# 离散化规则配置
# 定义所有连续因子的分箱规则
# 包含暴雨灾害链和地震灾害链的全部因子
#
# 2026-06-06: 基于1201个样本的实际数据分布，采用分位数分箱（等频分箱）
# 替代原有等宽分箱，使每个区间样本量更均匀

# ============================================
# 暴雨触发层离散化规则（保持气象标准不变）
# ============================================

rain_intensity:
  description: "降雨强度等级"
  unit: "mm/h"
  # 中国气象局降雨等级标准
  bins: [0, 0.2, 5, 15, 30, 70, 140, 99999]
  labels: [no_rain, light, moderate, heavy, storm, downpour, extreme]

duration:
  description: "持续时间"
  unit: "h"
  bins: [1, 3, 12, 99999]
  labels: [short, medium, long]

accum_rain:
  description: "累计降雨量"
  unit: "mm"
  # 中国气象局降水等级标准
  bins: [0, 10, 25, 50, 100, 99999]
  labels: [trace, light, moderate, heavy, extreme]

# ============================================
# 地震触发层离散化规则
# ============================================

magnitude:
  description: "地震震级"
  unit: "Richter"
  # 基于Keefer (1984) 地震触发地质灾害的震级阈值
  bins: [0, 4.0, 5.0, 6.0, 7.0, 8.0, 10.0]
  labels: [minor, light, moderate, strong, major, great]

epicenter_distance:
  description: "震中距"
  unit: "km"
  # 地震地质灾害影响范围（Keefer 1984）
  bins: [0, 30, 100, 300, 99999]
  labels: [very_near, near, moderate, far]

seismic_intensity:
  description: "地震烈度"
  unit: "中国烈度表"
  # GB 18306-2015 中国地震动参数区划图
  bins: [0, 5, 7, 9, 12, 99]
  labels: [minor, light, moderate, severe, extreme]

# ============================================
# 环境层离散化规则（暴雨/地震共享）
# 基于1201个样本的分位数分箱
# ============================================

elevation:
  description: "高程"
  unit: "m"
  # 数据: [356, 1934], 均值764.3±317.89, 偏度0.973
  # 分位数: [356, 470, 624, 792, 1016, 1934]
  bins: [356, 470, 624, 792, 1016, 1934]
  labels: [very_low, low, medium, high, very_high]

slope:
  description: "坡度"
  unit: "度"
  # 数据: [0.11, 47.14], 均值9.42±8.57, 偏度1.433
  # 分位数: [0.11, 1.81, 5.43, 9.48, 15.13, 47.14]
  bins: [0.11, 1.81, 5.43, 9.48, 15.13, 47.14]
  labels: [very_low, low, medium, high, very_high]

aspect:
  description: "坡向"
  unit: "度"
  # 数据: [0.86, 359.12], 均值173.47±104.96, 偏度0.126
  # 分位数: [0.86, 57.39, 135.0, 230.19, 297.83, 359.12]
  bins: [0.86, 57.39, 135.0, 230.19, 297.83, 359.12]
  labels: [flat, north, east, south, west]

soil_type:
  description: "土壤分类（中国土壤分类系统）"
  unit: "分类代码"
  mapping:
    0: ultisol        # 老成土
    6: entisol        # 初育土
    11: fluvo_aquic   # 潮土
    18: yellow_brown  # 黄棕壤
  default: entisol

lithology:
  description: "岩性（中国地质分类）"
  unit: "分类代码"
  mapping:
    1: acid_rock          # 酸性侵入岩（花岗岩等，SiO₂>66%）
    3: basic_rock         # 基性侵入岩（辉长岩等，SiO₂ 45-52%）
    4: basic_rock         # 基性火山岩（玄武岩等，合并入基性岩）
    5: carbonate          # 碳酸盐岩（石灰岩、白云岩）
    10: metamorphic       # 变质岩（片麻岩、大理岩）
    11: mixed_clastic     # 混合碎屑沉积岩（砂岩+泥岩互层）
    13: terrigenous       # 陆源碎屑岩（砂岩、粉砂岩）
    14: unconsolidated    # 松散堆积物（黄土、冲洪积）
  default: unconsolidated

landuse:
  description: "土地利用类型"
  unit: "分类代码"
  mapping:
    10: forest       # 林地
    30: farmland     # 农田
    40: urban        # 城市
    50: water        # 水域
    60: barren       # 裸地
    80: farmland     # 耕地（合并入农田）
  default: farmland

