USGS Science⌕ Search
Home/Archive/USGS

USGS · 70259740

Testing spatial out-of-sample area of influence for grain forecasting models

Frank Davenport·Donghoon Lee·Shraddhanand Shukla·Greg Husak·W. Chris Funk·Michael Budde·James Rowland
View original source ↗Export metadata ↓

Abstract

We examine the factors that determine if a grain forecasting model fit to one region can be transferred to another region. Prior research has proposed examining the area of applicability (AoA) of a model based on structurally similar characteristics in the Earth Observation predictors and weights based on the model derived feature importance. We expand on and evaluate this approach in the context of grain yield forecasting in Sub-Saharan Africa. Specifically, we evaluate an AoA methodology established for generating raster surfaces and apply it to vector supported grain data. We fit a series of ensemble tree models both within single countries and across multiple sets of countries and then test those models in countries excluded from the training set. We then calculate and decompose AoA measures and examine several different performance metrics. We find that the spatial transfer accuracy does not vary across season but does vary by average rainfall and across high, medium, and low yielding regions. In general, areas with higher yields and medium to high average rainfall tend to have higher accuracy for both model training and transfer. Finally, we find that fitting models with multiple countries provides more accurate out-of-sample estimates when compared to models fitted to a single country.

Explore related subjects

Enivronmental Research LettersArticle
90° N90° S · 180° W ← longitude → 180° E
Source-reported bounding extent: -16.8013° to 15.11616° latitude; -5.47056° to 51.13387° longitude. This indicates report coverage, not an exact sampling location. View area on OpenStreetMap.

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Frank Davenport, Donghoon Lee, Shraddhanand Shukla, Greg Husak, W. Chris Funk, Michael Budde, James Rowland. 2024-10-18. Testing spatial out-of-sample area of influence for grain forecasting models. https://doi.org/10.1088/1748-9326%2Fad845e

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related USGS reports

USGS2013

Reorganization of vegetation, hydrology and soil carbon after permafrost degradation across heterogeneous boreal landscapes

The diversity of ecosystems across boreal landscapes, successional changes after disturbance and complicated permafrost histories, present enormous challenges for assessing how vegetation, water and soil carbon may respond to climate change in boreal regions. To address this complexity, we used a chronosequence approach to assess changes in vegetation composition, water storage and soil organic carbon (SOC) stocks along successional gradients within four landscapes: (1) rocky uplands on ice-poor hillside colluvium, (2) silty uplands on extremely ice-rich loess, (3) gravelly–sandy lowlands on ice-poor eolian sand and (4) peaty–silty lowlands on thick ice-rich peat deposits over reworked lowland loess. In rocky uplands, after fire permafrost thawed rapidly due to low ice contents, soils became well drained and SOC stocks decreased slightly. In silty uplands, after fire permafrost persisted, soils remained saturated and SOC decreased slightly. In gravelly–sandy lowlands where permafrost persisted in drier forest soils, loss of deeper permafrost around lakes has allowed recent widespread drainage of lakes that has exposed limnic material with high SOC to aerobic decomposition. In peaty–silty lowlands, 2–4 m of thaw settlement led to fragmented drainage patterns in isolated thermokarst bogs and flooding of soils, and surface soils accumulated new bog peat. We were not able to detect SOC changes in deeper soils, however, due to high variability. Complicated soil stratigraphy revealed that permafrost has repeatedly aggraded and degraded in all landscapes during the Holocene, although in silty uplands only the upper permafrost was affected. Overall, permafrost thaw has led to the reorganization of vegetation, water storage and flow paths, and patterns of SOC accumulation. However, changes have occurred over different timescales among landscapes: over decades in rocky uplands and gravelly–sandy lowlands in response to fire and lake drainage, over decades to centuries in peaty–silty lowlands with a legacy of complicated Holocene changes, and over centuries in silty uplands where ice-rich soil and ecological recovery protect permafrost.

Alaska↗