Optimisation of forestry machinery operations to avoid problems of soil degradation and compaction

Scientific leaders: Noémie Pousse <noemie.pousse@onf.fr> (ONF, Chambéry), André Chanzy <andre.chanzy@inrae.fr> (INRAE, UMR EMMAH, Avignon) and Stéphane Ruy <stephane.ruy@inrae.fr> (INRAE, UMR EMMAH, Avignon).

The ForestOAD model has been developed in the VSoil platform. It is a tool to optimise the forestry machinery operations and avoid problems of soil degradation and compaction. ForestOAD is essentially based on an existing water balance model, for which minor simplifications and modifications have been made. One of the objectives is to be able to use weather forecasts to estimate changes in humidity over the coming days. As weather forecasts are updated regularly, it was essential to take them into account and therefore to be able to run the simulations again with a new weather forecast. In this context, the initial situation of the modules must correspond to the state of the variables calculated in a previous run for the start time of the new forecasts. A "STOP'n'RESUME" function has been implemented, which authorises the simulation, saves it at selected times and allows it to be restarted from one of the save dates. This required a number of modifications to be made to the VSoil platform and to the modules that use initial situations. The model also uses pedotransfer functions to calculate hydrodynamic properties. The final decision support model will therefore use databases to obtain the predicted climate and its variability, and soil databases to estimate the hydrodynamic properties. One of VSoil's functionalities is the ability to run sets of simulations using parameter distributions. This capability will provide estimates of the variables of interest and their sensitivity to climate and soil uncertainties.

French version:

Le modèle ForestOAD a été développé dans la plateforme VSoil. Il s'agit d'un outil permettant d'optimiser les opérations des machines forestières et d'éviter les problèmes de dégradation et de compactage du sol. Il repose essentiellement sur un modèle de bilan hydrique déjà existant et pour lequel de petites simplifications ou modifications ont été apportées. Un des objectifs est de pouvoir utiliser des prévisions météorologiques pour estimer l’évolution de l’humidité dans les jours qui arrivent. Les prévisions météo étant mises à jour régulièrement, il était impératif d’en tenir compte et donc de pouvoir refaire les simulations avec une nouvelle prévision météo. Dans ce cadre, la situation initiale des modules doit correspondre à l’état des variables calculées dans un précédent run pour le temps de début des nouvelles prévisions. Il a été mis en place une fonctionnalité « STOP’n’RESUME » qui autorise la simulation, qui fait des sauvegardes à des temps choisis et qui permet de repartir d’une des dates de sauvegarde. Ceci a nécessité l’ajout de quelques modifications dans la plateforme et dans les modules qui utilisent des situations initiales. Le modèle fait aussi appel à des fonctions de pédotransfert pour calculer les propriétés hydrodynamiques. Le modèle final d’aide à la décision utilisera donc des bases de données pour obtenir le climat prévu ainsi que sa variabilité, et des bases de données sol pour l’estimation des propriétés hydrodynamiques. Une des fonctionnalités de VSoil permet de faire des ensembles de simulations en utilisant des distributions de paramètres. Cette capacité permettra d’avoir des estimations des variables d’intérêt avec leur sensibilité aux incertitudes sur le climat et le sol.

Illustration of the degradation of a forest soil

Details of the work carried out under this theme are presented below.

• Emma Schmitt (master's internship from March to September 2023)

Emma Schmitt (AgroParisTech, Nancy) did her master's internship at the ONF in Chambéry (March - September 2023). Her objective was to test the climate service prototype developed in the VSoilForOAD project (see work done by Manon Martin below).

Emma firstly attended a training session on the use of the VSoil platform and the model developed in the context of the VSoilForOAD project. Using rut depths and soil textures measured on the field and moisture conditions predicted by the climate service prototype, Emma showed it is possible to predict trafficability on the designated paths with a simple field setup and meteorological data. To assess the reliability of weather forecasts in terms of practicability, Shell scripts were developed to launch for 35 sites a series of simulations using both observed and previsional meteo data from ECMWF. To significantly shorten model runtimes, the « stop and resume » mecanism was developed in the VSoil platform allowing to stop a simulation, save its complete state and restart the simulation using this saved state. Using this mecanism, the moisture conditions of each day of the 35 work sites were simulated several times with increasing reliability. The use of the « stop and resume » mecanism and of a multi-processor computing server reduced the total simulation time from over one week to less than one hour.

