Image ID Intents Chart appropriateness Visual variable problem Level-of-detail Color map problem Clutter Distortion Comparison complexity Communication gap Problem Descriptions
ScatterplotD0 Compare the distribution of GPP for prognostic and diagnostic models. Look at the variability within the prognostic and diagnostic models mismatch choice granularity overlap clutter:due to over plotting of symbols; encoding problem due to combination of color and orientation that makes it difficult to visually search and find a particular model; the level-of-detail creates a problem as one as to mentally compute the spread. There can be a better way to show spread, hence mismatch.
ScatterplotD1 Compare the spread of NEP, GPP and Rh for different eco-regions and analyze the difference in variability for the prognostic and diagnostic models. For each variable, identify which models belong to which eco-region. mismatch granularity color map:qualitative overlap grids clutter:due to over plotting of symbols; encoding problem due to combination of color and orientation that makes it difficult to visually search and find a particular model; the level-of-detail creates a problem as one as to mentally compute the spread. Mismatch, as the intent of finding which model belongs to which ecoregion is not conveyed.
ScatterplotC1 Compare the the skill of CMIP2 (ca 2000) and CMIP3 (ca 2005) models for predicting precipitation etc. There is a slight improvement from CMIP2 to CMIP3, but the spread among models is much greater than the improvement in the average model skill. ambiguity legend, grids encoding: use of marks is ambiguous, why use red squares and blue rhombus?; Location of mean marks is not clear; legend: difficult to interpret what the shapes and filled or unfilled areas mean.
ScatterplotKeenan1 Compare the errors (quantified in two ways) of different models in reproducing GPP, NEE, and RE for two plant functional types. Shows both that there is a lot of spread across models, and also that there is maybe slightly smaller errors in modeling the DBF plant functional type than the EVG plant functional type. mismatch scale inconsistency lack of explicit encoding legend,annotation labelling: placement of labels (NEE, GPP, RE); scale: non-aligned scales (see axes ticks) make comparison difficult and may lead to misinterpretation; intent-chart mismatch: chart type issue. a bivariate plot (scatter plot) is not necessary for this intent if the real intent is to see the distribution with respect to each type of error;
ScatterplotGlecker1 Shows the relationship between model errors in predicting precipitation and LW radiation, and between model errors in predicting sea level pressure and geopotential height, across a variety of models. The figure shows that the errors made by the models in predicting precipitation is very correlated to the errors that they are making for sea level pressure. The other two types of errors are not related to each other, on the other hand. labelling: what is being measured here is error but that is not mentioned in the labels
ScatterplotGlecker2 Plot shows model metrics (precipitation and surface pressure) evaluated across a variety of models, and the models are ranked according to this metric. For each model, they then also show the metric evaluated using alternative setups, so see whether the variation in model skill is large or small relative to the different between models. The figure shows that for the vast majority of models, the variation across different setups is at least as wide as the variation between models, meaning that model ranking is rather ambiguous except for the very "best" and very "worst" models. configuration choice overlap legend clutter: due to over plotting; encoding: shape vs color; legend: no legend for symbols in the chart; chart configuration: improve readibility by swapping the X and Y axes.
ScatterplotGlecker3 The fact that there is a lot of variability across the individual variables that contribute to the index tells me that the model ranking is probably very sensitive to how these multiple variables are integrated / averaged / compiled into the single final index. Model mean and model median outcompete any of the individual models. Also, the overall model rank is very different from the model rank if we were to look at just one or a few variables. choice overlap clutter: there are too many overplotting symbols; encoding: color and orientation are used together to denote different categories (variabels) but they are not separable, therefore it's hard to select something (what variables are outliers) (e.g., lots of cyan dots on the top but they represent different variables)
ScatterplotGlecker4 Show the variation of climate performance index Vs variability index for different models. Confusing whether "0" is good, or whether negative numbers are even better. Shows that some models are good at both climate and variability. For the other models, they tend to be good at one and poor at the other, i.e. no models are terrible at both. choice annotation encoding: same as above; annotation: the chart could be improved by manually annotating the observed trends (the message looks like some models are good/bad for both criteria) (Note: from the chart it's not clear whether nevative numbers are good or bad)
ScatterplotJimenez1 Figure looks at the variability across models in the predictions of key variables for snow covered regions in two different seasons. In the summer, there seems to be a positive correlation between the two variables. scale inconsistency annotaion scale: inconsistent scale makes it very hard to comapre between summer and winter charts; encoding: either use labels to annotate data points or symbols with legends describing them. Using both causes clutter.