terrain:
  description: "地形分类（中国地形分类体系）"
  unit: "分类代码"
  mapping:
    1: mountain       # 断裂山麓地带（秦岭北麓）
    2: plain          # 平坦平原（渭河平原）
    3: deep_valley    # 高山深峡谷（秦岭腹地）
    4: hill           # 丘陵（黄土塬）
    5: gentle_hill    # 低缓丘陵（塬边过渡带）
    6: low_mountain   # 低山（骊山等）
    7: flat_plain     # 平缓平原（冲积平原）
  default: hill

impervious:
  description: "不透水率"
  unit: "百分比"
  # 数据: [0.0, 97.2], 均值16.40±25.99, 偏度1.787
  # 26.9%为0.0（无硬化地表），非零值右偏分布
  # 分箱策略：0单独一类，其余4等分（分位数分箱）
  # 分位数(非零): [2.0, 9.95, 31.8, 97.2]
  bins: [0.0, 0.01, 2.0, 10.0, 32.0, 97.2]
  labels: [none, very_low, low, medium, high]

ndvi:
  description: "植被指数"
  unit: "NDVI值"
  # 数据: [1.25, 38.68], 均值20.67±5.87, 偏度-0.106
  # 分位数: [1.25, 17.09, 20.3, 22.4, 25.2, 38.68]
  bins: [1.25, 17.09, 20.3, 22.4, 25.2, 38.68]
  labels: [very_low, low, medium, high, very_high]

sand_content:
  description: "土壤含沙量"
  unit: "百分比"
  # 数据: [23.0, 52.0], 均值34.43±4.29, 偏度0.538
  # 分位数: [23.0, 31.0, 33.0, 35.0, 37.0, 52.0]
  bins: [23.0, 31.0, 33.0, 35.0, 37.0, 52.0]
  labels: [very_low, low, medium, high, very_high]

ph:
  description: "土壤PH值"
  unit: "PH值"
  # 数据: [59.0, 81.0], 均值71.79±4.14, 偏度-0.398
  # 分位数: [59.0, 68.0, 72.0, 74.0, 76.0, 81.0]
  bins: [59.0, 68.0, 72.0, 74.0, 76.0, 81.0]
  labels: [very_low, low, medium, high, very_high]

soil_moisture:
  description: "土壤湿度"
  unit: "百分比"
  # 数据: [0.0, 41.1], 均值32.02±14.92, 偏度-1.676
  # 约10%为0.0（缺失/极端干燥），其余集中在37-41
  # 分位数: [0.0, 37.7, 38.6, 38.9, 39.4, 41.1]
  bins: [0.0, 37.0, 38.5, 39.5, 41.1]
  labels: [very_low, low, medium, high]

organic_carbon:
  description: "有机碳"
  unit: "百分比"
  # 数据: [0.0, 73.0], 均值38.36±19.14, 偏度-1.187
  # 分位数: [0.0, 34.0, 41.0, 47.0, 53.0, 73.0]
  bins: [0.0, 34.0, 41.0, 47.0, 53.0, 73.0]
  labels: [very_low, low, medium, high, very_high]

dist_to_river:
  description: "距离河道距离"
  unit: "米"
  # 数据: [12.21, 29904.99], 均值11003.92±6582.23, 偏度0.271
  # 分位数: [12.21, 5165.0, 9003.0, 12424.97, 16431.82, 29904.99]
  bins: [12.21, 5165.0, 9003.0, 12424.97, 16431.82, 29904.99]
  labels: [very_close, close, moderate, far, very_far]

dist_to_fault:
  description: "距离断裂带距离"
  unit: "米"
  # 数据: [1.74, 14542.53], 均值3448.52±3406.56, 偏度1.055
  # 分位数: [1.74, 476.69, 1433.62, 3334.87, 6502.28, 14542.53]
  bins: [1.74, 476.69, 1433.62, 3334.87, 6502.28, 14542.53]
  labels: [very_close, close, moderate, far, very_far]

pipe_density:
  description: "供水管网密度"
  unit: "m/m²"
  # 数据: [0.0, 0.07], 约80%为0.0，90%分位数0.013，95%分位数0.023
  # 分位数: [0.0, 0.013, 0.023, 0.065]
  bins: [0.0, 0.013, 0.023, 0.065]
  labels: [none, low, medium, high]