I am downloading clipped ortho_analytic_sr image and its udm for the list of geojson geometries. I need to apply several filters (date, clear_percent, and asset) before downloading. I also need to clip the images to reduce my storage cost. However, I cannot find a way to clip the image first and apply the clear_percent filter without downloading the whole image first.

By filtering first and clipping results, since my geometry is very small, I am getting clipped images with noise despite selecting a high clear_percent.

I need in clipping the image first and select the clipped image with high clear_percent value. Thanks in advance!

This is the code I am using,

sfilter = data_filter.and_filter([

        data_filter.permission_filter(),

        data_filter.date_range_filter('acquired', gte=start_date, lte=end_date),

        data_filter.geometry_filter(geojson_geometry),

        data_filter.asset_filter('ortho_analytic_8b_sr'),

        data_filter.range_filter('clear_percent', gte=80)

    ])

    ids = []

    async def main_search():

        async with Session() as sess:

            cl = sess.client('data')

            print("Starting search...")

            items = [i async for i in cl.search(['PSScene'], sfilter, limit=0)]

            print(f'{len(items)} results found for geometry {index}')

            for item in items:

                ids.append([item['properties']['acquired'],

                          item['id'],

                          item['properties']['instrument'],

                          item['properties']['cloud_percent'],

                          item['properties']['clear_percent'],

                          item['properties']['clear_confidence_percent'],

                          {'geometry': item['geometry'], 'properties': {}}])

    await main_search()

    df_ids = gpd.GeoDataFrame(ids, columns = ['time', 'id', 'instrument', 'cloud_percent', 'clear_percent', 'clear_confidence_percent', 'geometry'])

    df_ids['geometry'] = gpd.GeoDataFrame.from_features(df_ids['geometry'])['geometry']

    df_ids = df_ids.set_geometry('geometry')

    df_ids.crs='EPSG:4326'

    #print(df_ids.head())

    feature = {'type': 'Feature', 'properties': {}, 'geometry': geojson_geometry}

    # Convert CRS to a projected CRS for accurate area calculations

    footprint = footprint.to_crs('EPSG:3395')  # Example: World Mercator

    df_ids['geometry'] = df_ids['geometry'].to_crs('EPSG:3395')

    # Calculate intersection area after CRS conversion

    overlay = df_ids.overlay(footprint, how="intersection")['geometry']

    df_ids['geometry_intersection'] = overlay.area / footprint.to_crs('EPSG:3395').loc[0, 'geometry'].area

    # Convert back to geographic CRS if needed

    df_ids = df_ids.to_crs('EPSG:4326')

    footprint = footprint.to_crs('EPSG:4326')

    df_ids = df_ids.sort_values(by=['clear_percent', 'cloud_percent'], ascending=[False, True])

    top_image = df_ids.iloc[0]

    print(top_image)

    request = order_request.build_request(

        name='test_order',

        products=[

            order_request.product(item_ids=[top_image['id']],

                                  product_bundle='analytic_sr_udm2',

                                  item_type='PSScene')

            ],

        tools=[

            order_request.harmonize_tool(target_sensor='Sentinel-2'),

            order_request.clip_tool(aoi=geojson_geometry)

        ]

    )

    #place the request with asynchronous function

    async def main():

        async with Session() as sess:

            cl = sess.client('orders')

            with reporting.StateBar(state='creating') as bar:

                #place order

                order = await cl.create_order(request)

                #wait

                bar.update(state='created', order_id=order['id'])

                await cl.wait(order['id'], callback=bar.update_state, max_attempts=0)

            #download

            await cl.download_order(order['id'], directory={dir to download}, progress_bar=True)

    await main()