The use of land- and satellite-based precipitation radar

The use of land- and satellite-based precipitation radar to forecast debris flows and water discharge:

case study from June 2nd, 2016 in southern Norway.Graziella Devoli (1,2), Zelalem T. Mengistu (1), Christoffer A. Elo (3),

Søren Boje(1), Snorre S. Rønning(3), Kolbjørn Engeland (1,2), and Cristian Lussana (3)

(1) Norwegian Water Resources and Energy Directorate (NVE), Forecast of flood and landslide hazard, Oslo, Norway (2) Department of Geosciences, University of Oslo, Oslo, Norway

(3) Norwegian Meteorological Institute (MET), Oslo, Norway

Studentdagen UMB& UiO7 November 2017

Norwegian Water Resources and Energy Directorate

High intensity and short duration rainfallsand daily landslide hazard assessment■ High intensity and short duration (less than 1 hour) rainfalls may cause sudden and

abundant runoff that can entrain large quantities of loose sediments and originate debris flows.

■ Intense convective rainstorms often develop quickly, especially during summer, and they are difficult to forecast and even to observe with a standard (synoptic) network of precipitation gauges.

■ In those cases, the forecaster on duty can send warning messages for a very large area (encompassing many counties and many municipalities), because of the large spatial uncertainty of the prognoses and amount of rain. A standard sentence in the warning message is always included, recommending to the population to monitor the evolution of the rainstorm with weather radar products, which are available on institutional websites.

■ In other cases, especially when the convective rainstorm is spatially confined in a small area and highly uncertain, the forecaster may choose to not issue any warning.

■ The first situation yields false alarms for some areas, while the second situation could result in a missing event, if a landslide actually occurs.


3

Number ofradar, positionand period ofoperation:

10 radar (more or less operational from 1992 to present)

only radar data from 2013 to test withlandslides

Buffer zones: 1 km, 9 km, 25 km, 120 km


Thursday 2.06.2016, prognosisForecastedDaily rainfall Forecasted

Landslidethresholds

Situation: Intense rainfall in Rogaland over 50 mm, in xgeo.no (supporting web tool)From met.briefing (probably convective but uncertain in space and amount, probably most offshore)


Daily rainfall from xgeo forecasted for Thursday 2.06.2016

No - warning- No evidence from threshold (Hydmet index) or other

parameters - Uncertain prognosis and spread shower activity- Too small area to warn, less than a county size- Poor historical landslide data in the area, Rarely

significant debris slides and debris flows have beenrecorded.

- Many false alarms in this area - South Rogaland not highly susceptibile to landslide

(only a few catchments more prone)

Yes - warning- Intense rainfalls 3 days before in the same area

ConclusionsGreen, «but if MET will send obs varsel, we will add the sentencetalking about possible landslide activity in the area»


Case study

Bilde: Politiet

Reported in many newspapers:NRK, Stavanger AftenbladGjesdalbuen

Ikke varslet

http://www.aftenbladet.no/nyheter/lokalt/gjesdal/Se-Aftenbladets-luftfoto-fra-rasomradet-3937350.html

Debris flow along Fv 45Rogaland, 2.06.2016Reported at ca. 16:30, but occurred before


Questions…

■ What kind of landslide type was?■ Why happened there? ■ What caused it?■ If it was rain? How much? ■ Could have been forecasted?

7


Bilde fra Jarle Aasland

Bilde fra Statens Vegvesen

Bilde. Politiet

Bilde fra Statens Vegvesen


Sentinel 219.05.2017


The area was detected to be landslide prone in thesusceptibility model proposedby NGU

The entire catchment is highly prone to landslide

Why here? Geological/Topographic settingSteep slope, Steep channelConcave topography in the upper part perfect to collectwater and snowMoraine and loose deposit available and easy to be eroded

Observed 24h rainfalldata at nearest stations

Søyland i Gjesdal

25 mm

Maudal

18 mm

Interpolated rainfall observations from rain gauges (grid cell 1km2)

Rain or other factors?


Last MODIS picture free-clouds is from 28th of May (5 days before), confirmed by other photos from NPRA

Still snow in the area? Probably not?

Source: Statens Vegvesen

This is in disagreement with the eye-witness that observed intense and very local showers.


Radar data

Intense rain?

Yes?...

…but how much?

16:15

10 mm/h?


Radar data: Preliminary analysis

Radar precipitation amounts: up to 45 mm (24h). The rain last approximately 2 hours (14:30-16:00) with intense peaks around 15:00-15:30 up to 35 mm. It was extremely local and therefore difficult to forecast.

Analyses of 3h radar data and discharge responsefor threecatchments during 2nd -3rd of June

Analysis of dischargeresponse at nearbycatchments

Radar, 3h values at 15:00 Interpolated rainfall observationsfrom rain gauges (grid cell 1km2)


Lightning data: number oflightning in a very localizedarea.


IMERG: Satellite based weather radar data from NASA-JAXA – kl. 12:00 16:00

6,30

59,35

IMERG = Integrated Multi-satellite Retrievals for GPM (Global PrecipitationMeasurement mission)

++Gratis data, Free of clouds, Cover Atlantic -- 4-5 timer data-delay; 10 km resolution, <60°N

https://svs.gsfc.nasa.gov/cgi-bin/details.cgi?aid=11784https://earthdata.nasa.gov/trmm-to-gpmhttp://www.livescience.com/50341-worldwide-water-watched-in-time-lapse-video.html

https://svs.gsfc.nasa.gov/cgi-bin/details.cgi?aid=11784

https://earthdata.nasa.gov/trmm-to-gpm

http://www.livescience.com/50341-worldwide-water-watched-in-time-lapse-video.html


Conclusions

■ Analysis of rainfall intensity estimated by both land-based and satellite-based (IMERG) radar data confirms the eye-witness observations, and it results in significantly higher values for the area where the debris flows occurred, if compared to precipitation interpolated from gauge observations.

■ This was also supported by discharge responses from three small catchments in the area.

■ Weather radar and lighting data are useful complements to the traditional analysis of landslide events made only by means of gauges, moreover they can be used: ■ in back analyses on rainfall and landslide events in order to improve landslide

thresholds;■ has a potential to assist in now-casting operations as supporting tool of a regional

warning, especially in summer season, and radar prediction can be used in the proximate hour to see the storm development.


Thanks

Documents

The use of land- and satellite-based precipitation radar