliance level land-cover

types within the tallgrass prairie regioninclude oak-hickory forest (~2%) and ash-elm-hackberry floodplain forest (~2%). TheKV database estimates that 2.8 million ha oftallgrass prairie remain and cover 13.1% ofthe state (Fig. 4c).

Figure 1. Küchler’s (1974) Potential Natural Vegetation (KPNV) map of Kansas. Classes arecoded to dominant prairie and forest types.

Table 1. Statewide areas and percentagesobtained from the KPNV database.


The KPNV map indicates mixed and tallgrassprairie formerly covered similar percentagesof the state (28.1% and 36%, respectively).The KV data show fewer hectares of mixedthan tallgrass prairie remain in Kansas; 2.01million ha of mixed prairie remain in Kansas,covering 9.8% of the state. According to theco-occurrence matrix, more mixed prairie hasbeen converted to cropland than tallgrassprairie. This is because most of the remainingtallgrass prairie is located on rocky substrate(e.g. in the Flint Hills), making it lessdesirable for cultivation and more suitable forlivestock grazing purposes. Other dominantalliance vegetation types currently foundwithin the mixed prairie region includewestern wheatgrass prairie (~4%) and mixedprairie disturbed (~2%) (Fig. 4b).

Additionally, the KPNV map indicatesshortgrass prairie once covered 16.7% ofKansas. Over 76% of this area is currently

used as cropland and over 8% is enrolled inCRP; the highest percent of CRP occurringamong the three dominant prairie regions(Table 2). The KV land-cover map shows thatover 757,000 ha of shortgrass prairie remain,covering almost 9% of the state (Fig. 4a).Other native grassland and shrublandvegetation types also occur within this regionincluding western wheatgrass prairie (~1%remains) and sandsage shrubland (~2%remains) (Fig. 4a).

Recent land-cover change

Both KLCP and KV show that cropland andgrasslands dominate the Kansas landscape(Table 3). The KLCP database mapped 52.8%of Kansas as cropland and 42.6% asgrasslands. The KV database mapped 48.4% ascropland and 41.7% as grasslands. The KLCPdatabase mapped 2.7% of Kansas as forest/woodland while the KV database mapped

Figure 2. Kansas Land-Cover Patterns (KLCP) map produced using single date Landsat ThematicMapper imagery. For the change detection analysis, the number of land-use/land-cover classeswas recoded from ten to six.


8.0%; the largest difference in percentagesmapped between the two databases. Urban andwater land-cover types increased in the KV

map slightly, suggesting that statewide urbanexpansion may be occurring at a relativelyslow pace.

Figure 3. Alliance level Kansas Vegetation (KV) map. The map contains 41 land-use/land-coverclasses and was derived using multitemporal Landsat Thematic Mapper satellite imagery. Thelegend shows dominant land-cover types.

Table 2. Co-occurrence matrix highlighting historical changes in land cover.


Figure 4. KV alliance level land-use/land-cover types mapped within Küchler’s (a) shortgrassprairie, (b) mixed prairie, and (c) tallgrass prairie regions.

Figure 5. Post-classification change detection image showing disagreement between the KLCP andKV maps.


The co-occurrence matrix quantified that 80.3%of Kansas has remained unchanged and 19.7%has potentially changed (Table 4). The changedetection image illustrates the spatial extent ofthe differences or changes in land-cover typespresented in the co-occurrence matrix (Fig. 5).The image shows that differences or areas ofdisagreement between the KLCP and KV dataoccur throughout Kansas. The largestdisagreement occurred between grassland andcropland vegetation classes.

In the change detection image most areas,especially in western Kansas, changing fromcropland to grassland appear as relativelylarge contiguous blocks of land, an expectedpattern if large tracts of cropland werereseeded to grassland (Fig. 6). A previous

study by Egbert et al. (1998) showed that post-classification change detection accuratelydelineated CRP lands in Finney County,Kansas. For our study, areas changing fromcropland to grassland or “predicted CRP” werecalculated for fifteen selected counties acrossthe state and compared to reported enrollments(USDA 2000) (Table 5). Several countiesshowed that “predicted CRP” closely matchedreported CRP enrollments. For the most part,however, “predicted CRP” overestimatedreported CRP enrollments (Table 5).

