Move figure out of prose

stories/NCEO-Africa-test.stories.mdx

@@ -26,17 +26,15 @@ The dataset consists of two files: i) Aboveground woody biomass raster, and ii)
 
 ## Methodology
 A Canopy Height Model (CHM) map for Africa was first generated by combining clusters of Global Ecosystem Dynamics Investigation (GEDI) footprints canopy height measurements with L-band SAR (ALOS-2 PALSAR-2) and Landsat Percent Tree Cover (PTC) (Hansen et al., 2012) by means of a Random Forests (RF) algorithm within a spatial k-fold calibration / validation framework. Clusters of GEDI footprints were used as reference data for the CHM estimation, by grouping 4 consecutive footprints along track. Then, an empirical model relating CHM to AGB, and developed using several airborne LiDAR AGB products, was used to estimate AGB. The PTC product was also used to constrain AGB estimations to pixels with PTC > 0 (discarding desserts, water bodies, etc.)
-  </Prose>
-</Block>
 
-<Block type='wide'>
-  <Prose>
 ## Uncertainty and Accuracy
 We first estimated the εCHM which is the standard deviation (SD) from our CHM retrieval based on RF and calculated as follows: εCHM = (ε2measurement + ε2temporal_difference + ε2sampling + ε2prediction)1/2, where εmeasurement is the SD arising from the measurement error of the GEDI footprint, εtemporal_difference is the SD from the use of GEDI footprints and EO imagery acquired at different time periods, and εsampling is the SD originating from the variability of CHM within the pixel. The εprediction corresponds to our model SD originated from the spatial k-fold framework. The εprediction also accounts for errors that arise if the sampling sites are not truly representative of the distribution of CHM in the region. The total SD from our AGB estimation at pixel level εAGB is composed of different sources of error, which are assumed to be random and independent. These are propagated using the following equation: εAGB = (ε2CHM + ε2LiDAR + ε2model)1/2, where εLiDAR is the SD from AGB LiDAR maps used as reference and includes field measurements, tree allometries and model errors. The εmodel is the error of AGB = f(CHM) empirical model.
 
 The AGB product is validated against a large dataset of in situ AGB estimations (i.e., forest inventory plots), and AGB estimated from airborne LiDAR data. Initial independent validation using ground measurements and airborne LiDAR shows RMSE = 48.5 Mg ha-1 and R2 = 0.83.
 </Prose>
+</Block>
 
+<Block>
   <Figure>
     <Map
       datasetId='nceo_africa_biomass'
@@ -87,4 +85,4 @@ This product is part of a larger dataset that covers the years 2007, 2008, 2009,
 
 Felis fames sed posuere lacinia neque vel pretium. Hendrerit vulputate blandit vel vivamus congue amet vehicula sit ullamcorper. Varius nostra fringilla vestibulum, amet vitae euismod ultrices. Ex leo consequat penatibus taciti suscipit nam lacus. Lorem suspendisse taciti placerat consectetur ultrices maecenas? Varius tincidunt vehicula ultricies faucibus cubilia volutpat ullamcorper. Curabitur imperdiet congue proin habitasse magnis curae interdum blandit. Commodo varius libero lobortis curabitur hac hac elementum fermentum.
   </Prose>
-</Block>
+</Block>