Geophysical measurements are commonly integrated with additional geospatial infromation in order to make appropriate agricultural management decisions. This additional geospatial information can be collected using field surveying methods, aerial photogrammetry, satellite imagery, and GPS (Global Positioning System). Field surveying is used to determine the positions or coordinates of features observed in the field. Field mapping is typically conducted when detailed surveys are required. Distances and angles are measured with surveying instruments—usually with total stations—to provide detailed x,y,z coordinates of ground features.
Aerial photogrammetry is used to develop topographic maps from a stereopair of aerial photographs. These maps can be quite detailed and at large scale, depending on the aircraft height and the aerial camera used. Topographic maps at a scale of 1:50,000 can be prepared from satellite images acquired by the French SPOT satellite because of its off-nadir viewing capability.
Satellite imagery ranging from Landsat (15, 30 m), to SPOT (5, 10, 20 m), to MODIS (250 m, 500 m, 1 km) is used for representing the earth's surface at various scales. Imagery from these satellite systems covers large areas of the earth. Land cover and land use maps can be prepared using standard image processing techniques. Other geospatial maps, such as NDVI (normalized difference vegetation map), elevation, temperature, soil moisture, snow moisture, and suspended sediment concentration maps, can also be prepared using selected bands from these satellite sensors. Other sensors provide images of the earth's surface in the microwave region at spatial resolutions ranging from 10 m to 100 m. These data are used to represent soil moisture conditions or structural rock features. GPS is used to collect point information of earth features. The data are collected in a format that can be directly input to a GIS database, using the shapefile format for ArcGIS.
GIS modeling uses a process of building models using spatial data. The GIS is a tool that can integrate different data sources, including maps, DEMs, GPS data, images, and tabular data. This makes GIS modeling particularly attractive for exploratory data analysis, data visualization, and database management. The models built with a GIS can be vector or raster based, depending more on the nature of the model, the data sources, and the computing algorithm. The distinction between raster-based or vector-based models does not prevent GIS users from integrating both types of data in the modeling process, because algorithms for converting data types are easily available in a GIS system. GIS modeling can take place within the GIS or may require linking the GIS to other computer programs, such as a statistical analysis package like SAS. Many GIS software programs have analytical functions for modeling.
There are four types of models available in a GIS system: binary models, index models, regression models, and process models (Chang, 2002). A binary model will use a logical expression to select map features from a composite map or from multiple grids, with the output being a binary map (1 [true] that satisfies a logical expression and 0 [false] for map features that do not). An index model uses an index value calculated from a composite map or multiple grids that are used to produce a ranked map. Usually the observed values of each variable on a map are evaluated and given numeric scores, then the relative importance (weighting) of a variable is evaluated against other variables to produce a final ranked map.
A regression model—either linear regression or logistic regression—is used to relate a dependent variable to a number of independent variables through the use of an equation, which can then be used for prediction or estimation. A process model integrates existing knowledge about environmental processes in the real world into a set of relationships and equations to quantify a physical process. The process model can offer both a predictive and an explanatory capability that is inherent in the proposed processes. The output from a process model is typically a set of equations that can be used for predictive purposes.
Was this article helpful?