ScatterplotPan1 Figure looks at NPP when atmospheric CO2 is doubled relative to NPP for current atmospheric CO2 (y-axis). The figure then sees whether the predicted change in NPP is related to the change in precipitation and temperature. The top plot shows that the change in NPP might be higher when the precipitation is low, but no impact thereafter. The bottom plot shows that the change in NPP is relatively insensitive to the mean annual temperature. It is a little hard to see what happens across models, but it looks like CENTURY consistently predicts a smaller change in NPP relative to the other models. choice superposition overload encoding: redundant use of shape and color; Decoding problem: here the intent is to show the variation within each model and not to compare models encoded in a single scatter plot. Use of small multiples will facilitate much more effective comparison
ScatterplotPan2 This is showing the same information as the last plot, except that all three variables are presented together, individually by model. mismatch potentially misleading due to the 3D plot
ScatterplotIPCC522 Compare the spread of carbon flux for differen regions. choice overlap clutter: due to overlap; legend: no description why only blue lines continue after 2040; annotation: also no descritpion of the gray bar; encoding: dashed lines and different shapes interfere making it very hard to discriminate one from the other.
ScatterplotIPCC554 Plots compare simulations of arctic and antarctic ozone to observations (in black). All models show time trends that are generally consistent with observations, but there is a lot of spread across models. Not clear how time of grey box was determined. choice, ambiguity overlap superposition overload legend,annotation legend: what do the symbols mean; encoding: unclear motivation for using color and shape together; annotation: what do the bars mean and what do the different categories mean
ScatterplotIPCC631 Plot compares the magnitude of different climate feedbacks across a variety of models. Hard to decipher what all the different symbols mean. Legend is not very informative and caption is complicated. Not clear how the symbols are binned into the three columns, i.e. what the three columns represent. ambiguity legend,annotation granularity: do we need to show all the data points to convey the message?; encoding: use of color as a quantitative channel, relative ranking according to similarity; comparison mechanism: is juxtaposing small multiples, that is 17 different maps the best way to compare similarity and dissimilarity; annotation: why data is missing
MapD1 For summer months during the period 2000-2005, analyze the degree of similarities and dissimilarities of the different models. choice granularity lack of explicit encoding annotation encoding: use visual variables that show relationship between mean and standard deviation more effectively, may be on a single map (refer bivariate maps -page 91, MacEachren) using glyphs?
MapD2 Based on provisional results. This looks at one variable and an ensemble average over 30 years. It seems that there is a relationship between mean GPP and standard deviation (areas with low mean GPP are associated with areas having low standard deviation and areas with high mean GPP are associated with higher standard deviations for the ensemble. Large variability among models in the GPP ensemble variability captured by mean and stddev. Not uncertainty but model spread. Larger the GPP, the larger the spread across models. granularity color map:quantitative lack of explicit encoding scale: huge difference in scale across the models
MapHT1 Compare the spatial distribution of an output variable for multiple models and deduce the degree of similarity and dissimilarity among them. choice granularity color map:quantitative scale inconsistency lack of explicit encoding scale: huge difference in scale across the models
MapHT2 Compare the spatial distribution of an output variable for multiple models and deduce the degree of similarity and dissimilarity among them. choice granularity color map:quantitative scale inconsistency lack of explicit encoding distortion due to map projection
MapU1 Compare the spatial distribution of an output variable with its historical behavior for one model ambiguity granularity color map:quantitative projection error encoding: not all colors in the color map appear in the map
MapU2 Compare the spatial distribution of an output variable with its historical behavior for one model ambiguity color map:quantitative projection error distortion due to map projection
MapU3 Compare the variability of two output variables for the same model for a particular year color map:quantitative projection error distortion due to map projection
MapU4 Compare the variability of two output variables for the same model for a particular year color map:quantitative projection error distortion due to map projection
MapU5 Compare the spatial distribution of an output variable with its historical behavior for one model ambiguity color map:quantitative projection error
MapU6 Compare the variability of two output variables for the same model for a particular year color map:quantitative
MapU7 Compare the variability of an output variable for two different months in a year. color map:quantitative
MapU8 Compare the variability of an output variable for two different months in a year. color map:quantitative distortion due to map projection
MapU9 Observe the temporal variability of an output variable over a period of time. projection error distortion due to map projection
MapU10 Observe the temporal variability of an output variable over a period of time. projection error distortion due to map projection
MapU11 Observe the temporal variability of an output variable over a period of time. projection error distortion due to map projection
MapU12 Observe the temporal variability of an output variable over a period of time. projection error clutter due to strong borders
MapMorton1 Compare the spatial distribution of three variables over a period of time. granularity color map:quantitative encoding: potential misinterpretation due to encoding p-value and r with the same color;
MapMorton2 Look at the spatial relationship between the correlation of burned area and climate variables over time. Looking at temporal correlation spatially. Also intent to show the pairwise correlation of a particular cell across the five maps comparing temporal correlation wrt burnt area with climate drivers and visualize them spatially. ambiguity encoding: potential misinterpretation due to edual encoding
MapMorton3 Compare the variation of different variables for different time periods. ambiguity
MapIPCC2 Compare seasonal values of ACE index for different regions. granularity scale inconsistency legend scaling: inconsistent scales; mismatch: use a line chart?; granularity: temporal granularity;legend: legend:what does the red line mean?
