docs.html

<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>About — Barred Owl Prioritization (GLEN proof-of-concept)</title>
<style>
:root {
    --ink: #241c14;
    --ink-soft: #5b4f43;
    --paper: #faf7f2;
    --rule: #e7dfd3;
    --accent: #8a5a2b;
    --accent-dark: #5e3c18;
    --accent-soft: #f0e7da;
    --link: #8a5a2b;
    --max: 880px;
}
* { box-sizing: border-box; }
html, body {
    margin: 0; padding: 0;
    background: var(--paper); color: var(--ink);
    font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", system-ui, sans-serif;
    font-size: 16px; line-height: 1.6;
}
a { color: var(--link); text-decoration: none; }
a:hover { text-decoration: underline; }
header.hero {
    background: linear-gradient(135deg, #3a2712 0%, var(--accent) 100%);
    color: #fff; padding: 3.5rem 1.5rem 2.5rem; border-bottom: 1px solid var(--rule);
}
.hero-inner { max-width: var(--max); margin: 0 auto; }
.hero h1 { font-size: 2.1rem; margin: 0 0 0.4rem 0; font-weight: 600; letter-spacing: -0.02em; }
.hero .tag { font-size: 1.05rem; opacity: 0.92; margin: 0 0 1.4rem 0; max-width: 62ch; }
.hero .cta {
    display: inline-block; background: rgba(255,255,255,0.15); color: #fff;
    padding: 0.55rem 1.1rem; border-radius: 6px; border: 1px solid rgba(255,255,255,0.35);
    font-weight: 500; transition: background 0.15s;
}
.hero .cta:hover { background: rgba(255,255,255,0.28); text-decoration: none; }
.hero .badge {
    display: inline-block; margin-left: 0.6rem; padding: 0.18rem 0.55rem; border-radius: 999px;
    background: rgba(255,255,255,0.18); font-size: 0.8rem; font-weight: 500;
    letter-spacing: 0.02em; vertical-align: middle;
}
main { max-width: var(--max); margin: 0 auto; padding: 2.5rem 1.5rem 4rem; }
section { margin-bottom: 2.6rem; }
h2 {
    font-size: 1.45rem; color: var(--accent-dark); border-bottom: 2px solid var(--accent-soft);
    padding-bottom: 0.4rem; margin: 0 0 1rem 0; font-weight: 600;
}
p { margin: 0.5rem 0 0.9rem; }
p.lede { font-size: 1.05rem; color: var(--ink-soft); }
.callout {
    background: #fff; border: 1px solid var(--rule); border-left: 4px solid var(--accent);
    border-radius: 8px; padding: 1rem 1.25rem; margin: 1.1rem 0;
}
ul { margin: 0.4rem 0 0.9rem; padding-left: 1.3rem; }
li { margin: 0.3rem 0; }
footer { max-width: var(--max); margin: 0 auto; padding: 1.5rem; color: var(--ink-soft); font-size: 0.9rem; border-top: 1px solid var(--rule); }
</style>
</head>
<body>
<header class="hero">
    <div class="hero-inner">
        <h1>Barred Owl Prioritization Tool <span class="badge">proof of concept</span></h1>
        <p class="tag">An independent demonstration of exploring the Barred Owl Tool's data
            through GLEN — an AI-assisted, cloud-native map interface. This is <strong>not</strong>
            the official tool.</p>
        <a class="cta" href="/">← Open the map</a>
    </div>
</header>
<main>

<section>
    <h2>About this project</h2>
    <p class="lede">This is a <strong>proof-of-concept</strong> built by the
        <a href="https://dse.berkeley.edu">Eric and Wendy Schmidt Center for Data Science &amp;
        Environment (DSE)</a> at UC Berkeley to explore the data behind the
        <strong>Barred Owl Tool</strong> inside the GLEN architecture. It is not affiliated with,
        endorsed by, or maintained by the tool's authors.</p>
    <p>The Barred Owl Tool is a decision-support tool that prioritizes where to concentrate
        invasive barred owl (<em>Strix varia</em>) population control to protect the northern
        spotted owl (<em>Strix occidentalis caurina</em>), complementing the USFWS Barred Owl
        Management Strategy. All of the science, modeling (via the <code>prioritizr</code>
        optimization package), and the original interactive tool are the work of its authors —
        full credit and authority rest with them.</p>

