Call for Papers: New Directions in Land Use/Land Cover (LULC) Analysis

Dear Colleagues,

This Special Issue on “Land Use/Land Cover (LULC) Analysis” welcomes submissions covering all areas of LULC, and papers that focus on the challenges associated with time series data, including classification, LULC change, and LULC error reporting are particularly encouraged.

The discipline of LULC has evolved from simply producing classifications of LULC at single points in time (e.g., national surveys for a particular year) to the generation of products to support and inform policy, which address issues regarding sustainability, vegetation health, and provides key inputs into climate change analyses and models. As a result, it embraces both traditional mapping of land cover (e.g., for forest resource inventories, REDD+) and land use (e.g., for modelling urbanization), and a number of related areas including ecosystem service, landscape function, land characterization, many of which are explicitly concerned with linking human distribution, vegetation condition/disturbance, as well as economics. This has led to increased interest in the multi-dimensional aspects of land use, as well as land cover.

A number of methodological challenges and opportunities have arisen due to the increased availability and volumes of remote sensing data (Big Data), the focus on LULC applications that link to landscape process and function and the increased demand for higher temporal resolution information, requiring time-series LULC analyses and analyses of LULC change. These challenges include how LULC are classified, how LULC change is measured, and how spatiotemporal error/accuracy in LULC are measured. They suggest the need to revisit some of the traditional methods used in these areas and to identify future opportunities and directions, especially as LULC is in the Big Data era.

For example, typically, LULC classifications are generated for a single point in time using data just for that time period (e.g., LULC classification for a particular year). The increased availability of time series data provides an opportunity to develop more temporally nuanced and informed approaches to classification. A standard single time classification is developed using the highest class likelihoods arising from data just that time period, independent of any other information. Error analysis is typically undertaken by comparing predictive against observed class from some validation exercise. Most current approaches to change analyses compare ‘predictions’ from different time periods (post-classification change). This suggests some interesting questions and opportunities:

  • Can multiple class likelihoods from a time series of likelihoods be used in classification instead of those from a single point in time?
  • Can information about temporal process be included in or even imposed on the classification? This is an old idea (see Comber et al. (2004)) but has its time come?
  • Can knowledge of local succession sequences, transitions or information from other LULC data be included?
  • How should we measure and informatively report on multi-temporal changes in LULC (including landscape and ecosystem function, characterization and services)?
  • Are new conventions needed for reporting change and error this based around soft measures (e.g., fuzzy sets, fuzzy change—see Fisher (2010))? For example, how should changes in the degree of ecosystem service be identified? Do soft classifications help in this matter? How do they alter the way that LULC change and error is approached?


Comber, A.J.; Law, A.N.R.; Lishman, J.R. Application of knowledge for automated land cover change monitoring. Int. J. Remote Sens. 2004, 25, 3177–3192.

Fisher, P.F. Remote sensing of land cover classes as type 2 fuzzy sets. Remote Sens. Environ. 2010, 114, 309–321.

Prof. Dr. Alexis  Comber
Guest Editor