Land use zones and land use patterns in the Atlantic Zone of Costa Rica

摘要: This thesis describes an approach to land use inventory at the sub-regional scale in Guacimo-Rio Jimenez-Siquirres (GRS) area Atlantic Zone of Costa Rica. Therefore, concept "land zones" is introduced. The zone (LUZ) plays a central role definition observational methodology as well for structuring dynamics use. Land described terms pattern (LUP). LUP denotes farming systems and utilization types (LUTs) occurring within zone.This formulates change that object-oriented data-driven. "Object-oriented" means expressed collection objects (land zones) with specific geometric thematic characteristics. A classification system developed so each class contains zones characteristic description, geometry, aggregation structure dynamics. handling such complex object information requires emphasis put on data model.For purposes satellite imagery aerial photos are used. these materials involves recognition. "data-driven" this case classes describe not a-priori but inductive, i.e. they result from process. data- driven strategy gain insight patterns. process unravelled into number sequentially ordered processing steps, various chapters.This consists three parts.The first part defines LUZs tool change. From comparison 1948-1952 1984 we learn LUZs, belong agricultural area, have stable boundaries. implies may serve reference monitoring change.Data farm size distribution composition were obtained by survey. show significant differences composition, basis which define That clear occur indicates LUZ serves spatial unit level. relate observed cover distributions. relation be inferred composite characteristics, can used tools inventory, if proper rules interpretation applied.The correspondence between under condition characteristics do Change 1986 1990 was investigated using imagery. Clear trends observed, when applied. These decrease cultivation maize pasture land, increase banana macadamia production reforestation. Besides changes crops, pastures plantations could indicated.The second recognition concerns identification LUZs. First, stratification GRS sub-regions described. pattern, determined size, form arrangement fields, key photo criterion identify LUZs.Once identified, their field determined. procedure per pixel classification. will guide image analyst task defining set training statistical properties suitable maximum likelihood Emphasis phases. presented here makes supervised unsupervised approaches.Special attention has been paid classes. Statistical methods different patterns Field One-way analysis variance multiple evaluated mean size. resulted five size.A hierarchical cluster performed evaluate difference To derive relevant groups corresponding results (represented dendrogram) critical distance defined. minimum (or LUZs) considered significantly respect composition. reflects accuracy map map. resulting provide LUZs.Land transformation categories mapping (also termed decision rules). Mapping assign conditional label object, whereby refers particular context. decisions. Insight gained putting order. tree LUPs. leads stepwise provides formalized description decisions LUZs.In Part Three bio-physical potentials. physiographic soil combined. boundaries corresponded high degree. But does agreement (bio-physical) Results 18 % risk degradation, while 51 potential more intensive Expert judgement determine suitability (LUTs). However, exact position type or LUT cannot scale, units being nature. introduces fundamental uncertainty statements suitability. study, therefore, exploratory character. figures denote expectations.In last chapter variation yield one plantation (representing LUZ) investigated. Soil survey explained 67 variation. Combining Landsat-TM did better estimation yields. remained %.