Enhanced Intensity Analysis to Quantify Categorical Change and to Identify Suspicious Land Transitions: A Case Study of Nanchang, China

Abstract

Conventional methods to analyze a transition matrix do not offer in-depth signals concerning land changes. The land change community needs an effective approach to visualize both the size and intensity of land transitions while considering possible map errors. We propose a framework that integrates error analysis, intensity analysis, and difference components, and then uses the framework to analyze land change in Nanchang, the capital city of Jiangxi province, China. We used remotely sensed data for six categories at four time points: 1989, 2000, 2008, and 2016. We had a confusion matrix for only 2016, which estimated that the map of 2016 had a 12% error, while the temporal difference during 2008–2016 was 22% of the spatial extent. Our tools revealed suspected errors at other years by analyzing the patterns of temporal difference. For example, the largest component of temporal difference was exchange, which could indicate map errors. Our framework identified categories that gained during one time interval then lost during the subsequent time interval, which raised the suspicion of map error. This proposed framework facilitated visualization of the size and intensity of land transitions while illustrating possible map errors that the profession routinely ignores.

Landscape Changes in the Southern Coalfields of West Virginia: Multi-Level Intensity Analysis and Surface Mining Transitions in the Headwaters of the Coal River from 1976 to 2016

“…Through aggregation across two time points, errors in the LCCs may produce mismatches that may be incorrectly identified as change. Indeed, LC products with levels of overall accuracy generally considered as satisfactory or adequate (≥85%), when combined, may be inadequate for accurate change detection [6]; an issue of particular concern for classes with low accuracy [53].…”

“…The Intensity Analysis framework has been recently updated [58] to include the Difference Components method and the Quantity and Allocation disagreement metrics [59]. Xie et al (2020) [6] integrated Intensity Analysis with Difference Components and error analysis in a unique framework to enable the evaluation of changes among LC classes and distinguish possible errors among land transitions.…”

“…Figure 4 reports the plots derived using PontiusMatrix42 [74], including the false alarm (commission or false positive) and miss (omission or false negative) components of error divided into the three parameters of Quantity, Exchange and Shift disagreement. PontiusMatrix42 allowed estimation and Comparison of the size of the errors to the size of the unbalanced samples used to validate the LCC [6].…”

“…They may contain suspected amounts of error that are difficult to assess when combined in multiple classifications. We will focus on this class more carefully in the results phase, using the framework proposed by [6] and accompanying the results' discussion with the high-resolution aerial imagery's visual assessment.…”

“…To achieve this goal, the Multi-Level Intensity Land-change Analysis method [5] was applied to examine land-change processes in the Land 2021, 10,748 Headwaters of the Coal River using high-resolution images (1-2 m) across two time intervals (1976-1996 and 1996-2016) spanning a total of 40 years. The comparison of the magnitude of the errors at each time point with the magnitude of temporal differences [6] applied in this study enabled the identification of the potential sources of errors to avoid misinterpretations of the land-change analysis results. Three land cover classifications (LCCs), consisting of five classes (developed areas, barren land, water, low vegetation, and forest), were derived from 1 and 2 m spatial resolution imagery over three time points (1976, 1996, and 2016).…”