Fuzzy Flow

Embed Size (px)

Citation preview

  • 8/10/2019 Fuzzy Flow

    1/59

    Flow Computations on

    Imprecise Terrains

    Anne Driemel,Herman Haverkort,Maarten Lofflerand Rodrigo Silveira p

    q

    EuroCG 2011, Morschach

  • 8/10/2019 Fuzzy Flow

    2/59

    Why study water flow on terrains?

    Analysis of flash floods

    Streamflow forecasting Erosion prediction for river beds

    ..

    Interesting geometric problems

  • 8/10/2019 Fuzzy Flow

    3/59

    p

    Assumption: Water flows downwards inthe direction of steepest descent.

  • 8/10/2019 Fuzzy Flow

    4/59

    q

    p

    The watershed W(q):the area that drains to q.

    Assumption: Water flows downwards inthe direction of steepest descent.

  • 8/10/2019 Fuzzy Flow

    5/59

    q

    p

    q

    p flows to q

    The watershed W(q):the area that drains to q.

    Assumption: Water flows downwards inthe direction of steepest descent.

  • 8/10/2019 Fuzzy Flow

    6/59

  • 8/10/2019 Fuzzy Flow

    7/59

  • 8/10/2019 Fuzzy Flow

    8/59

    q

    p

  • 8/10/2019 Fuzzy Flow

    9/59

    q

    p

  • 8/10/2019 Fuzzy Flow

    10/59

    qq

    p

  • 8/10/2019 Fuzzy Flow

    11/59

    q

    p

  • 8/10/2019 Fuzzy Flow

    12/59

    p

    q

  • 8/10/2019 Fuzzy Flow

    13/59

    q

    p

  • 8/10/2019 Fuzzy Flow

    14/59

    p

    q

  • 8/10/2019 Fuzzy Flow

    15/59

    p?

    q

    Can we computefuzzy watersheds?

  • 8/10/2019 Fuzzy Flow

    16/59

    p?

    q

    Can we computefuzzy watersheds?In this model:NP-hard

  • 8/10/2019 Fuzzy Flow

    17/59

    In a grid elevation model..

  • 8/10/2019 Fuzzy Flow

    18/59

    In a grid elevation model..

  • 8/10/2019 Fuzzy Flow

    19/59

    In a grid elevation model..

  • 8/10/2019 Fuzzy Flow

    20/59

    Each grid cell drainsto one of its eightneighbors.

    In a grid elevation model..

  • 8/10/2019 Fuzzy Flow

    21/59

    Each grid cell drainsto one of its eightneighbors.

    In a grid elevation model..

    q

    p

  • 8/10/2019 Fuzzy Flow

    22/59

    Flow network

    q

    p

  • 8/10/2019 Fuzzy Flow

    23/59

    Flow network

    Watersheds arediscrete point sets

    q

    p

  • 8/10/2019 Fuzzy Flow

    24/59

    Definitions

    An imprecise terrain is a graph in which each node

    v has the form v= (x,y, [z, z+]).

  • 8/10/2019 Fuzzy Flow

    25/59

    Definitions

    An imprecise terrain is a graph in which each node

    v has the form v= (x,y, [z, z+]).

  • 8/10/2019 Fuzzy Flow

    26/59

    Definitions

    An imprecise terrain is a graph in which each node

    v has the form v= (x,y, [z, z+]).

  • 8/10/2019 Fuzzy Flow

    27/59

    Definitions

    An imprecise terrain is a graph in which each node

    v has the form v= (x,y, [z, z+]).

  • 8/10/2019 Fuzzy Flow

    28/59

    Definitions

    A realization R is the graph with v= (x , y , z),

    such that z[z, z+].

    R

  • 8/10/2019 Fuzzy Flow

    29/59

    Definitions

    In a fixed realization water flows from a node to

    its steepest descent neighbor.

    p

    R

  • 8/10/2019 Fuzzy Flow

    30/59

    Definitions

    q

    The potential watershed :

    W(q) :=R

    {p:p flows to q in R}

  • 8/10/2019 Fuzzy Flow

    31/59

    Definitions

    q

    The core watershed :

    W(q) :=R

    {p:p flows to q in R}

  • 8/10/2019 Fuzzy Flow

    32/59

    Results

    The potential watershed :

    The core watershed :

    W(Q) :=R

    {p:p flows to q in R, qQ}

    W(Q) :=

    R

    {p:p flows to q in R, qQ}

    q

  • 8/10/2019 Fuzzy Flow

    33/59

    Results

    The potential watershed :

    The core watershed :

    W(Q) :=R

    {p:p flows to q in R, qQ}

    W(Q) :=

    R

    {p:p flows to q in R, qQ}

    q

    We can compute bothin O(n log n) time;on grid terrains: O(n)

    C

  • 8/10/2019 Fuzzy Flow

    34/59

    Core watersheds do not give a good definition of

    persistent water flow..

