Hazard

Helpers for querying Hazard Realizations and related dynamodb models.

Provides efficient queries for the models: HazardAggregation, OpenquakeRealization & ToshiOpenquakeMeta.

Functions:

• get_hazard_metadata_v3() - returns iterator of matching ToshiOpenquakeMeta objects.
• get_rlz_curves_v3() - returns iterator of matching OpenquakeRealization objects.
• get_hazard_curves() - returns iterator of HazardAggregation objects.

get_hazard_curves(locs, vs30s, hazard_model_ids, imts, aggs=None)

Query the HazardAggregation table for matching hazard curves.

Examples:

>>>  get_hazard_curves(
        locs=['-46.430~168.360'],
        vs30s=[250, 350, 500],
        hazard_model_ids=['NSHM_V1.0.4'],
        imts=['PGA', 'SA(0.5)']
    )
>>> <generator object get_hazard_curves at 0x7f115d67be60>

Parameters:

Name	Type	Description	Default
locs	Iterable[str]	coded location strings e.g. ['-46.430~168.360']	required
vs30s	Iterable[int]	vs30 values eg [400, 500]	required
hazard_model_ids	Iterable[str]	hazard model ids e.. ['NSHM_V1.0.4']	required
imts	Iterable[str]	IntensityMeasureType values e.g ['PGA', 'SA(0.5)']	required
aggs	Union[Iterable[str], None]	aggregation values e.g. ['mean', '0.9']	None

Yields:

Type	Description
HazardAggregation	an iterator of the matching HazardAggregation models.

Source code in toshi_hazard_store/query/hazard_query.py

def get_hazard_curves(
    locs: Iterable[str],
    vs30s: Iterable[int],
    hazard_model_ids: Iterable[str],
    imts: Iterable[str],
    aggs: Union[Iterable[str], None] = None,
) -> Iterator[HazardAggregation]:
    """Query the HazardAggregation table for matching hazard curves.

    Examples:
        >>>  get_hazard_curves(
                locs=['-46.430~168.360'],
                vs30s=[250, 350, 500],
                hazard_model_ids=['NSHM_V1.0.4'],
                imts=['PGA', 'SA(0.5)']
            )
        >>> <generator object get_hazard_curves at 0x7f115d67be60>

    Parameters:
        locs: coded location strings e.g. ['-46.430~168.360']
        vs30s: vs30 values eg [400, 500]
        hazard_model_ids:  hazard model ids e.. ['NSHM_V1.0.4']
        imts: IntensityMeasureType values e.g ['PGA', 'SA(0.5)']
        aggs: aggregation values e.g. ['mean', '0.9']

    Yields:
        an iterator of the matching HazardAggregation models.
    """
    aggs = aggs or ["mean", "0.1"]

    log.info("get_hazard_curves( %s" % locs)

    # table classes may be rebased, this makes sure we always get the current class definition.
    mHAG = openquake_models.__dict__['HazardAggregation']
    log.debug(f"mHAG.__bases__ : {mHAG.__bases__}")

    def build_condition_expr(loc, vs30, hid, agg):
        """Build the filter condition expression."""
        grid_res = decimal.Decimal(str(loc.split('~')[0]))
        places = grid_res.as_tuple().exponent

        res = float(decimal.Decimal(10) ** places)
        loc = downsample_code(loc, res)

        expr = None

        if places == -1:
            expr = mHAG.nloc_1 == loc
        elif places == -2:
            expr = mHAG.nloc_01 == loc
        elif places == -3:
            expr = mHAG.nloc_001 == loc
        else:
            assert 0

        return expr & (mHAG.vs30 == vs30) & (mHAG.imt == imt) & (mHAG.agg == agg) & (mHAG.hazard_model_id == hid)

    # TODO: use https://pypi.org/project/InPynamoDB/
    total_hits = 0
    for hash_location_code in get_hashes(locs):
        partition_hits = 0
        log.info('hash_key %s' % hash_location_code)
        hash_locs = list(filter(lambda loc: downsample_code(loc, 0.1) == hash_location_code, locs))

        for hloc, hid, vs30, imt, agg in itertools.product(hash_locs, hazard_model_ids, vs30s, imts, aggs):

            sort_key_first_val = f"{hloc}:{vs30}:{imt}:{agg}:{hid}"
            condition_expr = build_condition_expr(hloc, vs30, hid, agg)

            log.debug('sort_key_first_val: %s' % sort_key_first_val)
            log.debug('condition_expr: %s' % condition_expr)

            results = mHAG.query(
                hash_key=hash_location_code,
                range_key_condition=mHAG.sort_key == sort_key_first_val,
                filter_condition=condition_expr,
                # limit=10,
                # rate_limit=None,
                # last_evaluated_key=None
            )

            log.debug("get_hazard_rlz_curves_v3: results %s" % results)
            for hit in results:
                partition_hits += 1
                yield (hit)

        total_hits += partition_hits
        log.info('hash_key %s has %s hits' % (hash_location_code, partition_hits))

    log.info('Total %s hits' % total_hits)

get_hazard_metadata_v3(haz_sol_ids, vs30_vals)

