The application of GIS in Remote Sensing, particularly the land cover classification has contributed a lot in the scientific community (read: What is the importance of GIS (Geographic Information Science) in Remote Sensing?) and at the same time has led to the need for accuracy assessment of the products because most scientists believe that these products are not free of uncertainties. These uncertainties may result from different sources including the atmospheric particles, clouds and cloud shadows, and even the classification approaches we adopt to classify the images. Some of the uncertainties such as atmospheric particles could be somehow rectified using atmospheric correction techniques (read: What is an Atmospheric Correction and why is it necessary?), however the uncertainties from the classification cannot be rectified and therefore, there is a need of accuracy assessment to examine how accurate our classification is.
- Congalton (2004) pointed out the need of accuracy assessment for these three reasons:
- To perform a self-evaluation and learn from own mistakes
- To compare method/algorithms/analysts quantitatively
- To use the resulting maps/spatial information in some decision-making-process
What is accuracy assessment?
“The accuracy of a land cover classification is the degree to which the map land cover agrees with the reference land cover classification (i.e. ground condition).”Stehman (2009)
In accuracy assessment, typically the land cover classification in the map is compared to the true land cover condition. This is done using the ‘ground truth’, or ‘reference data’. Since the ground truth data is not practically attainable, researchers use the reference data such as higher quality data. Again, it is not possible to obtain the reference land cover classification for the entire region of interest, therefore, a statistical sampling method is used to produce a ‘sample’ i.e. subset or portion of the region mapped, and accuracy assessment is carried out. Most common method for accuracy assessment at present is the error matrix where a confusion matrix displaying the proportion of area that is correctly classified and mis-classified for the different land cover types is produced. It is helpful in estimating overall accuracy, user and producer accuracy, errors of omission and commission, and Kappa Coefficient.
What is the need for an Accuracy Assessment?
At present, accuracy assessment has become one of the integral components of land cover classification, why, because, both the researchers and readers have become aware of the uncertainties and fallback of the remotely sensed image classification. The interpretation of the remote sensing images could be misleading if they are not analyzed and presented well. The best way to check this disparity, producing the best thematic maps, and bringing the research to a certain standard, there is a need for accuracy assessment.
- There are several other needs for accuracy assessment. Some of them are listed below:
- To quantify the total error in a spatial data
- To measure the thematic errors
- To minimize the locational errors
- To identify the error magnitude
- To provide better estimations while performing land cover change analysis
- To identify the sources of confusion and deal with ambiguity/variation in remotely sensed maps
- To identify the agreement and disagreements between the land cover classification and reference data
- To make both descriptive and analytical evaluation of the spatial data
It is in the researcher’s court to decide their sampling size and criteria based on their objectives of their research.
- Congalton, R. G. (2004). Putting the map back in map accuracy assessment. Remote sensing and GIS accuracy assessment, 1-11.
- Stehman, S. V. (2009). Sampling designs for accuracy assessment of land cover. International Journal of Remote Sensing, 30(20), 5243-5272.