Just as the change detection image showedareas changing from cropland to grassland,there were also areas changing from grasslandto cropland. It should be noted that CRPenrollments do not automatically constituteone-for-one replacements of cropland withgrassland. Conservation programs such asCRP frequently suffer from “slippage,” acondition describing the simultaneousconversion of grasslands and fallow fields tocropland by producers to offset the cropacreage lost to conservation programs (e.g.Erickson and Collins 1985; Riddell and Skold1997). According to Leathers and Harrington(2000), slippage rates attributable primarilyto CRP in 14 counties in southwestern Kansasaveraged 53% annually from 1988 to 1994.

While it is probable that grasslands have beenplowed and cultivated in the past 8-10 years, it

Table 3. Statewide land-cover estimates fromthe KLCP and KV databases.

Table 4. Co-occurrence matrix comparing the KLCP and KV databases.


is unlikely that slippage accounts for 1.06million ha that have been converted asindicated in the change detection image. Oneshould not assume all areas of disagreementbetween the two databases indicate true land-

cover change. Registration errors, differencesin map class definitions, and imageclassification errors may also be contributingfactors. Registration errors (pixelmisalignment) result from comparing two

Figure 6. Potential CRP land (shown in black) in southwestern Kansas identified from the post-classification change detection image.

Figure 7. Small slivers of disagreement occurring along cropland or grassland field boundariesmay not indicate areas of change, but are likely registration errors between the two maps.


spatial databases created at different timesusing different preprocessing techniques(Gordon 1980). In this study, registrationerrors may account for a large percentage ofthe area reported as changing from grasslandto cropland. Using a generalization technique,we eliminated areas changing from grasslandto cropland that were less than 2 ha. Most ofthese areas were small slivers occurring alongfield boundaries (Fig. 7). As a result, areasshown as changing from grassland to croplanddecreased by 19%. However, a large number of

registration artifacts still remained after thegeneralization process indicating themisalignment of pixels likely account formore than the 19% of areas changing fromgrassland to cropland. While a largerthreshold would eliminate more “false”change, identifying the appropriate thresholdis somewhat subjective and could alsoeliminate areas of true change.

To determine whether classification errorscontributed to the large extents reported aschanging from grassland to cropland, 170areas shown as changing from grassland tocropland were randomly selected throughoutthe state and compared on-screen with thesatellite imagery used to generate the KVmap. Of the 170 selected sites, 95% werevisually interpreted on the satellite imagery asbeing accurately mapped as cropland in theKV map. While overall accuracy levels wererelatively high for both land-cover maps (91%and 88% for the KLCP and KV land-covermaps, respectively), the results suggest thatthe KLCP map may have misclassified someextents of cropland as grassland. The KLCPmap used a single-date classification approachwhile the KV map used a multitemporalclassification approach. Previous research atthe KARS Program has shown that amultitemporal approach increases the spectralseparability of vegetation types based ondifferences in vegetation phenologythroughout the growing season (Egbert et al.1995). The single-date classification approachmay have limited the spectral separability ofgrassland and cropland resulting inclassification errors. Furthermore, a study byWardlow and Egbert (2001) showed that theKV database closely matched the amount ofcropland in Kansas reported by the USDA.The USDA reported 10,090,775 ha ofcropland in Kansas and the KV databasemapped 10,293,104 ha during the same timeperiod.

Another major difference detected betweenthe two land-cover maps was the area

Table 5. A comparison of “predicted CRP” andreported CRP values (in hectares) for 15selected counties.

* Reported ha from 1986-1993 (Source:Economic Resource Service, 2000).


classified as forest/woodland land-cover types.The KLCP database mapped 2.7% of the stateas woodland and forest while the KV databasemapped 8.0%. A study by the USDA(Leatherberry et al. 1999) reported that from1981 to 1994, forested land in Kansasincreased by 13%. Furthermore, the studyestimated that there were 1.37 million ha offorest and land containing at least one tree peracre in Kansas in 1994. The 1.71 million haof woodland and forest mapped in the KVdatabase more closely matches the USDA’sestimate of forest and woodland in Kansasthan the 0.58 million ha mapped by the KLCPdatabase.

While an increase for forest and woodlandland-cover types due to natural vegetationgrowth (e.g. old field succession, juniperinvasion, and fire suppression) may beexpected, it is questionable that naturalvegetation growth is the single contributingfactor to the differences observed.Additionally, the KLCP database mappedareas as forest/woodland land-cover types thatthe KV database mapped as other land-covertypes. Several factors contributed to theobserved differences in mapped forest/woodland land-cover types.

