What: High quality basemaps (Google, ESRI, Mapbox, etc) If users cannot install any plugins, we suggest they switch to Module 1.2.2 Training Data Collection Using Google Earth Engine. Use of Planet and Google Earth Engine provides more flexible reference data, however the plugins are experimental and require accounts with both Earth Engine and Planet. Important Note: We suggest users starting first with QuickMap Services, as it does not require an external account and is the most straightforward to install. Here, the process is demonstrated for Colombia and for the year 2018. Users should consider the options below and decide a source of reference data that matches the time period and geographic extent of your study region. However, users of this tutorial should follow the instructions from the plugin creators to ensure proper installation. ![]() Each plugin works with QGIS Version 3.1.10. The following plugins offer access to reference imagery. The specific imagery to use will depend on your study period and region, but in general it is recommended to use as much data as possible. Luckily, there are numerous sources of high resolution reference imagery available directly within QGIS. The time of the reference imagery overlaps the input data used for classification. The target classes can be distinguished through visible interpretation. Two critical factors in the selection of reference data are: We will add to this layer in creating the training data.Ĭritical to the collection of training data is reference data, and for most purposes it is sufficient to use high-resolution imagery. Since the type is Whole number we do not need to worry about precision.Ĭlick Ok and you should see that a new layer is added to the ‘Layers’ panel on the left of the screen. ![]() The Length of the field corresponds to the number of characters you can input, and the Precision is the number of digits. Choose a folder to save it in that you can easily access and remember.Ĭhoose EPSG:4326 - WGS 84 for the projection.ĭelete the ‘id’ field by selecting it in the Fields List and selecting Remove Field.Īdd a new field of type Whole number named ‘class’ by selecting Add to fields list. Select Layer -> Create Layer -> New Shapefile Layer….Ĭhoose a filename to save your training data (in the example below I use the name ‘training_data_colombia_v1.shp’. To start, we will need to define a new shapefile layer. A similar process can be used with polygon data, but keep in mind that it is generally recommended to have more diverse training regions to minimize the effect of spatial autocorrelation. This tutorial will demonstrate how to create training data that are point geometries. Here, the process is demonstrated for the country of Colombia and for a simple legend of four land cover classes: Forest, Water, Herbaceous, and Developed. Users should adjust the various components to match their project objectives. This tutorial will demonstrate how to collect categorical training data for land cover classification using QGIS. percent tree cover) can be predicted using continuous training labels. ![]() This is an example of a categorical classification and the training labels are therefore categorical. ![]() Based on these labels, the classifier can predict the most likely land cover class for each pixel in an image. For example, land cover training data will contain examples of each class in the study’s legend. The trained classifier can then be applied to new data to create a classification. The training dataset is a labeled set of data that is used to inform or “train” a classifier. Training data is instrumental to supervised image classification.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |