diff --git a/joss/paper.md b/joss/paper.md
index af87980..de81c91 100644
--- a/joss/paper.md
+++ b/joss/paper.md
@@ -82,9 +82,6 @@ The data processing for data category 4 corresponds to a bottom-up approach. Her
 
 If no heat demand input data is available, the heat demand can be estimated using cultural data such as population density, landuse, and building-specific heat usage [@novosel; @meha] which will be implemented in a later development stage.
 
-![The main steps of the methodology to process the provided HD polygons for the heat demand data category 2 (top) and category 3 (bottom). \label{fig3}](../docs/images/fig3.png)
-
-
 ## Processing Heat Demand Map Data
 
 Heat demand maps may contain millions of cells. Evaluating each cell would not be feasible. Therefore, **PyHeatDemand** utilizes the rasterstats package [@rasterstats] returning statistical values of the heat demand map for further analysis and results reporting.
@@ -100,6 +97,8 @@ The following resources are available for **PyHeatDemand**:
 
 We welcome contributions of users in the form of questions on how to use **PyHeatDemand**, bug reports, and feature requests. 
 
+![The main steps of the methodology to process the provided HD polygons for the heat demand data category 2 (top) and category 3 (bottom). \label{fig3}](../docs/images/fig3.png)
+
 # Acknowledgements
 
 We would like to thank the open-source community for providing and constantly developing and maintaining great tools that can be combined and utilized for specific tasks such as working with heat demand data. The original codebase was developed within the framework of the Interreg NWE project DGE Rollout (Rollout for Deep Geothermal Energy) by Eileen Herbst and Elias Khashfe [@herbst]. It was rewritten and optimized for **PyHeatDemand**.