Differences in the methodologies used togenerate the land-cover maps may be onecontributing factor. As previously mentioned,the KLCP land-cover map used anunsupervised classification approach.Following the unsupervised classification,remote sensing analysts overlaid each spectralclass on satellite imagery and used aerialphotos to assign a land-cover type to eachspectral class. When determining whether aspectral class should be assigned as woodland,density of the stand was one of the guidelinesused. The density of the stand had to begreater than 80% cover to be labeled aswoodland. Therefore, estimates of woodlandfrom the KLCP map appear conservative.While woodland and forests were mapped inthe KLCP database using stand density and

visual interpretation, woodland and forests inthe KV database were mapped usingsupervised classification. While post-hoctechniques were used to refine theclassification of floodplain forest andwoodland types in the land-cover map, theclassification was ultimately based on trainingsites collected in the field and the maximumlikelihood classifier. However, in an effort touse recent cloud-free data, several TM scenesused in the generation of the KV map werefrom the flood year of 1993. As a result, therewere multiple areas mapped as forest orwoodland that appear to be inundated croplandand grassland on the satellite imagery. Forexample, areas surrounding Tuttle Creek nearManhattan, Kansas were classified as forestand woodland, but upon visual interpretationof non-flood year satellite imagery, most ofthese areas were actually cropland andgrassland-cover types. Other areas that wereerroneously classified as forest/woodland-cover types include contiguous areas in theFlint Hills that appear to be grassland covertypes. It seems that such areas were spectrallyconfused with forest and/or woodland covertypes due to land management practices,namely spring burning. These results indicatethat the KV map likely overestimated forestand woodland cover types in several areas ofKansas.

CONCLUSIONS

The use of co-occurrence matrices and post-classification change detection analysissuccessfully identified historical and recentland-cover change in Kansas. Conversion ofgrassland to cropland was the most prominenthistorical land-cover change and some prairietypes have been reduced in area moredramatically than others. While there havebeen dramatic land-cover changes in Kansassince European settlement, the data show thatlarge tracts of natural vegetation still remain.

The most important recent land-cover changehas been the conversion of cropland to


grassland. Most of these areas were identifiedas land taken out of cultivation and enrolledin CRP. Meanwhile, changes in forest andwoodland cover types were attributable tonatural vegetation growth and differences inmapping techniques that resulted inconservative and liberal estimates of forestand woodland extents in the KLCP and KVmaps, respectively.

Registration errors, classificationmethodologies (i.e. unsupervised vs.supervised), dates of satellite imagery, andclass definitions used to generate the initialland-cover maps were additional factorsexamined in this paper. These factorscontributed to reported areas of change inKansas. Results from this study illustrate theimportance of understanding how land-coverdatabases are generated in order to accuratelyinterpret results from post-classificationchange detection analysis.

ACKNOWLEDGEMENTS

We thank the National Aeronautics and SpaceAdministration (NASA) for financial supportprovided through the Great Plains RegionalEarth Science Applications Center (GP-RESAC) at the Kansas Applied RemoteSensing Program of the University of Kansasand through NASA’s Earth Science Enterpriseprogram (grants NAGW 3810 and NAG 5-4990). We further acknowledge financial andother assistance from numerous state andfederal partners who made the KansasVegetation Map possible: Kansas StateUniversity; the Kansas GIS Policy Board; theKansas Department of Wildlife and Parks; theKansas Biological Survey; the NationalAeronautics and Space Administration; theBiological Resources Division of the U.S.Geological Survey; the U.S. EnvironmentalProtection Agency, Region 7; and theNational Park Service.

LITERATURE CITED

Anderson, J.R., Hardy, E.E, Roach, J.T. andWitmer, W.E. 1976. A land use and landcover classification system for use withremote sensor data. U.S. GeologicalSurvey, Professional Paper 964, 27 p.

Berlanga-Robles, C.A. and Ruiz-Luna, A.2002. Land use mapping and changedetection in the coastal zone of NorthwestMexico using remote sensing techniques.Journal of Coastal Research, 18(3), p.514-522.

Egbert, S.L., Peterson, D.L. Stewart, A.M.,Lauver, C.L., Blodgett, C.F., Price, K.P.and Martinko, E.A. 2001. The KansasGAP land cover map final report. KansasBiological Survey, Report 99, 38 p.