MapIPCC3 Compare temporal change of two variables and show the regions corresponding to the changes. granularity color map:quantitative scale inconsistency scale: skewed scale applied to color leading to potential misinterpretation; granularity: also aplies to the number of different colors that are chosen
MapIPCC4 Compare temporal variation of temperature for different reconstructions and showing spatial locations of temerature sensitive proxy records choice overlap clutter due to interference with the background and size of the symbols
MapVEMAT1 Compare occurrence and geographic distribution of specific vegetation type color map:qualitative color map: distances between hues are not sufficiently large for distinguishing them. Use the Tableau 20 color palette?
MapVEMAT2 Comparing annual temperature change difference for different variables color map:qualitative ok but can be improved by avoiding black
MapVEMAT3 Comparing annual temperature change difference for different variables color map:qualitative
MapSchimel1 Comparing net carbon storage for multiple models for different regions. color mixing lack of explicit encoding encoding: mean can also be represented; color interference; clutter due to strong black lines
MapYates1 Compare vegetation classification of different regions in the US ambiguity color map:qualitative encoding due to ambiguous color map; color map due to inseparability
MapIPCC181 Compare land cover classifications. color map:qualitative grids
MapIPCC257 Comparing annual, regionwise mean trends for precipitation. color mixing
MapIPCC263 Compare drought index for different regions. encoding: overlapping dots and dashes cause clutter. Models can be separated into their own charts.
LineChartD1 Show how 4 different combinations of climate drivers affect inter-annual variability and trend of global total GPP. 2. Compare between two different models: LPJ and ORCHIDEE choice overlap superposition overload clutter due to color mixing; scaling: insufficent resolution or aspect ratio
LineChartD2 1. Show latitudinal average of various variables (GPP, TotSoilCarb, and TotLivBiom) of multi-model mean 2. Compare latitudinal average variability across variables 3. Compare latitudinal average variability across different simulation conditions (initial conditions v.s. best estimate) 4. Show range and variability of simulations across models. color mixing encoding due to chart type; clutter due to color mixing
LineChartD3 Show and compare additive impacts of different drivers on global total GPP along time mismatch color mixing encoding due to chart type; clutter due to color mixing
LineChartD4 Compare the annual temporal variability of the different models with respect to the ensemble mean and with respect to each other. mismatch color mixing superposition overload
LineChartC2 Analyze variance of model errors. mix of dots and dashes, use lines with different opacity.
LineChartC3 Analyze spectral power of model error choice legend legend for multiple black lines; annotation for dotted line; encoding due to choice of colors
LineChartDietze1 Compare different model spectra and global spectra to null spectra. clutter due to lne thickness; annotation: not clear what b and c are, have to find from the caption, superposition overload
LineChartDietze2 Compare different model errors. choice color map:qualitative legend,annotation
LineChartDietze3 Compare temporal changes of multiple output variables for multiple models annotation
LineChartHT1 Compare annual temporal variability of burnt area for different scenarios.
LineChartHT2 Compare temporal variability of wildfire emissions for different continents. annotation:
LineChartHT3 Global aggregate and not different continents. Not sure what the bottom panel adds to the top panel. Bottom one is rescaled and the top one could also be rescaled.