Query the ToshiOpenquakeMeta table

Parameters:

Name	Type	Description	Default
haz_sol_ids	Iterable[str]	list of solution ids to fetch	required
vs30_vals	Iterable[int]	vs30 values eg [400, 500]	required

Yields:

Type	Description
ToshiOpenquakeMeta	an iterator of the matching ToshiOpenquakeMeta objects

Source code in toshi_hazard_store/query/hazard_query.py

def get_hazard_metadata_v3(haz_sol_ids: Iterable[str], vs30_vals: Iterable[int]) -> Iterator[ToshiOpenquakeMeta]:
    """Query the ToshiOpenquakeMeta table

    Parameters:
        haz_sol_ids: list of solution ids to fetch
        vs30_vals: vs30 values eg [400, 500]

    Yields:
        an iterator of the matching ToshiOpenquakeMeta objects
    """
    mOQM = openquake_models.__dict__['ToshiOpenquakeMeta']

    total_hits = 0
    for tid, vs30 in itertools.product(haz_sol_ids, vs30_vals):
        sort_key_val = f"{tid}:{vs30}"
        log.info('sort_key_val: %s' % sort_key_val)

        for hit in mOQM.query(
            "ToshiOpenquakeMeta",  # NB the partition key is the table name!
            range_key_condition=(mOQM.hazsol_vs30_rk == sort_key_val),
        ):
            total_hits += 1
            yield (hit)

    log.info('Total %s hits' % total_hits)

get_rlz_curves_v3(locs, vs30s, rlzs, tids, imts)

Query the OpenquakeRealization table.

Parameters:

Name	Type	Description	Default
locs	Iterable[str]	coded location codes e.g. ['-46.430~168.360']	required
vs30s	Iterable[int]	vs30 values eg [400, 500]	required
rlzs	Iterable[int]	realizations eg [0,1,2,3]	required
tids	Iterable[str]	toshi hazard_solution_ids e.. ['XXZ']	required
imts	Iterable[str]	imt (IntensityMeasureType) values e.g ['PGA', 'SA(0.5)']	required

Yields:

Type	Description
OpenquakeRealization	HazardRealization models

Source code in toshi_hazard_store/query/hazard_query.py

def get_rlz_curves_v3(
    locs: Iterable[str],
    vs30s: Iterable[int],
    rlzs: Iterable[int],
    tids: Iterable[str],
    imts: Iterable[str],
    # model=None,
) -> Iterator[OpenquakeRealization]:
    """Query the OpenquakeRealization table.

    Parameters:
        locs: coded location codes e.g. ['-46.430~168.360']
        vs30s: vs30 values eg [400, 500]
        rlzs: realizations eg [0,1,2,3]
        tids: toshi hazard_solution_ids e.. ['XXZ']
        imts: imt (IntensityMeasureType) values e.g ['PGA', 'SA(0.5)']

    Yields:
        HazardRealization models
    """

    # table classes may be rebased, this makes sure we always get the current class definition.
    mRLZ = openquake_models.__dict__['OpenquakeRealization']

    def build_condition_expr(loc, vs30, rlz, tid):
        """Build the filter condition expression."""
        grid_res = decimal.Decimal(str(loc.split('~')[0]))
        places = grid_res.as_tuple().exponent

        res = float(decimal.Decimal(10) ** places)
        loc = downsample_code(loc, res)

        expr = None

        if places == -1:
            expr = mRLZ.nloc_1 == loc
        elif places == -2:
            expr = mRLZ.nloc_01 == loc
        elif places == -3:
            expr = mRLZ.nloc_001 == loc
        else:
            assert 0

        return expr & (mRLZ.vs30 == vs30) & (mRLZ.rlz == rlz) & (mRLZ.hazard_solution_id == tid)

    total_hits = 0
    for hash_location_code in get_hashes(locs):
        partition_hits = 0
        log.debug('hash_key %s' % hash_location_code)
        hash_locs = list(filter(lambda loc: downsample_code(loc, 0.1) == hash_location_code, locs))

        for hloc, tid, vs30, rlz in itertools.product(hash_locs, tids, vs30s, rlzs):

            sort_key_first_val = f"{hloc}:{vs30}:{str(rlz).zfill(6)}:{tid}"
            condition_expr = build_condition_expr(hloc, vs30, rlz, tid)

            log.debug('sort_key_first_val: %s' % sort_key_first_val)
            log.debug('condition_expr: %s' % condition_expr)

            results = mRLZ.query(
                hash_location_code,
                mRLZ.sort_key == sort_key_first_val,
                filter_condition=condition_expr,
            )

            # print(f"get_hazard_rlz_curves_v3: qry {qry}")
            log.debug("get_hazard_rlz_curves_v3: results %s" % results)
            for hit in results:
                partition_hits += 1
                hit.values = list(filter(lambda x: x.imt in imts, hit.values))
                yield (hit)

        total_hits += partition_hits
        log.debug('hash_key %s has %s hits' % (hash_location_code, partition_hits))

    log.info('Total %s hits' % total_hits)