Technology News: Digital Soil Mapping and Ecological State Change
Submitted by mrlevi21 on Mon, 04/21/2014 - 13:54
The path toward environmental sustainability and ecological resilience starts with maps.
Different land areas present different risks and opportunities, so we need to be able to classify those land areas and know where the classes occur. One new approach to this task fuses soil science and vegetation ecology by linking digital soil mapping (DSM) to state-and-transition models (STMs). DSM generally refers to any approach linking environmental data, such as Landsat reflectance or topographic variables from digital elevation models, to soil types or soil properties. New sensor technologies and statistical approaches continue to improve predictive power and resolution. Perhaps the biggest charge to the DSM community is the GlobalSoilMap project which aims to produce a uniform dataset of gridded soil properties at multiple depths for the entire globe. The objectives of the project are to provide a dataset that can be used in planning for global food production, climate change adaptation, and restoration.
Because the goal of DSM has been to map soil, many approaches attempt to minimize vegetation signals in reflectance indices in order to isolate the influence of soil properties. While this can be helpful for detecting soil properties, it overlooks a considerable amount of information provided by vegetation about soil and by the soil about vegetation. Plant-soil relationships are especially obvious in water-limited ecosystems, which cover more than 40 % of the global land area, because plants respond to subtle variation in soil properties affecting water availability.
Ecological sites exploit these relationships to describe soils and vegetation simultaneously. Ecological sites are landscape units with similar soil, topography, and climate that can support a specific suite of plant communities or land uses. STMs linked to ecological sites describe the possible ecological ‘states’ of vegetation and ‘transitions’ between states that are most likely to occur as a result of disturbance regimes (fire, climate, grazing, etc.)1 and land uses; for example, a grassland vs. an eroding shrubland. The ecological states can be mapped using the ever-increasing availability of remotely-sensed imagery2. By mapping the spatial extent of ecological states in addition to soils, we can use an understanding of vegetation change processes and ecological thresholds in management planning.
Ecological state maps provide a framework for applying specific management practices to parts of the landscape where they will be most effective: the restoration analog of “precision agriculture”. A good example is the selection of shrub-invaded grassland units for shrub removal. A simple query for ecological states that are currently shrub-invaded with minimal erosion and sufficient water holding capacity to support perennial grass could quickly identify areas that are the best candidates.
Current methods of producing ecological state maps in the Chihuahuan Desert area are time-intensive and costly because they require hand digitizing of ecological states followed by field visits. This approach is largely dependent on existing soil survey data; thus, any discrepancies in soil survey data can emerge in ecological state maps. Many soil maps in rangeland consist of individual map units that mix soil components (distinct soil types). In some cases these soil components are similar with regard to their effects on plants so they are grouped within the same ecological site; however, soil components often represent different ecological sites resulting in different management considerations; hence, uncertainty in soil maps complicates conventional state mapping efforts.
DSM can be a useful approach to state mapping because it can differentiate unique soil components independently from existing coarse-grained soil maps (Fig. 1). Combining environmental data and spatial modeling can enhance ecological state mapping with optimal field sample designs, powerful prediction models, and estimates of model uncertainty. Advanced classification algorithms can greatly reduce the time needed to produce ecological state maps because they provide a means of grouping pixels of landscape information into similar units; thereby reducing the burden of hand digitizing. DSM approaches can also be scaled up or down to meet desired management objectives which is currently difficult to do with soil survey maps. Another benefit of coupling DSM and state mapping is the identification of vegetation response thresholds related to soil properties that may inform management decisions and improve STMs.
Linking spatial patterns with landscape processes is essential to advance our understanding of ecosystem function and to interpret land conditions. DSM and ecological state mapping provide a quantitative approach to this linkage. The concept of ‘the critical zone’, or the zone between the tops of the trees to the bottom of the water table, provides us with a new way to view the interactions of soil, geology, vegetation, fauna, and the atmosphere. In spite of increasing attention to critical zone processes, there has been little effort toward mapping land to reflect these processes. Although soil maps describe the consequences of environmental interactions for soil formation3, they are limited in their ability to provide information on those interactions directly. Therefore, a system of evaluating landscapes in an integrated fashion is needed to ensure that multiple management objectives can be met to improve land stewardship. Coupled DSM and ecological state mapping is emerging as a robust approach for evaluating pattern and process feedbacks of landscapes that may reshape the way we view and manage land.
--- Matt Levi, Postdoctoral Research Scientist, Jornada Experimental Range
1 Westoby, M., Walker, B., Noy-Meir, I., 1989. Opportunistic management for rangelands not at equilibrium. Journal of Range Management 42(4), 266-274.
2 Steele, C.M., Bestelmeyer, B.T., Burkett, L.M., Smith, P.L., Yanoff, S., 2012. Spatially explicit representation of state-and-transition models. Rangeland Ecology & Management 65(3), 213-222.
3 Jenny, H. 1941. Factors of soil formation. McGraw-Hill Book Company, Inc., New York.