LineChartHT4 Compare simulated vs observation data for variability of burnt area. jaggedness annotation,grids encoding: add a regression line and code NEP lines separately
LineChartHT5 Compare temporal variation of temperature with that of multiple observations of NEP encoding: due to the redundant red dots and prominence of black line over blue lines; clutter for the same reason; scale due to non-zero baseline
LineChartIPCC1 Compare temporal variability of GPP, temperature and precipitation jaggedness overlap superposition overload,lack of explicit encoding clutter: red line invisible; annotation: annotate the temperatures differently as it is misleading
LineChartIPCC2 Compare observed temporal variability in temperature, sea level and snow cover, and the uncertainty choice overlap scale inconsistency
LineChartIPCC3 Compare observed and simulated changes in surface temperatures. overlap annotation
LineChartIPCC4 Compare multimodel global averages of temperatures for different scenarios, along with their standard deviation. clutter due to thickness of line
LineChartIPCC5 Understand variance of global mean temperatures. legend: dashed red one not in the legend, colors in the legend dont appear in the chart; clutter: lines are too thick; mismatch due to using lines at the bottom ,
LineChartIPCC6 Compare different PDF estimates for climate sensitivity. color map:qualitative encoding: small multiple vs large single
LineChartIPCC7 Temporal variability of temperature reconstructions. Unclear message, difficult to interpret these spaghetti graphs. mismatch color map:qualitative encoding: due to use of shapes; clutter due to thickness of lines; scale inconsistency
LineChartIPCC8 Compare interannual variability of MTBS and GFED for different regions jaggedness overlap superposition overload scale inconsistency
LineChartMorton1 Compare variability of Fire season PE for different regions choice scale inconsistency
LineChartMorton2 Compare variation of soil water and temperature with respect to transpiration choice scale inconsistency
LineChartPan1 Analyze the annual scalar for the effect of nitrogen availability on GPP as a function of mean annual temperature and study the ratio of the resulting regression lines clutter due to color mixing, the mean line is almost invisible; color map as different seasonal cylces have the same color; encoding as use of shapes is not necessary if different colors are used; encoding for stacking the line charts horizontally and vertically
LineChartPan2 Compare multiple time series for different seasonal cycles Compare multiple annual cycles by displaying the ensemble mean, standard deviation around the ensemble mean same as above except for orientation
LineChartJimenez1 Compare multiple time series for different seasonal cycles configuration ambiguity color mixing clutter due to line thickness; colormap due to poor selection of hues, white dots on white background is avoidable; encoding: no dots for the bottom plot
LineChartJimenez2 Compare variability of seasonal cycles of LAI from different models and observations. Messages: 1. There are some outliers 2. Seasonal cycles of models somewhat agree with observation, but magnitude varies a lot. ambiguity color mixing clutter due to line thickness and the red dotted line at the bottom panel almost invisible
LineChartRichardson1 Compare variability of multiple output variables for different simulations. Messages: 1. CO2 increase (SG3/BG1) has a big positive impact on GPP and NPP, thus promotes the carbon uptake (NEE decrese) 2. Land use (SG2) slightly reduced NPP. color hues can be better chosen
LineChartTR60 Comparing temporal variations of temperature reconstructions and visualizing the overlap of their uncertainty ranges color map:qualitative annotation: not clear what the blue and brown lines mean; encoding: why do the different colors have varying widths?; encoding: different forcings can be encoded with differently colored lines
LineChartTR6 Intents: 1. Compare temperature anomaly time series across models/simulations and with reconstructed temperature 2. Relate modeled temperature variation with different forcings choice annotation encoding: stack the line charts for the same variable vertically
LineChartKeenan1 Compare the monthly variability of different output variables. Messages: 1. Models tend to underestimate GPP/NEE/Re in May 2. Models uncertainties maximize during summer configuration clutter due to overlap, mixing of green line with green shaded area; color map: green and grey are not good choices; encoding: do the grey vertical lines serve any purpose?
LineChartKeenan2 Compare the modelled and observed monthly variability for different output variables. Intents: 1. Compare time series of observed and modeled variables 2. Show uncertainties of model simulations 3. Link monthly variability of observations with mean choice overlap, color mixing annotation for grey bars and legend for the different colors, what does the red line mean?
LineChartTR1 Comapre the historical, temporal variability of multiple variables and indicate warm periods. annotation
LineChartTR2 Compare linear trends of precipitable water. Intents: 1. Show spatial pattern of global linear trends of column water vapour 2. Show monthly time series of anomalies and linear trends for global mean column water vapour and another variable (...) 3. Compare monthly time series of anomalies and linear trends for global mean column water vapour and another variable (...) jaggedness encoding due to color mixing; legend for colors
LineChartTR3 Compare observed and reconstructed sea level changing patterns.Intents: 1. Display spatial pattern of short-term trends in mean sea level 2. Link point location on a map jaggedness color mixing
LineChartTR4 Intents: 1. Compare model simulations (with different forcings) with observations 2. Show uncertainties of model simulations 3. Link events with model simulations and observations
LineChartIPCC27 Analyze annual changes in global mean carbon dioxide concentration and visualize the uncertainties annotation: of grey bars and dotted lines;color map: yellow stands out
LineChartIPCC525 Intents: 1. Show global precip long-term trends spatial pattern 2. Show and compare annual mean precip trends for different regions 3. Compare two annual mean precip trends between two different data sets ambiguity color mixing annotation clutter due to overlap of black and red, encoding because it is hard to detect any patterns due to absence of logical grouping of line charts
LineChartIPCC257 Compare year-to-year anomalies in carbon dioxide fluxes by visualizing different ensemble means. Messages: 1. Carbon fluxes on land have much bigger inter-annual variability than in ocean. 2. El Nino has positive impacts on carbon fluxes 3. Cooling period of volcano has negative impact on carbon fluxes, especially on northern and tropical regions. 4. Tropical region has biggest contribution to global carbon fluxes configuration overlap superposition overload