Find in Library

<p>Soil moisture is key for understanding soil–plant–atmosphere interactions. We provide a soil moisture pattern recognition framework to increase the spatial resolution and fill gaps of the ESA-CCI (European Space Agency Climate Change Initiative v4.5) soil moisture dataset, which contains <span class="inline-formula">&gt;</span> 40 years of satellite soil moisture global grids with a spatial resolution of <span class="inline-formula">∼</span> 27 km. We use terrain parameters coupled with bioclimatic and soil type information to predict finer-grained (i.e., downscaled) satellite soil moisture. We assess the impact of terrain parameters on the prediction accuracy by cross-validating downscaled soil moisture with and without the support of bioclimatic and soil type information. The outcome is a dataset of gap-free global mean annual soil moisture predictions and associated prediction variances for 28 years (1991–2018) across 15 km grids. We use independent in situ records from the International Soil Moisture Network (ISMN, 987 stations) and in situ precipitation records (171 additional stations) only for evaluating the new dataset. Cross-validated correlation between observed and predicted soil moisture values varies from <span class="inline-formula"><i>r</i>=</span> 0.69 to <span class="inline-formula"><i>r</i>=</span> 0.87 with root mean squared errors (RMSEs, m<span class="inline-formula">³</span> m<span class="inline-formula">⁻³</span>) around 0.03 and 0.04. Our soil moisture predictions improve (a) the correlation with the ISMN (when compared with the original ESA-CCI dataset) from <span class="inline-formula"><i>r</i>=</span> 0.30 (RMSE <span class="inline-formula">=</span> 0.09, unbiased RMSE (ubRMSE) <span class="inline-formula">=</span> 0.37) to <span class="inline-formula"><i>r</i>=</span> 0.66 (RMSE <span class="inline-formula">=</span> 0.05, ubRMSE <span class="inline-formula">=</span> 0.18) and (b) the correlation with local precipitation records across boreal (from <span class="inline-formula"><i>r</i>=</span> <span class="inline-formula">&lt;</span> 0.3 up to <span class="inline-formula"><i>r</i>=</span> 0.49) or tropical areas (from <span class="inline-formula"><i>r</i>=</span> <span class="inline-formula">&lt;</span> 0.3 to <span class="inline-formula"><i>r</i>=</span> 0.46) which are currently poorly represented in the ISMN. Temporal trends show a decline of global annual soil moisture using (a) data from the ISMN (<span class="inline-formula"><math xmlns="http://www.w3.org/1998/Math/MathML" id="M19" display="inline" overflow="scroll" dspmath="mathml"><mrow><mo>-</mo><mn mathvariant="normal">1.5</mn><mo>[</mo><mo>-</mo><mn mathvariant="normal">1.8</mn><mo>,</mo><mo>-</mo><mn mathvariant="normal">1.24</mn><mo>]</mo></mrow></math><span><svg:svg xmlns:svg="http://www.w3.org/2000/svg" width="87pt" height="12pt" class="svg-formula" dspmath="mathimg" md5hash="658304eb811f540d4734f0008720da5b"><svg:image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="essd-13-1711-2021-ie00001.svg" width="87pt" height="12pt" src="essd-13-1711-2021-ie00001.png"/></svg:svg></span></span> %), (b) associated locations from the original ESA-CCI dataset (<span class="inline-formula"><math xmlns="http://www.w3.org/1998/Math/MathML" id="M20" display="inline" overflow="scroll" dspmath="mathml"><mrow><mo>-</mo><mn mathvariant="normal">0.87</mn><mo>[</mo><mo>-</mo><mn mathvariant="normal">1.54</mn><mo>,</mo><mo>-</mo><mn mathvariant="normal">0.17</mn><mo>]</mo></mrow></math><span><svg:svg xmlns:svg="http://www.w3.org/2000/svg" width="99pt" height="12pt" class="svg-formula" dspmath="mathimg" md5hash="03108aeb725109f0ee57a38ba7fcec94"><svg:image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="essd-13-1711-2021-ie00002.svg" width="99pt" height="12pt" src="essd-13-1711-2021-ie00002.png"/></svg:svg></span></span> %), (c) associated locations from predictions based on terrain parameters (<span class="inline-formula"><math xmlns="http://www.w3.org/1998/Math/MathML" id="M21" display="inline" overflow="scroll" dspmath="mathml"><mrow><mo>-</mo><mn mathvariant="normal">0.85</mn><mo>[</mo><mo>-</mo><mn mathvariant="normal">1.01</mn><mo>,</mo><mo>-</mo><mn mathvariant="normal">0.49</mn><mo>]</mo></mrow></math><span><svg:svg xmlns:svg="http://www.w3.org/2000/svg" width="99pt" height="12pt" class="svg-formula" dspmath="mathimg" md5hash="bd7293501d0bf895bac5fd77943d3807"><svg:image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="essd-13-1711-2021-ie00003.svg" width="99pt" height="12pt" src="essd-13-1711-2021-ie00003.png"/></svg:svg></span></span> %), and (d) associated locations from predictions including bioclimatic and soil type information (<span class="inline-formula"><math xmlns="http://www.w3.org/1998/Math/MathML" id="M22" display="inline" overflow="scroll" dspmath="mathml"><mrow><mo>-</mo><mn mathvariant="normal">0.68</mn><mo>[</mo><mo>-</mo><mn mathvariant="normal">0.91</mn><mo>,</mo><mo>-</mo><mn mathvariant="normal">0.45</mn><mo>]</mo></mrow></math><span><svg:svg xmlns:svg="http://www.w3.org/2000/svg" width="99pt" height="12pt" class="svg-formula" dspmath="mathimg" md5hash="6c561bdbe7d55e8d47f8578a0ce27fbe"><svg:image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="essd-13-1711-2021-ie00004.svg" width="99pt" height="12pt" src="essd-13-1711-2021-ie00004.png"/></svg:svg></span></span> %). We provide a new soil moisture dataset that has no gaps and higher granularity together with validation methods and a modeling approach that can be applied worldwide (Guevara et al., 2020, <a href="https://doi.org/10.4211/hs.9f981ae4e68b4f529cdd7a5c9013e27e">https://doi.org/10.4211/hs.9f981ae4e68b4f529cdd7a5c9013e27e</a>).</p>