    <div class="callout">
        <p style="margin-top:0"><strong>Use the official tool.</strong> For authoritative
        results, methods, and guidance, please visit the original application and its
        documentation:</p>
        <ul>
            <li><a href="https://barredowltool.russell.wisc.edu">barredowltool.russell.wisc.edu</a>
                — the official Barred Owl Tool (Background, Methods, How-to guide)</li>
            <li><a href="https://www.fws.gov/project/barred-owl-management">USFWS Barred Owl
                Management</a> — the management strategy this tool supports</li>
        </ul>
        <p style="margin-bottom:0">Coproduced by the University of Wisconsin–Madison
        <a href="https://peery.russell.wisc.edu">Peery Wildlife Ecology + Conservation Lab</a>
        and the US Fish &amp; Wildlife Service, with support from
        <a href="https://appliedsciences.nasa.gov/what-we-do/ecological-conservation">NASA's
        Ecological Conservation</a> program.</p>
    </div>

    <p>This demo mirrors the tool's outputs into cloud-native formats and pairs them with an
        AI assistant. Because it re-processes and re-hosts the data, values or styling here may
        differ from the official tool — treat this as an exploratory interface, not a source of
        record.</p>
</section>

<section>
    <h2>Understanding the tool</h2>

    <h3>The conservation problem</h3>
    <p>The <strong>northern spotted owl (NSO)</strong>, <em>Strix occidentalis caurina</em>, is
        a threatened old-forest raptor of the Pacific Northwest. The <strong>barred owl</strong>,
        <em>Strix varia</em> — a larger, more aggressive relative from eastern North America — has
        expanded into the spotted owl's range and now outcompetes it for territory and food. This
        invasion is one of the leading drivers of the spotted owl's continued decline. The USFWS
        Barred Owl Management Strategy calls for <strong>targeted removal of barred owls</strong>
        where it will most benefit spotted owls. But the range is vast and resources are limited,
        so the core question is <em>where</em> to act — a spatial prioritization problem.</p>

    <h3>What a "priority run" is</h3>
    <p>The tool answers that question with <strong>prioritizr</strong>, a systematic
        conservation-planning package that casts the map as an optimization problem. Given a set
        of input layers — where spotted owls are, where barred owls are, habitat quality, fire
        risk, accessibility, and so on — and a fixed <strong>budget</strong> (roughly the area
        that can realistically be managed), prioritizr finds the allocation of ~5&nbsp;km² hex
        cells that best meets the objective: maximizing the weighted coverage of the things we
        care about. Every hex is then given a <strong>priority score from 0 to 1</strong> —
        <code>0</code> for cells the solution leaves out, up to <code>1</code> for the
        highest-priority cells. One such run is a <em>priority run</em>. The score is
        <strong>relative</strong>: it compares cells <em>within a chosen planning area</em>,
        rather than measuring an absolute amount — so scores from different planning areas aren't
        directly comparable.</p>

    <h3>What the scenarios are</h3>
    <p>There is no single "correct" prioritization — it depends on management goals and
        assumptions. The tool therefore produces <strong>2,754 scenarios</strong>, combining four
        choices:</p>
    <ul>
        <li><strong>Where you're planning</strong> (management unit): the whole spotted owl range,
            one of 10 physiographic provinces, or one of 40 general management areas. The
            optimization is re-solved and rescored <em>within</em> each unit, so a cell's
            range-wide score differs from its score inside a province or management area.</li>
        <li><strong>What you're protecting</strong> (threat mode): <em>current</em> spotted owl
            populations, <em>projected future</em> populations, or nest/roost <em>habitat</em>.</li>
        <li><strong>Which factor you emphasize</strong> (tuning parameter): the weight on spotted
            owls, or the penalty/incentive for fire risk, fire refugia, federal reserve lands,
            distance to roads, or distance to trails.</li>
        <li><strong>How strongly</strong> (weight level): low, medium, or high emphasis on that
            factor.</li>
    </ul>
    <p>Comparing scenarios is the point: cells that score high across <em>many</em> scenarios are
        robust priorities, while cells that score high only under one set of assumptions are more
        sensitive to how the problem is framed.</p>
</section>