Egbert, S.L., Price, K.P., Nellis, M.D. andLee, R. 1995. Developing a land covermodelling protocol for the High Plainsusing multi-seasonal Thematic Mapperimagery. In Proceedings, ASPRS/ACSMAnnual Meeting, 27 February - 2 March,Charlotte, North Carolina, p. 836-845.

Egbert, S.L., Lee, R., Price, K.P., Boyce, R.,and Nellis, M.D. 1998. MappingConservation Reserve Program (CRP)grasslands using multi-seasonal ThematicMapper imagery. Geocarto International13(4), p. 17-24.

Erickson, M.H. and Collins, K. 1985.Effectiveness of acreage reductionprograms. In Agricultural Food PolicyReview, p. 166-184. Research Service,U.S. Department of Agriculture.

Gordon, S.I. 1980. Utilizing LANDSATimagery to monitor land-use change: a casestudy in Ohio. Remote Sensing ofEnvironment 9, p. 189-196.

Howarth, P.J. and Wickware, G.M. 1981.Procedures for change detection usingLandsat digital data, International Journalof Remote Sensing 2(3), p. 277-291.

Jensen, J.R. 1996. Introductory digital imageprocessing: A remote sensing perspective.


Prentice Hall, Upper Saddle River, NewJersey, p. 107-124.

Küchler, A.W. 1974. A new vegetation map ofKansas. Ecology 55, p. 586-604.

Lauver, C.L., Kindscher, K., Faber-Langendoen, D. and Schneider, R. 1999. Aclassification of the natural vegetation ofKansas. The Southwestern Naturalist44(4), p. 421-443.

Leatherberry, E.C., Schmidt, T.L., Strickler,J.K., and Aslin, R.G. 1999. An analysis ofthe forest resources of Kansas. U.S.Department Agriculture, Research PaperNC-334, 114 p.

Leathers, N. and Harrington, L.M.B. 2000.Effectiveness of conservation reserveprograms and land “slippage” insouthwestern Kansas. ProfessionalGeographer 52(1), p. 83-93.

Loveland, T.R., Sohl, T., Sayler, K., Gallant,A., Dwyer, J., Vogelmann, J., Zylstra, G.,Wade, T., Edmonds, C., Chaloud, D. andJones, K.B. 1999. U.S. land use and landcover trends: rates, causes, andconsequence of late-twentieth centurychange, EPA Research Plan. <http://www.epa.gov/nerlesd1/land-sci/pdf/trend.pdf> (21 April 2003).

Mas, J.F. 1999. Monitoring land-coverchanges: a comparison of change detectiontechniques. International Journal ofRemote Sensing, 20(1), p. 139-152.

Palandro, D., Andréfouët, S., Dustan, P. andMuller-Karger, F.E. 2003. Changedetection in coral reef communities usingIkonos satellite sensor imagery and

historic aerial photographs. InternationalJournal of Remote Sensing, 24(4), p. 873-878.

Riddell, M. and Skold, M. 1997. Croplandretirement and supply control in the GreatPlains. Journal of Production Agriculture10(1), p. 106-10.

USDA, Environmental Research Service<http://www.ers.usda.gov/data/archive/96004/> (December 2000).

Stewart, A.M., Egbert, S.L., Lauver, C.L.,Martinko, E.A., Price, K.P., Peterson, D.,Park, S., Blodgett, C.F., and Cully, J. 2000.Landcover mapping for GAP: A hybridclassification approach to identifying thevegetation of Kansas. In: ASPRSTechnical Papers, 2000 AnnualConference, American SocietyPhotogrammetry & Remote Sensing,Washington, D.C., May 24-26.

Wardlow, B.D., and Egbert, S.L. 2001. Acomparison of the GAP and RegionalMRLC land cover data sets for the state ofKansas - classification systems, methodsand results. In Proceedings, ASPRS 2001:Gateway to the New Millenium. 27-23April, St. Louis, Missouri.

Whistler, J.L., Jakubauskas, M.E., Egbert,S.L., Martinko, E.A., Baumgartner, D.W.,and Lee, R. 1995. The Kansas state landcover mapping project: regional scale landuse/land cover mapping using LandsatThematic Mapper data. In Proceedings,ASPRS/ACSM Annual Meeting, 27February - 2 March, Charlotte, NorthCarolina, p. 773-785.

identifying historical and recent land-cover changes in

Documents