import os
import rasterio
import numpy as np

def calculate_cumulative_area(folder_path):
    total_area = 0.0
    
    for filename in os.listdir(folder_path):
        if filename.endswith('.tif') or filename.endswith('.tiff'):
            filepath = os.path.join(folder_path, filename)
            
            with rasterio.open(filepath) as src:
                # Get pixel size (resolution) in meters
                transform = src.transform
                pixel_width = transform[0]
                pixel_height = -transform[4]  # Note the negative sign to convert to positive meters
                
                # Calculate pixel area in square meters
                pixel_area_sqm = pixel_width * pixel_height
                
                # Read the data and count the valid (non-NaN) pixels
                data = src.read(1)  # Read the first band
                valid_pixels = np.count_nonzero(~np.isnan(data))
                
                # Calculate the area of valid pixels in square kilometers
                image_area_sqm = valid_pixels * pixel_area_sqm
                image_area_sqkm = image_area_sqm / 1e6  # Convert to square kilometers
                
                # Add to the total cumulative area
                total_area += image_area_sqkm
    
    return total_area

# Example usage:
folder_path = 'Downloads/gavl_processed_data/gavl_processed_data/tci/'
cumulative_area = calculate_cumulative_area(folder_path)
print(f'Total cumulative area: {cumulative_area} sq km')