<section>
    <h2>How it works</h2>
    <p>This app is powered by <a href="https://schmidtdse.github.io/glen/">GLEN</a>,
        DSE's open-source framework for AI-assisted geospatial exploration. Open datasets stream
        straight to your browser from cloud-native formats — no tile server in between — and an
        AI assistant answers questions by writing live queries against the same data. Technical
        details are in the <a href="https://schmidtdse.github.io/glen/">GLEN
        documentation</a>.</p>
    <p>🌱 <strong>Low-carbon, open models.</strong> GLEN uses more environmentally-friendly AI
        models — faster than most humans, if not the very largest commercial models. See the
        <a href="https://carbon-dashboard.nrp-nautilus.io">carbon dashboard 🌱</a> for the
        CO₂/energy use of this deployment.</p>
</section>

<section>
    <h2>What you can explore here</h2>

    <h3>On the map — the model inputs</h3>
    <p>Each map layer is one of the <strong>ingredients</strong> the prioritization draws on,
        shown on the native ~5&nbsp;km² hex grid. These are the same in every scenario — only the
        priority scores change. From coarse to fine:</p>
    <ul>
        <li><strong>NSO carrying capacity</strong> (0–20) — how many spotted owl territories a
            cell could support.</li>
        <li><strong>NSO pair occupancy</strong> and <strong>nest/roost habitat</strong> (0–1) —
            where spotted owls are, and where suitable breeding habitat is.</li>
        <li><strong>Barred owl occupancy</strong> (0–1) — where the invading barred owl is
            established.</li>
        <li><strong>Large-fire suitability</strong> vs. <strong>fire refugia</strong> (0–100) —
            wildfire risk against climate-buffered refuges.</li>
        <li><strong>Federal reserve allocation</strong> — federal conservation land, which can
            raise or lower priority depending on the scenario.</li>
    </ul>

    <h3>Through the assistant — the priority scores</h3>
    <p>The 2,754 priority runs are far too many to list as map toggles, so they live in a
        queryable table that the assistant reads on demand. Ask it to <strong>map or rank any
        scenario</strong>, or to <strong>compare across scenarios</strong>. It can also drive the
        map and answer quantitative questions about the inputs. Try:</p>
    <ul>
        <li>"Where is northern spotted owl carrying capacity highest?"</li>
        <li>"Show the range-wide current-threat scenario at high NSO weight."</li>
        <li>"Compare priority scores between the current-threat and future-threat scenarios."</li>
        <li>"Which physiographic province has the most high-priority area?"</li>
    </ul>
    <p style="color:var(--ink-soft)"><strong>Reading the scores:</strong> higher = higher
        management priority, and scores are relative <em>within</em> each management unit — so
        don't compare a province's raw scores against the range-wide run as if they were on one
        scale.</p>
</section>

<section>
    <h2>Data &amp; sources</h2>
    <p>Data mirror:
        <a href="https://radiantearth.github.io/stac-browser/#/external/s3-west.nrp-nautilus.io/public-barred-owl/stac-collection.json">STAC
        collection</a> (<code>public-barred-owl</code>) — processed by the
        <a href="https://github.com/boettiger-lab">Boettiger Lab</a> from the
        <a href="https://barredowltool.russell.wisc.edu">Barred Owl Tool</a>.
        Underlying inputs include NSO pair occupancy (Lesmeister et al.), spotted-owl nest/roost
        habitat (Davis et al. 2022), barred-owl landscape use (Jenkins et al. 2025,
        <em>J. Biogeography</em>), USGS GAP species, large-fire suitability (Davis et al. 2017),
        OSU Eco-Vis fire refugia, and NWFP federal reserve allocations. See the official tool's
        Methods page for full provenance.</p>
</section>

</main>
<footer>
    A GLEN proof-of-concept by DSE / Boettiger Lab · Data &amp; original tool © their respective
    authors · <a href="/">Open the map →</a>
</footer>
</body>
</html>