• Mostafa Moradzadeh  (postdoctoral fellow from April 2022 to April 2023)

Mostafa Moradzadeh (INRAE, EMMAH, Avignon) worked on the VSoilForOAD project from April 2022 to April 2023 in replacement of Manon Martin who found a new job (see work done by Manon below).

Mostafa more precisely carried out some sensitivity analysis in order to identify the relative influence of some parameters of the model developed in this project (climatic, soil, vegetation, phenological and mulch parameters, and boundary conditions) on a set of model outputs produced by the module implementing the "soil praticability" process (saturation index, praticability index and drying slope). To do that, two methods implementing the Monte Carlo estimation of the Sobol' indices, named soboljansen and sobolSalt (from "sensitivity" R package, see here), were added in the list of R sensitivity analysis methods available in VSoil.

• Manon Martin (postdoctoral fellow from January 2020 to Mai 2022)

Manon Martin (INRAE, EMMAH, Avignon) started her postdoctoral period in January 2020 as part of the VSoilForOAD project (ADEME project) which aims to develop a prototype of a climate service capable of predicting the practicability of the traffic routes, named skid trails, for logging machinery. This work is a continuation of the development of the ForestOAD model she carried out during her PhD defended in 2019 (see below).

The objective of Manon's postdoctoral period was to find strategies to simplify the parameterisation of this model and to use easily accessible data. To do so, it is necessary to (i) identify the pedotransfer function (empirical relation allowing to determine soil hydraulic properties that are difficult to measure directly, from simple variables such as granulometry) that is most adapted to forest soils and (ii) evaluate the key parameters for predicting water dynamics of forest soils using a global sensitivity analysis method (calculation of Sobol indices).

For this purpose, Manon developed several modules that have been integrated to the ForestOAD model: (1) a Fortran module implementing several pedotransfer functions to be evaluated, (2) a C++ module that implements the "climate" process and reads climatic data from different sources: SAFRAN, Météo-France, ECMWF (European Centre for Medium-Range Weather Forecasts) or local data, (3) a C++ module that implements the "evapotranspiration" process and has the capacity to read hourly PET (potential evapotranspiration) data or calculate PET values from climatic data using the ASCE standardized Penman-Monteith formulation, (4) a Fortran module that implements the "crop development" process and calculates a leaf area index profile (using read or calculated phenological dates) and root lengths, and (5) a C++ module that implements the new created "soil praticability" process and that calculates several soil saturation and soil praticability indexes.

Moreover, the Sobol sensitivity analysis method (from "sensitivity" R package, see here) was added in the list of R sensitivity analysis methods available in VSoil in order Manon can run sensitivity analysis with experimental design built with both continuous and discrete parameters (an example of discrete parameter is the origin of climatic data: SAFRAN, Météo-France, ECMWF or local).

Finally, a new feature was developed by the VSoil team to stop and resume a simulation in order Manon can use VSoil to provide real-time water dynamics by coupling weather forecasts and soil-plant characteristics.

• Manon Martin (PhD student from October 2016 to October 2019)

Manon Martin (FCBA technological institute) developed the ForestOAD model during her PhD (October 2016 - October 2019) entitled "Development of an hydrologic model of forest soils compacted partially usable in a situation of sparse datas" (the abstract of the thesis is available here in french and english versions).

This model is a tool to optimise the forestry machinery operations and avoid problems of soil degradation and compaction. It is a 1D mechanistic soil transfer model based on the PASTIS model (F. Lafolie. 1991. Modeling water flow, nitrogen transport and root uptake including physical non-equilibrium and optimization of the root water potential. Fertilizer Research, 27:215-231 and A. Findeling, P. Garnier, F. Coppens, F. Lafolie, S. Recous, 2007. Modelling water, carbon and nitrogen dynamics in soil covered with decomposing mulch. EJSS, 58, 196-206) and implemented in the VSoil platform. This model was adapted to the forestry context during her thesis.