Climate-based Pre-screening of Self-sustaining Regreening Opportunities in Drylands: A Case Study for Saudi Arabia
Katja Froehlich · 2026
Read Climate-based Pre-screening of Self-sustaining Regreening Opportunities in Drylands: A Case Study for Saudi Arabia by Katja Froehlich (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-04206.txt.
Introduction
Katja Froehlich and five co-authors present a way to find dryland sites where native plants might regrow and survive without long-term watering. Their case study is Saudi Arabia, where many past planting efforts have relied on costly irrigation in a water-scarce land.
Using climate records, satellite data, and expert-chosen reference sites, they build a climate score and map it across the country. Sites where the climate looks good but plants are still sparse are flagged. The screening then narrows the search to thirteen priority places. Nearby healthy ecosystems serve as targets for what regrowth could reach.
Catalog record
| CREATOR | Katja Froehlich |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 53 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.04206
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Climate-based Pre-screening of Self-sustaining Regreening Opportunities in Drylands: A Case Study for Saudi Arabia Katja Froehlich Abstract Abstract Large-scale restoration in drylands is widely promoted to address land degradation and biodiversity loss, yet many efforts rely on long-term irrigation, limiting sustainability in water-scarce regions. A key challenge is identifying locations where native vegetation can persist without intensive management while minimizing costly field campaigns. A scalable pre-screening framework is presented that integrates climate and remote sensing data to enable cost-efficient site selection in arid environments using Saudi Arabia as a case study. A Climate Suitability Score (CSS), derived from machine learning models trained on expert-curated reference sites, captures complex climatic dependencies on vegetation persistence. Using multi-year ERA5-Land data for Saudi Arabia, national-scale prediction maps are generated and combined with vegetation indices to identify areas where climate is favorable, but vegetation remains underdeveloped. Multi-criteria screening reduces candidates to thirteen priority locations. Climatically analogous intact ecosystems provide benchmarks for restoration targets and indicate that an average 2.5 fold increase in vegetation coverage is a realistic target for restoration efforts. Overall, this approach narrows the search space, reduces costs, and supports resilient ecosystem recovery planning in water-limited regions. 1. Introduction Desertification is defined as land degradation in hyper-arid to dry sub-humid areas caused mainly by land mismanagement or adverse climatic conditions, e.g., overgrazing and drought 43 , 47 . These stressors lead to loss of vegetation, which destabilizes the soil layers and exposes the land to erosion 37 . Desertification leads to long-term destruction of native ecosystems and, subsequently, to the displacement of populations that rely on these ecosystems 29 . As a result, human intervention is crucial for the restoration and maintenance of native ecosystems 23 . Land restoration efforts, specifically regreening, have spurred global projects to combat desertification, including the Bonn Challenge, the Great Green Wall launched by the African Union, the Three Norths Shelterbelt Program in China, and the Saudi Green Initiative 15 , 34 . However, restoration success in water-limited environments remains highly variable, and many projects fail to establish persistent vegetation cover or require continued human intervention to remain viable. A recurring limitation is that restoration planning often prioritizes visible short-term greening rather than long-term ecological persistence. These projects often rely on continued intensive maintenance, such as irrigation with extracted groundwater 13 . As restoration commitments expand across drylands, there is an urgent need to distinguish locations capable of supporting self-sustaining vegetation recovery from those that remain dependent on external inputs 23 , 34 . Otherwise, precious resources are wasted, and the existing ecosystem could even degrade further 34 , 42 . In the past, many studies have been conducted to find efficient solutions for regreening; however, they have mostly focused on specific, relatively small regions 40 , 17 , 35 . For large-scale restoration projects, not all regions are equally suitable for regreening, necessitating a complex selection process. One option is to select locations based on Geographic Information System (GIS) maps, using metrics such as the Normalized Difference Vegetation Index (NDVI) to assess changes in vegetation cover at large scales 3 , 22 . A reduction of the NDVI over time indicates a loss of vegetation that could potentially be restored through regreening initiatives. Generally, NDVI values between 0.1 and 0.2 are associated with barren land, while values between 0.2 and 1 correspond to vegetation 24 . However, in arid climates, NDVI analysis can be misleading because vegetation naturally has low visible cover due to smaller leaf areas despite extensive root systems, making standard thresholds unreliable 24 , 6 , 25 . Additionally, NDVI cannot effectively differentiate between orchards and natural perennial vegetation, often leading to overestimation of natural vegetation potential. Finally, low NDVI values are ambiguous, as they may represent either true desert conditions or degraded land that could still be suitable for land restoration. Rather than relying solely on the NDVI, suitable locations for land restoration should therefore be selected based on additional environmental factors such as climate, soil, and water quality, as well as accessibility for plant deployment and monitoring 23 , 34 . However, including these factors increases the cost of the selection analysis as it necessitates on-site analysis for accurate soil and water quality assessment. In this work, we therefore propose to employ an additional pre-selection step that relies solely on climate factors, for which global datasets are readily available. For example, the ERA5 Land dataset provides globally consistent, multi-year climate data with high temporal resolution 32 . Since plant survival depends on a complex interplay of factors rather than on simple criteria such as specific temperature intervals, it is challenging to explicitly compute the long-term survival potential of a given location. Therefore, we opt for an implicit, machine learning-based approach, in which the training data captures the underlying complexity of the survival criteria. A model is trained on expert-curated reference locations representing persistent natural vegetation, degraded landscapes and climatically unsuitable areas. It then predicts a Climate Suitability Score for Vegetation Survival (CSS). A promising land restoration location is then characterized by a high CSS but low current vegetation cover. We apply our framework to Saudi Arabia as a model for dryland systems. Here, annual declines of 1.2–1.5% in shrub and tree cover within rangelands have been reported 3 , 30 . The Saudi Green Initiative pledges to restore 74 million ha of land by planting 10 billion trees across the country as part of its Vision 2030 39 . We generate national-scale opportunity maps, prioritize feasible restoration sites, and identify climatically analogous intact ecosystems that provide realistic reference conditions and restoration targets. Our results establish a scalable approach for directing restoration effort toward landscapes with higher long-term ecological potential in water-limited regions worldwide. 2. Results 2.1. Climate Suitability Score (CSS) Model Training Figure 1 : Map of Saudi Arabia. The 230 sample locations and their classes are shown according to climatic suitability for vegetation and prevalence of vegetation. Saudi Arabia encompasses strong climatic gradients ranging from hyper-arid deserts to comparatively mesic mountain systems in the south-west, providing a stringent test case for climate-based restoration screening. We trained two complementary machine-learning models on expert-curated reference locations representing combinations of climatic suitability and vegetation prevalence. Multi-year ERA5-Land time series were transformed into compact feature representations and used to predict a continuous climate suitability score across the national landscape. Saudi Arabia exhibits pronounced hydroclimatic heterogeneity, reflecting its predominantly arid to semi-arid climate and complex topography. Climatic and topographic constraints contribute to the patchy distribution of natural vegetation. Most vegetation cover is concentrated in the southwestern highlands, whereas in other regions it is largely restricted to wadis, stream networks, and other favorable microhabitats (Figure S1 ). Extreme aridity, high solar radiation, and sandy soils create inhospitable conditions across much of the country, selecting species with high tolerance to heat and drought. To capture this variability, we constructed a nationwide training dataset comprising 230 locations distributed across all major climate zones and landforms (Figure 1 , Figure S1 ). 23 climate variables from the ERA5-Land hourly data set 32 were selected, and three-hourly data spanning the years 2020–2024 were extracted for each site. Only variables irrelevant for our region, such as snow density or lake ice temperature