Can the Temporal Profile chart on the Multidimensional toolbar be scripted?

[DougBrowning]

I built out a couple Multidimensional Multivariate Rasters and then use the Multidimensional toolbar to make Temporal Profiles super fast and it is pretty slick.

I now want to be able get a temporal profile in script for a given polygon to add to our reporting tool.  So basically simulate the above gif and get back a SVG of the chart that I add to our larger PDF.

I am not finding a way in script.  Is this possible?

I found the arcpy.raster one

temporal_profile(points=[], time_field=None, variables=[], bands=[0], time_extent=None, dimension=None, dimension_values=[], show_values=False, trend_type=None, trend_order=None, plot_properties={})

But it is not giving me anything and not sure how points fits in.  I got a chart to display just giving it time but its blank.

Thanks a lot

[DougBrowning]

In addition if I try to just use the charting from arcpy none of the code in the help makes much sense.

https://pro.arcgis.com/en/pro-app/latest/arcpy/charts/line.htm 

Like here chart is not defined anywhere?

import arcpy

aprx = arcpy.mp.ArcGISProject('current')
censusLayer = aprx.listMaps()[0].listLayers('Census Block Groups')[0]

# Set data source of chart object to a layer within current project
chart.dataSource = censusLayer

# Save the chart to file with dimensions width=500, height=500
chart.exportToSVG('populationByState.svg', 500, 500)

There is a arcpy.charts.line and arcpy.Chart that seem to intermingle some parts of the code samples?

Then I tried the info from this article  https://www.esri.com/arcgis-blog/products/arcgis-pro/analytics/using-arcgis-notebooks-and-arcpy-char...

myChart = arcpy.Chart("test")
myChart.type = 'line'
myChart.dataSource = mdRaster

myChart.exportToSVG(r"C:\temp\Charts\test.svg", 500, 500)

And I just get RuntimeError: Invalid dataset: on my crf.

Honestly I have been trying these charts for years but this is when I usually give up and just use Seaborn or some other package like that. 

Appreciate any links to something that can work for a Multidimensional raster and get me a temporal profile.  Maybe @ChristopherAllen ?

Its so easy in Pro I just need to automate it.  

thanks

[ChristopherAllen]

Hi @DougBrowning ,

Thanks for the feedback regarding charts documentation/code samples. The blog post you referenced was created before the new charts module (https://pro.arcgis.com/en/pro-app/latest/arcpy/charts/what-is-the-charts-module.htm) was introduced, and it will be updated to reflect the new API. We'll also look into making the code samples in the documentation clearer.

Thanks again

[HongXu]

The temporal profile itself does not have its own chart type, it basically gets a time/value array from the multidimensional raster and draw a line chart. To obtain that time/value array, you can run Sample tool, which takes point or polygon as input, and create an output table which can be used in creating a line chart.

[DougBrowning]

I ended up using Zonal stats as table since I get so much more info.   Then load that into a panda then seaborn to chart it.  It was literally two lines of code for a single var.  But splitting all 9 variables in my raster out then combing back into one chart is a bit of a pain in code.  

Access to the temporal profile tool would be much easier to chart all the vars and I like the way it handles the labels better and the chart points.  Please consider opening it up.  Thanks

[DougBrowning]

I tested the Sample tool and I like the output there better actually and it can feed directly into a panda all in memory.  Thanks for that idea!  (I thought the service did not work but it was parallelProcessingFactor="100%" that was causing it to fail - not sure why.

I built a CRF out for climate data yesterday and that Temporal Profile tool is a home run.  Just so fast to process so much data.  It may be worth its own toolbox.

Seems to work with a CRF directly or a service.  Pretty fast.

# first use the sample tool to pull out the means for a given polygon. setting extent seems to help speed a lot
with arcpy.EnvManager(parallelProcessingFactor="100%", extent=polygonAOI):
arcpy.sa.Sample(
in_rasters=rcmapCRF,
in_location_data=polygonAOI,
out_table=os.path.join("memory", "tempTable"),
resampling_type="NEAREST",
process_as_multidimensional="ALL_SLICES",
acquisition_definition=None,
statistics_type="MEAN",
percentile_value=None,
buffer_distance=None,
layout="ROW_WISE",
generate_feature_class="TABLE"
)

#Convert NumPy array to pandas DataFrame.
fc_np = arcpy.da.FeatureClassToNumPyArray(os.path.join("memory", "tempTable"), "*") 

df = pd.DataFrame(fc_np)

I was able to use arcpy charting to get one line chart but was not figuring out how to stack multiple vars in one chart.

It does not seem like I can give it more than one var for x and if I stack them like plot works I just get 1 line.  There must be a way since the Temporal Profile tool can do it but I cannot find any samples.  It is way easier to avg by year.  How do I chart multiple vars from a CRF?

chart = arcpy.charts.Line(x='StdTime', y='AvgTempC', aggregation="mean", timeIntervalSize=1, timeIntervalUnits='YEARS')

chart = arcpy.charts.Line(x='StdTime', y='AvgPrecipitationMM', aggregation="mean", timeIntervalSize=1, timeIntervalUnits='YEARS')

So I just went with mathplotlib.

f, ax = plt.subplots(1, 1)

ax.plot(df["StdTime"], df["AnnualHerbaceous"], color="blue", label="Annual Herbaceous", linestyle="-", marker='.')

Thanks a lot

[ChristopherAllen]

Hi @DougBrowning ,

If I understand correctly, you want to create a single chart displaying one line for each of your variables. To do this you would pass the `y` argument a list of all your field names. Using your example, this would be:

chart = arcpy.charts.Line(x='StdTime', y=['AvgTempC', 'AvgPrecipitationMM'], aggregation="mean", timeIntervalSize=1, timeIntervalUnits='YEARS')

Hope this helps!

Chris

[DougBrowning]

Yep that works.  I figured that would be the way but the help says string.  I should have tested just busy.  Sorry about that.

[ChristopherAllen]

Awesome, glad to hear that worked. Thank you for following up.