Premium
Applying Pattern Oriented Sampling in current fieldwork practice to enable more effective model evaluation in fluvial landscape evolution research
Author(s) -
Briant Rebecca M.,
Cohen Kim M.,
Cordier Stephane,
Demoulin Alain J.A.G.,
Macklin Mark G.,
Mather Anne E.,
Rixhon Gilles,
Veldkamp Tom,
Wainwright John,
Whittaker Alex,
Wittmann Hella
Publication year - 2018
Publication title -
earth surface processes and landforms
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.294
H-Index - 127
eISSN - 1096-9837
pISSN - 0197-9337
DOI - 10.1002/esp.4458
Subject(s) - field (mathematics) , fluvial , identification (biology) , data collection , sampling (signal processing) , computer science , key (lock) , range (aeronautics) , data science , environmental resource management , data mining , environmental science , geology , ecology , engineering , geomorphology , structural basin , statistics , mathematics , computer security , filter (signal processing) , aerospace engineering , pure mathematics , computer vision , biology
Field geologists and geomorphologists are increasingly looking to numerical modelling to understand landscape change over time, particularly in river catchments. The application of landscape evolution models (LEMs) started with abstract research questions in synthetic landscapes. Now, however, studies using LEMs on real‐world catchments are becoming increasingly common. This development has philosophical implications for model specification and evaluation using geological and geomorphological data, besides practical implications for fieldwork targets and strategy. The type of data produced to drive and constrain LEM simulations has very little in common with that used to calibrate and validate models operating over shorter timescales, making a new approach necessary. Here we argue that catchment fieldwork and LEM studies are best synchronized by complementing the Pattern Oriented Modelling (POM) approach of most fluvial LEMs with Pattern Oriented Sampling (POS) fieldwork approaches. POS can embrace a wide range of field data types, without overly increasing the burden of data collection. In our approach, both POM output and POS field data for a specific catchment are used to quantify key characteristics of a catchment. These are then compared to provide an evaluation of the performance of the model. Early identification of these key characteristics should be undertaken to drive focused POS data collection and POM model specification. Once models are evaluated using this POM/POS approach, conclusions drawn from LEM studies can be used with greater confidence to improve understanding of landscape change. © 2018 John Wiley & Sons, Ltd.