无法读取带.ovr/.tfw的.tif波罗的海压力指数图像，求解决方案

Hey there, let's tackle this problem head-on. First, let's clear up what those extra files are, since understanding them might help diagnose the NaN issue:

• .tfw: This is a world file—it stores geographic coordinate information for your TIFF, telling GIS tools how to map pixel positions to real-world coordinates. It's harmless and just complements the TIFF's spatial data.
• .ovr: This is an overview pyramid file—it contains lower-resolution versions of your TIFF to speed up zooming in/out in GIS software. It's also non-essential for reading the full-resolution data.

Now, onto the all-NaN problem. Here are actionable steps to diagnose and fix this:

1. First, Inspect the TIFF's Metadata #

Before diving into code, use GDAL's command-line tool to check if the file has valid data and proper metadata. Run this in your terminal:

gdalinfo "波罗的海压力指数图像.tif"

Look for these key details in the output:

• NoData Value: If this is set incorrectly, it might be masking all your data as NaN.
• Band X sections: Check for Minimum and Maximum values—if both are listed as nan, the file might be corrupted or have no valid pixel data.
• Data Type: Confirm it's a numeric type (like Float32 or Int16) that your libraries can handle.

2. Adjust NoData Handling in Rasterio/GDAL #

Sometimes the NoData value isn't auto-detected correctly. Try explicitly overriding it when reading:

For Rasterio: #

import rasterio
import numpy as np

with rasterio.open("波罗的海压力指数图像.tif") as src:
    # Print metadata to verify settings
    print("Metadata:", src.meta)
    # Read data without applying default NoData masking
    data = src.read(1, masked=False)
    # Check if there are non-NaN values
    print(f"Non-NaN pixels: {np.count_nonzero(~np.isnan(data))}")
    
    # If you suspect the NoData value is wrong, set it manually
    src.nodata = None  # Disable NoData masking entirely
    data_no_mask = src.read(1)
    print(f"Pixels after disabling NoData: {np.count_nonzero(data_no_mask)}")

For GDAL Python: #

from osgeo import gdal
import numpy as np

ds = gdal.Open("波罗的海压力指数图像.tif")
band = ds.GetRasterBand(1)
# Check the current NoData value
current_nodata = band.GetNoDataValue()
print(f"Current NoData value: {current_nodata}")

# Read data without applying NoData
data = band.ReadAsArray()
print(f"Non-NaN pixels: {np.count_nonzero(~np.isnan(data))}")

# If needed, override NoData and re-read
band.SetNoDataValue(None)
data_no_mask = band.ReadAsArray()

3. Try a Non-GIS Image Library (PIL/Pillow) #

GIS libraries like GDAL/Rasterio might be tripped up by spatial metadata issues, but PIL can read the raw pixel data directly. This helps confirm if the file actually has valid pixels:

from PIL import Image
import numpy as np

try:
    img = Image.open("波罗的海压力指数图像.tif")
    img_array = np.array(img)
    print(f"Image shape: {img_array.shape}")
    print(f"Pixel min: {np.min(img_array)}, max: {np.max(img_array)}")
    print(f"Non-zero pixels: {np.count_nonzero(img_array)}")
except Exception as e:
    print(f"PIL error: {e}")

If PIL reads valid data, the issue is likely with how GIS libraries are interpreting the TIFF's spatial metadata or NoData settings.

4. Repair the TIFF File #

If the file is corrupted (common with incomplete downloads or broken pyramids), use gdal_translate to create a fresh copy:

gdal_translate "波罗的海压力指数图像.tif" "repaired_baltic_index.tif"

This will rebuild the TIFF's internal structure and ignore any broken overview data (you can regenerate overviews later with gdaladdo if needed). Then try reading the repaired file with your original code.

5. Check for Multi-Band Data #

If your TIFF has multiple bands, you might be reading a band that's empty. Use gdalinfo to confirm the number of bands, then loop through all bands to check for valid data:

import rasterio
import numpy as np

with rasterio.open("波罗的海压力指数图像.tif") as src:
    for band_idx in range(1, src.count + 1):
        band_data = src.read(band_idx)
        non_nan_count = np.count_nonzero(~np.isnan(band_data))
        print(f"Band {band_idx} non-NaN pixels: {non_nan_count}")

If none of these steps work, it's possible the original TIFF was saved incorrectly (e.g., with all pixels set to NoData) or is severely corrupted. But in most cases, adjusting NoData settings or repairing the file should resolve the all-NaN issue.

内容的提问来源于stack exchange，提问作者Melih Durmaz