Page 8 - i1052-5173-29-6
P. 8

A

















                                                         B














                      Figure 3. Landslide density distributions and geologic formations. (A) Map showing geologic formations with >10% by
                      area high landslide density and with >10% by area absent landslides. (B) Chart showing formations with >10% by area
                      with high landslide density. Ku—granodiorite; Thp—hornblende quartz diorite; Kap—basalt lava and tuffaceous sand-
                      stone; Kco—basaltic lava and volcanic breccia; TKdi—diorite; Trhp—rhyodacite porphyry; Kav—andesitic lava;
                      Kmam—andesite lava and volcanic sandstone and siltstone; TKn—crystal tuff and tuffaceous clastics; Kln—basaltic
                      breccia and basalt lava; Kja—amphibolite; Kper—basalt lava and tuff, volcanic sandstone (Bawiec, 1998).

          COMPARISON WITH                    from 287 to 579 mm, and although these   CONCLUSIONS
          PAST EVENTS                        values vary by as much as a factor of two,   Although landsliding is frequent on
            To place Hurricane Maria in the context   even the minimum is nearly twice as large   steep hillslopes in Puerto Rico, the abun-
          of previous landslide-generating TCs, we   as the median MMR for all previous wide-  dance and widespread extent of landslides
          compiled rainfall data and information on   spread landsliding events (152 mm).  that occurred during Hurricane Maria
          the extent and characteristics of landslides   Although TC-triggered widespread land-  were unprecedented in comparison to
          from 72 TCs that affected Puerto Rico   sliding occurs frequently in Puerto Rico,   previously recorded hurricane or tropical
          between 1960 and 2017 (see [B] in the GSA   the number and spatial extent of landslides   storm-triggered landsliding events.
          Data Repository [see footnote 1]). Based on   triggered by Hurricane Maria were unprec-  Hurricane Maria provided a unique data
          this information, we classified the extent of   edented in comparison to previously docu-  set to examine the characteristics and con-
          landsliding as either none, isolated, fre-  mented events. For example, Hurricane   tributing factors of landslides with variable
          quent, or widespread for each TC (Table S2   Hugo (1989) and the precursor to Tropical   failure types and material properties as a
          [see footnote 1]; see Table 1 for classifica-  Storm Isabel (1985) both produced large   means of improving future hazard assess-
          tion criteria). In addition to reporting maxi-  amounts of rain but only triggered land-  ments throughout Puerto Rico.
          mum rainfall (Fig. 4A), we calculated the   slides that were confined to localized areas   While we have shown that the average
          average rainfall in mountainous terrain   (Jibson, 1989; Larsen and Torres-Sanchez,   rainfall from Hurricane Maria in mountain-
          (mean mountain rainfall [MMR], Fig. 4B)   1992). Other TC-triggered landslide events   ous areas was greater than that of any other
          for 56 TCs for which data were available.   have not been documented systematically,   hurricane or tropical storm in Puerto Rico
          We used contoured rainfall maps (National   and typically refer to “numerous” landslides   since 1960, we also found that inconsisten-
          Centers for Environmental Prediction,   in one or more municipalities or regions of   cies among rainfall data sets and the coarse
          2018) to extract the average rainfall in areas   Puerto Rico (Table S2 [see footnote 1]).   resolution of available data did not allow for
          with slopes >20° for each historical TC to   Based on our island-wide landslide density   a meaningful assessment of the correlation
          compare with all three rainfall data sets   map (Fig. 1B), we estimate that the number   between rainfall amount and landslide dis-
          from Hurricane Maria. On average,   of landslides triggered by Hurricane Maria   tribution. Satellite-based SMAP measure-
          Hurricane Maria produced more rain than   (at least 40,000) was approximately two   ments from both before and after Hurricane
          any other TC to affect Puerto Rico in the   orders of magnitude greater than the hun-  Maria showed larger differences between
          58-year record (Fig. 4). Our calculated   dreds of landslides reported from previous   low and high landslide density areas, indi-
          MMR values for Hurricane Maria range   detailed inventories.          cating that, unsurprisingly, soil moisture

       8  GSA Today  |  June 2019
   3   4   5   6   7   8   9   10   11   12   13