Omar,

Your query is very pertinent. Landslides seldom occur in areas closely monitored for precipitation. Not only that, many landslides occur as the result of intense rainfall of short duration related to thunderstorm activity. I have seen one debris flow triggered on one side of a narrow valley where debris flow activity was historically very low. Yet on the other side of the valley where historically frequent debris flow activity occurred, there was none at at all. Clearly the intense rainfall was highly localised and no rain gauge could have picked it up unless located within a couple of hundred metres of the flow.

There is a way round the problem and that is to look out for river/stream flow patterns over the time when the landslide happened. It may be that the rivers are being monitored on a continuous basis, in which case they may provide circumstantial evidence of the time of the landslide occurrence. I have had recourse to that before now. Otherwise tracking down meaningful rainfall data can prove elusive.

regards

George Strachan

Very informative dear George J. Strachan "Clearly the intense rainfall was highly localized and no rain gauge could have picked it up unless located within a couple of hundred meters of the flow".

• Yes, and that is geographically dependent (for instance, in Korea we might not have the localized precipitation. However, the topographical settings play the main role.
• After I scanned the previous studies, I noticed that, taking in mind mainly 3 components (horizontal and vertical distance, and watershed basin) as main driving forces for better selection of related rain gauges.
• The details about the horizonal distance is well introduced, while the vertical distance and the related issue, like, whether the difference in vertical distance should be taken with reference to rain gauge and landslide location (which is higher is importance).
• Still more assumption need to discussed here.

Best regards,

I agree with Dr. Strachan, and can relate personally to his responses and opinions. Our local and national records include real time or near real time Doppler coverage of storm events, which has estimates of storm intensity and rainfall. But I am not sure how this data might be stored for use relative to previous events. If your weather agency collects and stores these Doppler data, it might be useful.

Unfortunately, many countries and localities do not collect frequent rainfall data with weighing or tipping bucket recording rainguages. And many landslide studies do not only not have detail rainfall data, but they in many or most instances do not include topics such as past or recent hydrological modifications or potential changes that include observations such as jackstrawed (tipping in various directions) trees, timber cutting or vegetation changes, various types of surface drainage changes, such as roads, animal or human trails, ditches or other activities such as with heavy equipment, geology, soils, stratigraphy of any faults and dips, bedrock, etc. Even recent earthquakes may contribute.

Rainfall amount, intensity and duration are probably the main factors we would look toward, but there are numerous other considerations or potential contributors. A few of my observations, opinions and information exchange with geologists and soil scientists were briefly mentioned in our paper on Watershed Restoration after Calamity in my Researchgate.

If you have access to the real time or historic weather Doppler, you will see that through time, not all severe storms come from the same direction, and in mountains where normally orographic increases on one side can change with direction of storm front or circulation. Some storms are very localized, and some spread out over larger areas. Other approaches such as krieging can be used if a network of raingauges was available, adjusted for topography in GIS when desired. Some storms are so localized that the stream gauging network that may sometimes be useful, would be unable to pick up local detail, or may pick up on other areas of intense rainfall. I don’t think there is any acceptable rainfall range that all would agree to. Meteorologists and hydrologists typically would review available data and its location, and then make any adjustments or calculations appropriate to apply to local situation, or use as just contributory background data. If a specific landslide prone area is being researched, then installing a network of raingauges seems prudent, but landslide prone terrain may stay stable for many years before failure. In reality, we will in most cases continue to use whatever rainfall or other information we can find to try to determine causal or contributory factors.