1.1 Use of Satellite Imagery and SAE

One of the pioneering articles in small area estimation was the paper by Singh, R, et al. (2002), which addressed crop yield estimation for the tehsils (sub-administrative units) of Rohtak district in Haryana, India.

Raster images represent the world through a set of contiguous equally spaced cells known as pixels. These images contain information like a geographic information system and a coordinate reference system. Images store an identifier, a value in each pixel (or a vector with different values), and each cell is associated with a color scale.

Images can be obtained in raw and processed forms. The former contains only color layers, while the latter also contains values that have been processed in each cell (vegetation indices, light intensity, type of vegetation).

Raw information can be used to train desired features (roads, crop types, forest/non-forest). Fortunately, in Google Earth Engine, we find many processed indicators associated with a pixel. These indicators can be aggregated at a geographical area level.