Search

Crop type mapping at the field level is critical for a variety of applications in agricultural monitoring, and satellite imagery is becoming an increasingly abundant and useful raw input from which to create crop type maps. Still, in many regions crop type mapping with satellite data remains constrained by a scarcity of field-level crop labels for training supervised classification models. When training data is not available in one region, classifiers trained in similar regions can be transferred, but shifts in the distribution of crop types as well as transformations of the features between regions lead to reduced classification accuracy. We present a methodology that uses aggregate-level crop statistics to correct the classifier by accounting for these two types of shifts. To adjust for shifts in the crop type composition we present a scheme for properly reweighting the posterior probabilities of each class that are output by the classifier. To adjust for shifts in features we propose a method to estimate and remove linear shifts in the mean feature vector. We demonstrate that this methodology leads to substantial improvements in overall classification accuracy when using Linear Discriminant Analysis (LDA) to map crop types in Occitanie, France and in Western Province, Kenya. When using LDA as our base classifier, we found that in France our methodology led to percent reductions in misclassifications ranging from 2.8% to 42.2% (mean = 21.9%) over eleven different training departments, and in Kenya the percent reductions in misclassification were 6.6%, 28.4%, and 42.7% for three training regions. While our methodology was statistically motivated by the LDA classifier, it can be applied to any type of classifier. As an example, we demonstrate its successful application to improve a Random Forest classifier. • Automated crop classifiers can suffer in regions lying outside the training region. • Label and feature distribution shift is a major hurdle for crop classification. • We propose a method that simultaneously corrects for both types of shift. • The method uses estimates of crop composition based on aggregate crop statistics. • The method leads to an increase in overall accuracy on two example datasets. [ABSTRACT FROM AUTHOR]

Terrestrial carbon stock estimates information has significant importance in planning decisions for global warming and climate change mitigation. This study aimed to estimate and analyze carbon stock changes in Kenya as consequence of land cover change (LCC) using free open data to provide affordable and timely information. Using Random Forest (RF) decision trees, the land cover for 2028 was modelled based on 2004 and 2016 land cover under a Business as Usual (BAU) and an alternative Reducing of Emissions from Forest Degradation and Deforestation (REDD+) scenarios. The InVEST carbon model was used for estimation and valuation of carbon stock between 2004 and 2028. Results show a 16% decline in carbon stock with a loss of 21 billion US$ under the BAU scenario. On a regional scale, results show a gradual decline in carbon stock in the Coastal and Central regions while other regions exhibited mixed results. This trend can be reversed by the implementation of a REDD + scenario with a possible increase of 1.6% between 2016 and 2028, translating to a gain of 1 billion US$. This study contributes to the understanding of spatiotemporal carbon stock changes under different scenarios for effective spatial planning aiming to a balanced natural resource utilization. [Display omitted] • We estimate carbon stock changes in Kenya as consequence of LULC changes. • We compare business as usual with REDD + scenario using InVEST tool. • A REDD + scenario increases carbon stock in 1.6%–2028 while BAU declines 16%. • We provide a ready-to-use methodology for spatial implementation of REDD + policy. [ABSTRACT FROM AUTHOR]