We find that in the second case, it shows a greater similarity between A and B than in the first. This is because, although the percentage of concordance is the same, the percentage of concordance that would occur “by chance” is significantly higher in the first case (0.54 versus 0.46). An important note about the non-system index is that, although it does not overlook any distortion, it does not mean that it corresponds to the correlation. The two are not equivalent to α in the case of λ. The difference can be detected by looking at a scatter cloud and turning it. This changes their correlation, but remains the same thanks to self-cutting. In the case of λ, positive values also appear for when , which can be interpreted as a noise measure in the data. It is a simple framework between the two distributions. We will come back to the question of the threshold for accepting the agreement as good – which turns out to be around 60% for most purposes.

Nevertheless, significant guidelines have appeared in the literature. Perhaps the first Landis and Koch,[13] the values < 0 were not compliant and 0-0.20 as low, 0.21-0.40 as fair, 0.41-0.60 as moderate, 0.61-0.80 as substantial, and 0.81-1 almost perfect. However, these guidelines are not universally recognized; Landis and Koch did not provide evidence to support this, but supported them on personal opinions. It was found that these guidelines could be more harmful than useful. [14] Fleiss`s[15]:218 equally arbitrary guidelines characterize kappas from over 0.75 as excellent, 0.40 to 0.75 as just right, and below 0.40 as bad. If we can accept the use of an index based on MAE and not on MAE, we argue that the right metric should be a slightly modified version of the Mielke index. This argument comes from the idea that, for an index constructed on the basis of the structure of equation (3), the objective should be to define the denominator μ as the maximum value that the meter can take δ. It is important to find the smallest value that maximizes the meter (i.e. its supremum) in order to guarantee an index with the maximum possible sensitivity. For indices based on the MSE, it is possible to show (see Additional information) that the counter can be rewritten as follows: The results are presented in the maps in Figure 5. All maps show expected patterns of temporal correspondence: areas with a strong dynamic NDVI signal, such as northern production areas, have greater convergence than desert areas where the signal is mainly made of noise.

However, there is a big difference where each metric provides negative values: the λ map does not show negative values, the Watterson M-metric map only accepts negative values if the correlation is negative, but ji & Gallo`s AC index map shows huge areas of negative values throughout the territory. . . .