Land Cover Mapping and Habitat Analysis Distinguishing vegetation communities
Learning Objectives Understand the difference between land cover, vegetation, ecosystems, and habitat Understand the general procedure used for mapping land cover using remote sensing Understand the importance of scale and MMU. Be able to construct a classification scheme for mapping. Understand the role of photointerpretation and interpretation keys. Understand the importance of accuracy assessment in mapping.
Some definitions… Vegetation Ecosystem Land Cover Habitat All of the plant life in a particular region (e.g., the vegetation of Wyoming) or period (e.g., Pleistocene vegetation) Ecosystem An interacting system of biotic (plants, animals, microorganisms) and abiotic factors (the environment) Land Cover All of the features occupying the land surface including vegetation, unvegetated areas, natural and human affected Habitat Habitat includes the physical requirements that a species requires to live.
Mapping land cover: General procedure Scope the project Acquire the remotely sensed imagery Develop a classification scheme (map legend) Explore the area on the ground (if possible) Develop an interpretation key and/or spectral tools Create land cover units (make the map) Assess the accuracy of the product Refine as necessary
Scoping the project What is the desired mapping scale? What is the MMU? What resources are available (data, money, etc.)? How much time is available to do the project? How accessible is the map area? Etc.
Minimum Mapping Units (MMU's) The minimum mapping unit is the smallest contiguous area that will be digitized or classified on your map. Important decision because it determines what patch sizes will be retained Will have different effects on map depending on the spatial configuration of the area (e.g., is the area homogeneous or patchy) Sometimes maps have different MMUs for some types (e.g., riparian) than others MMU lower limit defined by resolution of the imagery
MMU determines which clearcuts get mapped Copper Mountain, Colorado (IKONOS image)
Acquiring imagery for mapping What qualities does the imagery need to have to accomplish the mapping task? Dates/times Spatial scale/resolution Spectral resolution (panchromatic, true color, false color IR, etc.) Geometric properties (orthorectified, vertical, oblique, etc.) Amount of overlap for stereo viewing, etc. Are suitable images available (e.g., NAIP, satellite)? Are contractors available to fly if no archive is suitable? Costs, timing, etc.
Land Cover Classification Schemes How many types do you want to map? How should you divide up the features you are interested in? What resources do you have? How will your interpretation be used? What do the funders want! Can be very controversial!
How would you classify this produce?
Characteristics of classification schemes Must be useful Types must be detectable using the data you have Should be hierarchical Categories must be mutually exclusive Require explicit definitions of each class
Example of a simple hierarchy Vegetated Forest Evergreen Spruce-fir forest i. Spruce-fir with winterberry understory Lodgepole pine forest etc. Deciduous Shrubland Non-Vegetated
Exploring the ground (field recon) Critical for understanding the distribution of land cover in the real world Helps you choose useful ancillary data Useful for understanding images back at the computer Nice to get outside!
How would you classify this vegetation and where are the boundaries??
Interpretation keys Interpretation keys should include the following: Explicit written description of each type in the classification Examples of each type as it appears in the aerial photography being used A list of key features of the type that can be used to distinguish it from other types Can take the form of a dichotomous key like you would use for keying out plants or animals Should be organized into an easy-to-use notebook or in some kind of digital format (e.g., hyperlinked web key)
Image interpretation key for forest types in Vermont
Photointerpretation/classification Classification can take two basic forms: Manual photointerpretation of imagery On hard copy using clear overlays or other tools On computer screen by digitizing with mouse On tablet using a stylus Per-pixel classification of digital imagery Not used as commonly with aerial photography because the spectral data are not as good as in multispectral satellite imagery. Will talk about this more in the context of satellite data
Detailed view of Wyoming GAP Land Cover Map
Wetland community mapping – University of Wisconsin
Horticulture map at University of Wisconsin Individual shrubs delimited with carefully rectified orthophoto
Assessing accuracy Accuracy assessment is crucial for any mapping project Requires extensive field work to compare what is on the ground to what is designated on the interpreted photo Can be expensive – field work costs money Can be quantified using a suite of accuracy assessment metrics
Product refinement If product accuracy is not sufficient, must refine Assess the original data – are problems related to the information content of the imagery? Assess the interpretive process – are problems related to inconsistent or poor image classification? Does the interpretive key need to be revised/amended? Assess the classification scheme – is the list of types appropriate and mappable. Would moving “up” the hierarchy be useful and if so would the map still be appropriate for users?
Using land cover data to create habitat maps Habitat is usually a combination of land cover with other spatially distributed environmental drivers Climate Soils Proximity to water Availability of cover Contiguity of cover types needed for different parts of the life cycle Availability of food/prey Etc. Typically combine these as spatial layers in a GIS
GIS Habitat model schematic for Wild Boar
Summary Land cover is the collection of features occupying the land surface Air photos are useful for mapping land cover, especially at fine scales Land cover mapping is a process with a series of important steps, each of which must be carefully executed Our ability to accurately map land cover can be limited by the quality of the photography or other data that we are interpreting Important to be realistic about what can be accomplished