Pierre Roudier

Mapping Daily Air Temperature for Antarctica Based on MODIS LST

Spatial predictions of near-surface air temperature ( T a i r ) in Antarctica are required as baseline information for a variety of research disciplines. Since the network of weather stations in Antarctica is sparse, remote sensing methods have large potential due to their capabilities and accessibility. Based on the MODIS land surface temperature (LST) data, T a i r at the exact time of satellite overpass was modelled at a spatial resolution of 1 km using data from 32 weather stations. The performance of a simple linear regression model to predict T a i r from LST was compared to the performance of three machine learning algorithms: Random Forest (RF), generalized boosted regression models (GBM) and Cubist. In addition to LST, auxiliary predictor variables were tested in these models. Their relevance was evaluated by a Cubist-based forward feature selection in conjunction with leave-one-station-out cross-validation to reduce the impact of spatial overfitting. GBM performed best to predict T a i r using LST and the month of the year as predictor variables. Using the trained model, T a i r could be estimated with a leave-one-station-out cross-validated R 2 of 0.71 and a RMSE of 10.51 ∘ C. However, the machine learning approaches only slightly outperformed the simple linear estimation of T a i r from LST ( R 2 of 0.64, RMSE of 11.02 ∘ C). Using the trained model allowed creating time series of T a i r over Antarctica for 2013. Extending the training data by including more years will allow developing time series of T a i r from 2000 on.

VNIR Soil Spectroscopy for Field Soil Analysis

The advent of affordable, ground-based, global positioning information (GPS)–enabled sensor technologies provides a new method to rapidly acquire georeferenced soil datasets in situ for high-resolution soil attribute mapping. Our research deployed vehicle-mounted electromagnetic sensor survey equipment to map and quantify soil variability (˜50 ha per day) using apparent electrical conductivity as an indirect measure of soil texture and moisture differences. A portable visible–near infrared (VNIR) spectrometer (350–2500 nm) was then used in the field to acquire hyperspectral data from the side of soil cores to a specified depth at optimized sampling locations. The sampling locations were derived by statistical analysis of the electromagnetic survey dataset, to proportionally sample the full range of spatial variability. The VNIR spectra were used to predict soil organic carbon (prediction model using field-moist spectra: R2 = 0.39; RPD = 1.28; and air-dry spectra: R2 = 0.80; RPD = 2.25). These point values were combined with the electromagnetic survey data to produce a soil organic carbon map, using a random forest data mining approach (validation model: R2 = 0.52; RMSE = 3.21 Mg C/ha to 30 cm soil depth; prediction model: R2 = 0.92; RMSE = 1.53 Mg C/ha to 30 cm soil depth). This spatial modeling method, using high-resolution sensor data, enables prediction of soil carbon stocks, and their spatial variability, at a resolution previously impractical using a solely laboratory-based approach.

Advances Towards Quantitative Assessments of Soil Profile Properties

In this paper, we present some advances in digital soil morphometrics techniques in New Zealand. A soil monolith extractor has been developed in house and facilitates the application of digital soil morphometrics techniques. Three distinct soil profiles have been sampled using the monolith extractor to test new ways to collect information from the soil profile. Digital images have been collected on these soil monoliths and calibrated using a set of reference colour chips. The spectral resolution of these images has been enhanced by combining the spatial resolution of the CCD images (1 mm) with the spectral resolution and range of an ASD FieldSpec 3 visible–NIR spectrometer (1 nm between 350 and 2500 nm). A processing chain combining image processing methods such as principal component (PC) analysis and image segmentation has been developed to support the delineation of soil horizons and collect information about the soil structure.

Soil legacy data rescue via GlobalSoilMap and other international and national initiatives

Legacy soil data have been produced over 70 years in nearly all countries of the world. Unfortunately, data, information and knowledge are still currently fragmented and at risk of getting lost if they remain in a paper format. To process this legacy data into consistent, spatially explicit and continuous global soil information, data are being rescued and compiled into databases. Thousands of soil survey reports and maps have been scanned and made available online. The soil profile data reported by these data sources have been captured and compiled into databases. The total number of soil profiles rescued in the selected countries is about 800,000. Currently, data for 117, 000 profiles are compiled and harmonized according to GlobalSoilMap specifications in a world level database (WoSIS). The results presented at the country level are likely to be an underestimate. The majority of soil data is still not rescued and this effort should be pursued. The data have been used to produce soil property maps. We discuss the pro and cons of top-down and bottom-up approaches to produce such maps and we stress their complementarity. We give examples of success stories. The first global soil property maps using rescued data were produced by a top-down approach and were released at a limited resolution of 1km in 2014, followed by an update at a resolution of 250m in 2017. By the end of 2020, we aim to deliver the first worldwide product that fully meets the GlobalSoilMap specifications.

