SP-3162: update for new templates (#28)

MelissaGraham · web-flow · commit bddcb369e829 · 2026-05-22T17:05:10.000-07:00
* update for new templates

* fix links

* try table auto width

* update to R/B v0.3

* ECB comments implemented
diff --git a/docs/index.rst b/docs/index.rst
@@ -24,8 +24,7 @@ Observational progress, live status reports, and known issues.
    :maxdepth: 2
    :titlesonly:
 
-   Observations <observations/index>
-   Known issues <known-issues/index>
+   Overview <overview/index>
 
 
 Data products
diff --git a/docs/overview/index.rst b/docs/overview/index.rst
@@ -0,0 +1,28 @@
+########
+Overview
+########
+
+Observational progress, live status reports, and known issues.
+
+Observations
+============
+
+Template coverage, alert stream status, and scheduler viewer.
+
+.. toctree::
+    :maxdepth: 2
+    :glob:
+
+    observations/index
+
+
+Known issues
+============
+
+Known issues and their mitigations.
+
+.. toctree::
+    :maxdepth: 2
+    :glob:
+
+    known-issues/index
diff --git a/docs/overview/known-issues/index.rst b/docs/overview/known-issues/index.rst
diff --git a/docs/overview/observations/index.rst b/docs/overview/observations/index.rst
@@ -1,6 +1,6 @@
-#######################
-Observations and status
-#######################
+############
+Observations
+############
 
 .. important::
 
@@ -26,12 +26,48 @@ Alert stream status
 A live dashboard showing alert processing over the last twelve hours is available at `ls.st/alert-stream-status <https://ls.st/alert-stream-status>`_.
 
 
+Template coverage
+=================
+
+On May 21 2026, a significant expansion to the template coverage available to Alert Production was deployed, extending it beyond the original `Deep Drilling Fields <https://survey-strategy.lsst.io/baseline/ddf.html>`_ templates.
+
+The best-covered subset of the templates generated as part of the forthcoming Data Preview 2 were selected.
+This increased the total sky area for alert production by more than a factor of ten (depending on the band).
+
+Current template coverage can be visualized with the `interactive alert production template coverage interface <https://s3df.slac.stanford.edu/people/ebellm/template_coverage/template_coverage.html>`_.
+More detailed programatic access to template coverage under development.
+
+.. list-table:: Current alert production coverage by filter.
+   :widths: auto
+   :header-rows: 1
+   :class: tight-table
+
+   * - Filter
+     - Sky area (sq. deg)
+   * - u
+     - 53
+   * - g
+     - 211
+   * - r
+     - 905
+   * - i
+     - 1444
+   * - z
+     - 1197
+   * - y
+     - 25
+
+
+
+Additional template images will be generated incrementally during the early science phase, and information about template regions will be added here.
+Details are available in :doc:`/processing/templates/index`.
+
+
 Survey strategy
 ===============
 
 Information about the strategy and cadence for the Legacy Survey of Space and Time can be found at `survey-strategy.lsst.io <https://survey-strategy.lsst.io/>`_.
 
-
 Nightly projection
 ------------------
 
@@ -45,50 +81,6 @@ Automated per-night reports are publicly available as `scheduler viewer reports
 Static files of LSSTCam visits metadata are available in the Commissioning :ref:`tutorials <tutorials-notebook>`.
 In the future, observational metadata will be available as queryable tables and documented under :doc:`/products/index`.
 
-Template coverage
------------------
-
-During early Alert Production, templates (and thus alerts) primarily exist in the `Deep Drilling Fields <https://survey-strategy.lsst.io/baseline/ddf.html>`_.
-
-As of February 2026, templates had been deployed in Alert Production for the fields listed in the table below.
-All of these are DDFs except the M49 (Virgo) field, which was part of the Rubin First Look images.
-
-.. list-table:: DDF templates
-   :widths: 20 20 20
-   :header-rows: 1
-   :class: tight-table
-
-   * - Name
-     - Coordinates
-     - Filters
-   * - COSMOS
-     - 150.11, 2.23
-     - ugrizy
-   * - ECDFS
-     - 52.98, -28.12 deg
-     - riz
-   * - ELAIS-S1
-     - 9.45, -44.02 deg
-     - griz
-   * - EDFS_a
-     - 58.9, -49.32 deg
-     - griz
-   * - EDFS_b
-     - 63.6, -47.6 deg
-     - griz
-   * - XMM-LSS
-     - 35.57, -4.82 deg
-     - iz
-   * - M49 (Virgo)
-     - 187.4, +8 deg
-     - ugri
-
-
-Additional template images will be generated incrementally during the early science phase, and information about template regions will be added here.
-Details are available in :doc:`/processing/templates/index`.
-
-.. important::
 
