While the accuracy of a 1995-era classification is hard to determine based on 2000-era ground truth, the 1995-era classification is composed of approximately 96.8% of the final 2000-era classification. This is because only the areas of change between 1995 and 2000 were classified for the early-date classification. Then the classified early-date layer was overlaid onto the late-date classification. Therefore the accuracy assessment for the late-date classification is still a good estimate of the accuracy of the early-date classification. According to accuracy assessment performed by Sanborn, the overall accuracy is 85.3% and 0.863 Kappa. The accuracy results shown below are from US Coastal zone 66. A total of 625 points are located in US Coastal zone 66.
Each class accuracy is as follows:
(Errors of Omission/Commission)
0 Background (N/A)
1 Unclassified (Cloud, Shadow, etc)(N/A)
2 High Intensity Developed (50%/73%)
3 Medium Intensity Developed (79%/52%)
4 Low Intensity Developed (25%/41%)
5 Open Spaces Developed (50%/100%)
6 Cultivated Land (86%/72%)
7 Pasture/Hay (77%/73%)
8 Grassland (61%/76%)
9 Deciduous Forest (95%/88%)
10 Evergreen Forest (99%/85%)
11 Mixed Forest (80%/93%)
12 Scrub/Shrub (75%/84%)
13 Palustrine Forested Wetland (75%/75%)
14 Palustrine Scrub/Shrub Wetland (68%/84%)
15 Palustrine Emergent Wetland (91%/72%)
16 Estuarine Forested Wetland (N/A)
17 Estuarine Scrub/Shrub Wetland (N/A)
18 Estuarine Emergent Wetland (93%/100%)
19 Unconsolidated Shore (90%/95%)
20 Bare Land (79%/96%)
21 Water (100%/100%)
22 Palustrine Aquatic Bed (100%/100%)
23 Estuarine Aquatic Bed (100%/100%)
24 Tundra (N/A)
25 Snow/Ice (N/A)
The validation points were both collected in the field and photo interpreted. The accuracy assessment selection methods were developed to minimize spatial autocorrelation between the training and accuracy assessment. The first pool of accuracy assessment sites came from field data and photo interpretation of black and white digital orthophotos and digital color infrared imagery (primarily Ikonos data). These sites were collected prior to initial mapping and were collected at the same time as the training data. The sites were selected to capture the physical and spectral diversity of the land cover. After the first criteria was met, the accuracy assessment sites were buffered to see if they fell within 1000 meters of another accuracy assessment site of the same class or within 1000 meters of a training site of the same class. Those that fell within the 1000 meter buffer were eliminated. All sites were to be from a homogeneous 3x3 area. After an analysis of the point distribution, it became clear that there were not enough samples for every class. The remaining points were selected from the initial draft final classification and had to be a homogeneous 3x3 area. Sampling was limited to areas where there was high resolution color infrared imagery. The imagery included the previous Ikonos imagery, but also included an additional 20 scenes of Ikonos imagery. The additional Ikonos imagery provided sampling areas across the entire study area. Classes which had a proportionally larger representative land area had larger numbers of samples . When possible, we tried to identify 50 samples of each of the classes. Exceptions were for the following classes (actual sample numbers): Cultivated Crops (44) -Pasture/Hay (43) - Grassland/Herbaceous (31)- Scrub/Shrub (36) - Palustrine Forest Wetland (36) - Palustrine Scrub/Shrub Wetland (40) Palustrine Emergent Wetland (23)- Estuarine Emergent Wetland (15) - unconsolidated Shore (21) - Bare Land (29) - Open Water (49) - Palustrine Aquatic Bed (2) - Estuarine Aquatic Bed (14). A total of 625 accuracy assessment points were used excluding urban classes. 697 accuracy assessment points were used including urban points. All classes have a minimum of 50 accuracy assessment points except for the classes mentioned above. These classes are limited in the study area and to some extent in the imagery that was available to sample from.
Post-Processing Steps: None
Known Problems: None
Spatial Filters: None