Comparison of regression methods for spatial downscaling of soil organic carbon stocks maps

Abstract This paper presents a refinement of the dissever algorithm, a framework for downscaling spatial information based on available environmental covariates proposed by Malone et al. (2012). While the original algorithm models the relationships between the target variable and the covariates using a general additive model (GAM), the modified procedure presented in this paper allows the user to choose between a wide range of regression methods. These developments have been implemented in an open-source package for the R statistical environment, and tested by downscaling soil organic carbon stocks (SOCS) maps available on two study sites in Australia and New Zealand using 4 different regression methods: linear model (LM), GAM, random forest (RF), and Cubist (CU). In this study, the spatial resolution of a set of reference maps were degraded to a coarser resolution, so to assess the performance of the different downscaling methods. On the Australian site, the 1-km SOCS coarse resolution map has been downscaled to a 90-m resolution. The best results were achieved using either CU or RF ( R 2 = 0.91 and 0.94 respectively). On the New Zealand site, the 250-m SOCS coarse resolution map has been downscaled to a 10-m resolution. The best results were achieved using GAM ( R 2 = 0.90 ). The results illustrate that the optimal regression methods for downscaling spatial information using dissever vary on a case-by-case basis. In particular, simpler approaches such as LM or GAM outperformed more complex approaches in cases where only a limited number of pixels are available to train the downscaling algorithm. This demonstrate the value of an implementation that facilitates testing of different regression strategies.

Integration of NIR on a multi-sensor platform to improve soil resource assessments

Soil is an intriguing and complex entity in terms of physics, chemistry and biology. It has solid, liquid and gas elements, mineral and organic parts and inert and living constituents. All these different components interact on multiple levels between themselves and with external factors such as weather and human activities. These interactions contribute ultimately to the large spatial and temporal variabilities in soil properties, which can be observed at different scales – from global to regional to even the individual field. The soil science community has made incredible progress over the last decades, but we are still far from fully understanding the dynamic complexity that governs soil attributes and processes. Harnessing soil variabilities, across space and time, is a critical step to the sustainable management of that resource. The precision agriculture paradigm provides an illustration of the value of such information: if sitespecific nutrient zones can be delineated in some way, then farmers have an evidence base to fine-tune their inputs (nutrients, water), and adapt these to the very local requirements of the crop. If erosion-prone areas can be identified, appropriate management and remediation procedures can be suggested. If the soils of a farm, or even of a field, can be split into contiguous zones according to their ability to hold onto water, sitespecific irrigation plans can be developed to optimise water use and minimise detrimental effects such as runoff or leaching. To do this, extensive soil sampling, at high temporal and spatial resolution, is required. The reference chemistry laboratory analyses used traditionally in soil science produce high-accuracy results but are not an appropriate solution due the detrimental costs in time and money. To overcome this issue, the soil science community started, years ago, to consider a trade-off of some degree of accuracy if that translates into more samples recorded.

Dissolved organic carbon concentration and denitrification capacity of a hill country sub-catchment as affected by soil type and slope

ABSTRACT Characterising the dissolved organic carbon (DOC) concentration and denitrification capacity of the soils and slopes in hill country is important in order to manage the leaching and availability of nitrate in ground and surface waters. This study investigated the DOC concentration and denitrification capacity of the soils and slope classes of a sub-catchment within a hill country farm, in Palmerston North, New Zealand. Fifty locations comprising of 2 soil orders (Pallic, Brown), 8 soil types (3 drainage classes) and 3 slope classes were sampled from different soil depths down to 1 m. The results suggest that compared to slope, soil type had a greater effect on denitrification capacity within the sub-catchment. The Ramiha soil had the highest DOC concentration (105 mg kg−1 within 0.3–0.6 m depth) and moisture content, and hence the highest denitrification capacity (10 µg kg−1 h−1). This suggests that farms or catchments with similar soil types may have a greater capacity to attenuate nitrogen losses to the environment.