Intercomparison on Four Irrigated Cropland Maps in Mainland China

Area of Science:

• Remote Sensing
• Agricultural Science
• Geographic Information Systems (GIS)

Background:

• Accurate irrigated cropland data is crucial for food security and water resource management.
• Several global irrigated cropland datasets exist, but their consistency and discrepancies remain largely unexamined.
• This lack of comparative analysis limits the effective use and improvement of these vital datasets.

Purpose of the Study:

• To conduct a comprehensive comparison of four major irrigated cropland datasets (GMIA, GRIPC, GlobCover, GFSAD) in mainland China.
• To identify areas of agreement and disagreement between the datasets, both quantitatively and spatially.
• To analyze the impact of mosaic cropland pixels on dataset accuracy, particularly for GlobCover and GFSAD.

Main Methods:

• Quantitative comparison using regression analysis to derive regional and provincial statistics.
• Spatial grid-based comparison to assess similarities and discrepancies.
• Analysis of mosaic cropland pixel contributions in GlobCover and GFSAD datasets.

Main Results:

• GMIA exhibited the lowest dispersion and best statistical correlation, followed by GRIPC.
• GlobCover and GFSAD showed similar performance, with significant contributions from mosaic pixels containing mixed information.
• Spatial agreement was higher in eastern China; GlobCover and GFSAD were the primary sources of disagreement.

Conclusions:

• A need exists for standardized quantitative classification systems and improved methods for heterogeneous agricultural regions.
• The study highlights the advantage of numerical restrictions in data analysis and the importance of addressing mosaic pixels.
• Future efforts should focus on integrating databases and developing advanced remote sensing techniques for accurate mapping of irrigation areas, especially in complex landscapes.