-   Tools for visualizing the current template holdings are under development.
 
 
diff --git a/docs/processing/filtering/figures/precision_recall_reliability_2026-02-13.png b/docs/processing/filtering/figures/precision_recall_reliability_2026-02-13.png
diff --git a/docs/processing/filtering/index.rst b/docs/processing/filtering/index.rst
@@ -38,33 +38,9 @@ The resulting ``reliability`` score ranges from 0-1, with larger scores correspo
 To minimize the number of false positives, ``DiaSources`` with reliability scores less than 0.5 are filtered out.
 Users may select alerts with higher reliability scores to maximize purity.
 
-
 .. important::
 
-   The performance information below is for the reliability model generated on 2026-02-13.  Continued training and refinement is underway.
-
-
-The current (2026-02-13) reliability model was retrained using detections collected between August and October of 2025.
-The detections are from both the AP and DRP pipeline products.
-The detections had a spatial crossmatch with either the Gaia variables catalog (`Rimoldini et al. 2023 <https://doi.org/10.1051/0004-6361/202245591>`_), the TNS catalog, Solar System objects detected by Rubin, and sources from the Rubin source catalog with an extendedness of 1.
-Additionally, some detections where the DP1 reliability model assigned a score between 0.5 and 0.9 were also used.
-The detections were labeled as Real or Bogus by volunteers through `Rubin Difference Detectives <https://www.zooniverse.org/projects/ebellm/rubin-difference-detectives>`_, a citizen science project on the Zooniverse platform.
-
-A series of analyses was performed on all the classifications provided by the volunteers. An approach similar to the one described in `Marshall et al. (2016) <https://ui.adsabs.harvard.edu/abs/2016MNRAS.455.1171M/abstract>`_ was implemented here; the main idea was to understand how the volunteers performed in the classification task by comparing the label provided by experts and the one provided by volunteers for the same detection.
-
-The initial classification used only detections with high-confidence labels assigned by volunteers.
-In total, 13,178 sources were used to fine-tune the DP1 model: 6,630 real and 6,548 bogus. 1,647 were used to validate, and another 1,647 to test the model.
-Given the small size of the training data set, in addition to the original images, two augmentations (vertical and horizontal flipping) were added to the training.
-
-For the test data set, at a threshold of 0.552, the purity (precision) is 95.0%, and completeness (recall) is 93.4%.
-Users can choose their own reliability threshold to trade off completeness vs. purity.
-
-
-.. figure:: figures/precision_recall_reliability_2026-02-13.png
-    :name: precision_recall_reliability_2026-02-13
-    :alt: Precision vs. Recall curves illustrating the performance of the machine learned reliability model.
-
-    Figure 1: Purity (precision) vs. completeness (recall) as a function of reliability threshold value for the 2026-02-13 model.
-    Note that performance is calculated on a test set and may not be fully representative of on-sky performance.
-
+   Continued training and refinement of the reliability model is underway, and the version of the model is reported in the alert packets.
 
+A technote describing the training and performance of the reliability models is available at (`DMTN-337 <https://dmtn-337.lsst.io/>`_).
+Interested users are encouraged to consult it to understand the impact of reliability selections on their objects of interest.
diff --git a/docs/processing/templates/index.rst b/docs/processing/templates/index.rst
@@ -15,7 +15,7 @@ In steady state operations, templates for alert production will be produced by t
 During Early Operations prior to DR1, templates will be built incrementally when sufficient suitable images are available in a given area of the sky.
 Once produced, templates will remain in place as long as possible to avoid repeated shifts of the difference light curves.
 
-For a list of the available templates, see the :doc:`/observations/index` page.
+For a list of the available templates, see the :doc:`/overview/observations/index` page.
 
 Coadds by patch
 ===============
diff --git a/docs/products/images/template_coadd.rst b/docs/products/images/template_coadd.rst
@@ -7,6 +7,7 @@ Template coadd (coming soon)
 .. important::
 
    Prompt images are not yet available.
+   Template coverage information is available on the :doc:`/overview/observations/index` page.
 
 
 The combination of processed images with the best seeing, for a patch of sky and for each of the six LSST filters, used as a template for difference imaging.
