SelectionEarthquake

Deprem kayitlarini AFAD ve PEER gibi veri saglayicilardan cekip ortak kolona normalize eden, arama kriterlerine gore puanlayan ve TBDY 2018 odakli kayit secimi yapan Python kutuphanesi.

Proje Metrikleri

Son yerel dogrulama:

• Surum: 1.2.1
• Test komutu: python -m pytest
• Coverage komutu: python -m pytest --cov=selection_service --cov-report=term-missing
• Test sonucu: 414 passed
• Toplam coverage: 81%
• Dogrulama ortami: Python 3.13.5, pytest 8.4.2, pytest-cov 7.0.0

Paket ve Import Adi

Paket kurulum adi:

pip install earthquake-selection

Python import paketi:

import selection_service

Yerel gelistirme kurulumu:

git clone https://github.com/muhammedsural/SelectionEarthquake.git
cd SelectionEarthquake
pip install -e ".[dev]"

Ozellikler

• Coklu veri saglayici destegi: AFAD, PEER
• Ortak SearchCriteria modeli ve provider bazli kriter donusumu
• Pipeline tabanli sorgu, birlestirme, puanlama ve secim akisi
• TBDY 2018 icin Gaussian tabanli secim stratejisi
• Kriter bazli skor kirilimi: SCORE_BREAKDOWN
• Her kayit icin secim/eleme aciklamasi: SELECTION_STATUS, SELECTION_REASON
• CSV, JSON rapor ve Pandas DataFrame ciktilari
• AFAD icin waveform indirme destegi
• Pytest tabanli test altyapisi

Hizli Baslangic

from selection_service.core.EarthquakeApi import EarthquakeAPI
from selection_service.enums.Enums import DesignCode, ProviderName
from selection_service.processing.Selection import (
    ScoringWeights,
    SearchCriteria,
    SelectionConfig,
    TBDYSelectionStrategy,
)

config = SelectionConfig(
    design_code=DesignCode.TBDY_2018,
    num_records=11,
    max_per_station=3,
    max_per_event=3,
    min_score=55,
)
strategy = TBDYSelectionStrategy(config=config)

criteria = SearchCriteria(
    start_date="2000-01-01",
    end_date="2025-09-05",
    min_magnitude=7.0,
    max_magnitude=8.0,
    min_vs30=300,
    max_vs30=400,
    mechanisms=["StrikeSlip"],
    weights=ScoringWeights.from_preset("tbdy_2018_record_selection"),
)

api = EarthquakeAPI(
    provider_names=[ProviderName.PEER],
    strategies=[strategy],
    use_cache=True,
)

result = api.run_sync(criteria=criteria, strategy_name=strategy.get_name())

if result.success:
    selected = result.value.selected_df
    report = result.value.report
    print(selected[["PROVIDER", "RSN", "EVENT", "MAGNITUDE", "SCORE", "SELECTION_REASON"]].head())
    print(report["selection_summary"])
else:
    print(result.error)

Komut Satiri Ornegi

Kurulumdan sonra uc uca arama, secim, CSV ve JSON rapor uretimi:

quake-sel --providers peer --num-records 11 --report-path selection_report.json --selected-csv selected_records.csv

AFAD ile waveform indirme akisini da baslatmak icin:

quake-sel --providers afad --download-waveforms --export-type mseed

AFAD API ag baglantisi gerektirir. PEER veri seti lokal flatfile ile calisir ve waveform indirme desteklemez; download adiminda desteklenmeyen provider atlanir.

Scoring Presetleri

Hazir agirlik setleri ScoringWeights.from_preset(...) ile secilir:

• balanced: Varsayilan dengeli agirliklar.
• tbdy_2018_record_selection: Magnitude, mesafe, Vs30 ve mekanizmayi one cikarir.
• site_response: Vs30, sure ve siddet olcutlerine daha fazla agirlik verir.

Preset aciklamalari:

from selection_service.processing.Selection import ScoringWeights

print(ScoringWeights.preset_descriptions())

Rapor Izlenebilirligi

PipelineResult.report asagidaki ek alanlari icerir:

• selection_summary: Secilen/reddedilen kayit sayilari ve eleme gerekceleri.
• score_breakdown: Secilen kayitlar icin kriter bazli hedef, deger, agirlik ve katki.

scored_df tum kayitlar icin su kolonlari tasir:

• SCORE_BREAKDOWN
• SELECTION_STATUS
• SELECTION_REASON

Mimari

selection_service/
  core/        Pipeline, API facade, config, error types
  services/    Query, provider registry, waveform download orchestration
  providers/   AFAD, PEER, cache, provider interfaces
  processing/  SearchCriteria, selection strategy, mappers, Result
  utility/     Data file loading helpers
  data/        Packaged CSV/XLSX data
tests/         pytest suite
examples/      Usage examples

Test

pytest

Pytest ayarlari tek kaynak olarak pyproject.toml icindedir.

Yol Haritasi

• FDSN waveform sonuclarini dosyaya aktarma ve secim akisiyla iliskilendirme.
• Eurocode stratejisini gercek kurallarla doldur veya deneysel olarak etiketle.
• Rapor ciktilarini HTML/PDF formatina genislet.

Lisans

MIT License