weiyu184/spectralrao-monitoring

The open-source repository called spectralrao-monitoring is implemented on Python3.

A function called spectralrao receives one or more rasters as input and converts each one of them to a NumPy 2d array and performs all the steps needed to retrieve the RaoQ index and saves the output as a GeoTIFF raster in the desired path. The function supports two different modes of use: (i) "classic" calculates RaoQ on a single raster layer (or band) (ii) multi-dimensional, making use of several bands defined by the user

The window size parameter given by the user is required to define an odd-sized square rolling window to preserve the integrity of the information - the movement of the window on the whole image allowing to iterate the procedures to define RaoQ.

The parameter "na.tolerance" determines the algorithm's behavior when encountering Not a Number (NaN) values in the input image. This parameter represents the minimum number of finite values accepted, enabling the calculation of the RaoQ in any given rolling window

There are two folders for a quick implementation of the Rao's Q index using both the "classic" and the "multidimensional" approaches.

A function called threshold_method allowed to define the threshold using the secant method. This is defined using two methods available in the Python package: calculateDistance used to calculate the distance between two points. It is mandatory to compute the threshold_method to determine the interest value to define the limit to certificate a change. This last method requires as input the NumPy type difference between the RaoQ for two periods of interest, being the results is a binary map (Change-NoChange)

The binary_change method allowed to define a single binary map where the change is confirmed only when it occurred in the same pixel of two different inputs

The method, priority_classification, requires as input the two binary maps extracted using RaoQ and NDVI information and allows to define the priority of the change for each pixels of the area of interest.