    Understanding Core Watersheds

    W(q) =

    RRT

    {p: p flows to q in R}

    U d d C W h d

  • 8/10/2019 Fuzzy Flow

    35/59

    Core watersheds do not give a good definition of

    persistent water flow..

    Understanding Core Watersheds

    W(q) =

    RRT

    {p: p flows to q in R}

    q

    upper terrain

    lower terrain

    U d d C W h d

  • 8/10/2019 Fuzzy Flow

    36/59

    Core watersheds do not give a good definition of

    persistent water flow..

    W(q)

    Understanding Core Watersheds

    W(q) =

    RRT

    {p: p flows to q in R}

    q

    upper terrain

    lower terrain

    realization

    U d di C W h d

  • 8/10/2019 Fuzzy Flow

    37/59

    Core watersheds do not give a good definition of

    persistent water flow..

    W(q) ={q} W(q)

    Understanding Core Watersheds

    .. only contains q!

    W(q) =

    RRT

    {p: p flows to q in R}

    q

    upper terrain

    lower terrain

    realization

    U d di C W h d

  • 8/10/2019 Fuzzy Flow

    38/59

    Core watersheds do not give a good definition of

    persistent water flow..

    W(q) ={q}

    Understanding Core Watersheds

    p

    .. only contains q!potential localminimum

    W(q) =

    RRT

    {p: p flows to q in R}

    q

    upper terrain

    lower terrain

    realization

    U d di C W h d

  • 8/10/2019 Fuzzy Flow

    39/59

    Understanding Core Watersheds

    W(q) =

    RRT

    {p: p flows to q in R}

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    U d di C W h d

  • 8/10/2019 Fuzzy Flow

    40/59

    Understanding Core Watersheds

    = RRT

    {p: p does not flow to q in R}c

    W(q) =

    RRT

    {p: p flows to q in R}

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    U d di C W h d

  • 8/10/2019 Fuzzy Flow

    41/59

    Understanding Core Watersheds

    = RRT

    {p: p does not flow to q in R}c

    W(q) =

    RRT

    {p: p flows to q in R}

    (ii) nodes outsideW(q)

    (i) potential local minima

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    Contained in this set:

    (iii) nodes with flow paths to (i) or (ii)

    U d di C W h d

  • 8/10/2019 Fuzzy Flow

    42/59

    Understanding Core Watersheds

    = RRT

    {p: p does not flow to q in R}c

    W(q) =

    RRT

    {p: p flows to q in R}

    (ii) nodes outsideW(q)

    (i) potential local minima

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    Contained in this set:

    (iii) nodes with flow paths to (i) or (ii)

    Vmin

    U d di C W h d

  • 8/10/2019 Fuzzy Flow

    43/59

    Understanding Core Watersheds

    = RRT

    {p: p does not flow to q in R}c

    W(q) =

    RRT

    {p: p flows to q in R}

    (ii) nodes outsideW(q)

    (i) potential local minima

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    Contained in this set:

    (iii) nodes with flow paths to (i) or (ii)

    Vmin (W(q))c

    U d t di C W t h d

  • 8/10/2019 Fuzzy Flow

    44/59

    Understanding Core Watersheds

    = RRT

    {p: p does not flow to q in R}c

    W(q) =

    RRT

    {p: p flows to q in R}

    (ii) nodes outsideW(q)

    (i) potential local minima

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    Contained in this set:

    (iii) nodes with flow paths to (i) or (ii)

    W ( Vmin (W(q))c )

    U d t di C W t h d

  • 8/10/2019 Fuzzy Flow

    45/59

    Understanding Core Watersheds

    = RRT

    {p: p does not flow to q in R}c

    W(q) =

    RRT

    {p: p flows to q in R}

    (ii) nodes outsideW(q)

    (i) potential local minima

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    Contained in this set:

    (iii) nodes with flow paths to (i) or (ii)

    W ( Vmin (W(q))c )

    Caution! Avoid theflow paths through q.

    U d t di C W t h d

  • 8/10/2019 Fuzzy Flow

    46/59

    Understanding Core Watersheds

    = RRT

    {p: p does not flow to q in R}c

    W(q) =

    RRT

    {p: p flows to q in R}

    (ii) nodes outsideW(q)

    (i) potential local minima

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    Contained in this set:

    (iii) nodes with flow paths to (i) or (ii)

    W\q ( Vmin (W(q))

    c )

    Caution! Avoid theflow paths through q.

    U d t di C W t h d

  • 8/10/2019 Fuzzy Flow

    47/59

    Understanding Core Watersheds

    = RRT

    {p: p does not flow to q in R}c

    W(q) =

    RRT

    {p: p flows to q in R}

    (ii) nodes outsideW(q)

    (i) potential local minima

    =

    W\q ( Vmin (W(q))

    c )

    c

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    Contained in this set:

    (iii) nodes with flow paths to (i) or (ii)

    U d t di C W t h d

  • 8/10/2019 Fuzzy Flow

    48/59

    Understanding Core Watersheds

    = RRT

    {p: p does not flow to q in R}c

    W(q) =

    RRT

    {p: p flows to q in R}

    (ii) nodes outsideW(q)

    (i) potential local minima

    =

    W\q ( Vmin (W(q))

    c )

    c

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    Contained in this set:

    (iii) nodes with flow paths to (i) or (ii)

    Alternative Definition:

    W(q) = W\q ( ( W(q) )

    c)

    c

    U d t di C W t h d

  • 8/10/2019 Fuzzy Flow

    49/59

    Understanding Core Watersheds

    = RRT

    {p: p does not flow to q in R}c

    W(q) =

    RRT

    {p: p flows to q in R}

    (ii) nodes outsideW(q)

    (i) potential local minima

    =

    W\q ( Vmin (W(q))

    c )

    c

    Core Watersheds are the complement of the set of

    nodes that have alternative destinations

    Contained in this set:

    (iii) nodes with flow paths to (i) or (ii)

    Alternative Definition:

    W(q) = W\q ( ( W(q) )

    c)

    c

    Persistent Watersheds

    P i t t W t h d P ti

  • 8/10/2019 Fuzzy Flow

    50/59

    Persistent Watersheds Properties

    In general, persistent watersheds and potential

    watersheds are not nested!

    Persistent Watersheds Properties

  • 8/10/2019 Fuzzy Flow

    51/59

    Persistent Watersheds Properties

    In general, persistent watersheds and potential

    watersheds are not nested!

    On regular terrains, we can prove:

    q

    p

    Let p W(q)

    after removing

    avoidable local minima

    Persistent Watersheds Properties

  • 8/10/2019 Fuzzy Flow

    52/59

    Persistent Watersheds Properties

    In general, persistent watersheds and potential

    watersheds are not nested!

    On regular terrains, we can prove:

    q

    p

    Let p W(q)

    (i) W(p)W(q)

    after removing

    avoidable local minima

    Persistent Watersheds Properties

  • 8/10/2019 Fuzzy Flow

    53/59

    Persistent Watersheds Properties

    In general, persistent watersheds and potential

    watersheds are not nested!

    On regular terrains, we can prove:

    q

    p

    Let p W(q)

    (i) W(p)W(q)

    (ii)W(p)W(q)

    after removing

    avoidable local minima

    Fuzzy watersheds

  • 8/10/2019 Fuzzy Flow

    54/59

    Fuzzy watersheds

    Fuzzy watersheds

  • 8/10/2019 Fuzzy Flow

    55/59

    Fuzzy watersheds

    Persistent

    Minimum

    Fuzzy watersheds

  • 8/10/2019 Fuzzy Flow

    56/59

    Fuzzy watersheds

    PotentialWatersheds

    Persistent

    Minimum

    Fuzzy watersheds

  • 8/10/2019 Fuzzy Flow

    57/59

    Fuzzy watersheds

    PotentialWatersheds

    Persistent

    Watershed

    Persistent

    Minimum

    Fuzzy watersheds

  • 8/10/2019 Fuzzy Flow

    58/59

    Fuzzy watersheds

    PotentialWatersheds

    Fuzzy ridge!

    Persistent

    Watershed

    Persistent

    Minimum

    Fuzzy watersheds

  • 8/10/2019 Fuzzy Flow

    59/59

    Fuzzy watersheds

    PotentialWatersheds

    Fuzzy ridge!

    Persistent

    Watershed

    Thank you!

    Persistent

    Minimum